diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,186016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.769964609939991,
+  "global_step": 31000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820516e-08,
+      "loss": 0.9406,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.102564102564103e-08,
+      "loss": 1.1806,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846154e-08,
+      "loss": 0.9356,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.205128205128206e-08,
+      "loss": 0.8198,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-07,
+      "loss": 1.0557,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2307692307692308e-07,
+      "loss": 0.8908,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.435897435897436e-07,
+      "loss": 0.89,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6410256410256413e-07,
+      "loss": 0.8708,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8461538461538464e-07,
+      "loss": 0.9641,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820514e-07,
+      "loss": 1.0983,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2564102564102568e-07,
+      "loss": 0.8763,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4615384615384616e-07,
+      "loss": 0.8658,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 0.9204,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.871794871794872e-07,
+      "loss": 0.8675,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-07,
+      "loss": 0.8215,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2820512820512825e-07,
+      "loss": 1.0958,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.487179487179487e-07,
+      "loss": 0.9418,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6923076923076927e-07,
+      "loss": 0.923,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.897435897435898e-07,
+      "loss": 0.977,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.102564102564103e-07,
+      "loss": 0.9048,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.307692307692308e-07,
+      "loss": 0.9865,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5128205128205136e-07,
+      "loss": 0.9995,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7179487179487187e-07,
+      "loss": 0.8733,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.923076923076923e-07,
+      "loss": 0.8785,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 0.7954,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 0.8758,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.53846153846154e-07,
+      "loss": 0.8867,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.743589743589744e-07,
+      "loss": 0.8683,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.948717948717949e-07,
+      "loss": 0.8474,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 0.9133,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.35897435897436e-07,
+      "loss": 0.9483,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.564102564102565e-07,
+      "loss": 0.9282,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.769230769230769e-07,
+      "loss": 0.992,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.974358974358974e-07,
+      "loss": 0.8505,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.179487179487179e-07,
+      "loss": 0.8873,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.384615384615385e-07,
+      "loss": 0.9843,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.58974358974359e-07,
+      "loss": 0.9362,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.794871794871796e-07,
+      "loss": 0.9963,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.806,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128206e-07,
+      "loss": 0.8419,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.410256410256411e-07,
+      "loss": 0.9232,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.615384615384616e-07,
+      "loss": 0.97,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.820512820512821e-07,
+      "loss": 0.9234,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.025641025641027e-07,
+      "loss": 0.9292,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-07,
+      "loss": 0.8779,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.435897435897437e-07,
+      "loss": 0.8868,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.641025641025642e-07,
+      "loss": 0.9965,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.846153846153847e-07,
+      "loss": 0.8918,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0051282051282053e-06,
+      "loss": 1.0048,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 0.9204,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0461538461538463e-06,
+      "loss": 0.8202,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 0.7746,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0871794871794873e-06,
+      "loss": 0.8725,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.107692307692308e-06,
+      "loss": 0.9133,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1282051282051283e-06,
+      "loss": 0.8561,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1487179487179487e-06,
+      "loss": 0.8109,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1692307692307693e-06,
+      "loss": 0.97,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1897435897435897e-06,
+      "loss": 0.8456,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2102564102564104e-06,
+      "loss": 0.8889,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 0.9299,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2512820512820514e-06,
+      "loss": 0.908,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.271794871794872e-06,
+      "loss": 0.9105,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2923076923076924e-06,
+      "loss": 0.8684,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.312820512820513e-06,
+      "loss": 0.9085,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.848,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3538461538461538e-06,
+      "loss": 0.8162,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3743589743589744e-06,
+      "loss": 0.8991,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3948717948717948e-06,
+      "loss": 0.9581,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4153846153846155e-06,
+      "loss": 0.881,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4358974358974359e-06,
+      "loss": 0.8085,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4564102564102565e-06,
+      "loss": 0.8634,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.476923076923077e-06,
+      "loss": 0.8271,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4974358974358977e-06,
+      "loss": 0.9274,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.517948717948718e-06,
+      "loss": 0.87,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.8193,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5589743589743591e-06,
+      "loss": 0.7907,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5794871794871797e-06,
+      "loss": 0.8706,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.8999,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6205128205128208e-06,
+      "loss": 0.8157,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6410256410256412e-06,
+      "loss": 0.7381,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6615384615384616e-06,
+      "loss": 0.8955,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6820512820512822e-06,
+      "loss": 0.7578,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7025641025641026e-06,
+      "loss": 0.7808,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7230769230769232e-06,
+      "loss": 0.6865,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7435897435897436e-06,
+      "loss": 0.903,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7641025641025642e-06,
+      "loss": 0.8453,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7846153846153846e-06,
+      "loss": 0.9533,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8051282051282054e-06,
+      "loss": 0.8293,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8256410256410258e-06,
+      "loss": 0.8582,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 0.7616,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 0.928,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8871794871794875e-06,
+      "loss": 0.8213,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.907692307692308e-06,
+      "loss": 0.8546,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9282051282051285e-06,
+      "loss": 0.9477,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.948717948717949e-06,
+      "loss": 0.9053,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9692307692307693e-06,
+      "loss": 0.8115,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9897435897435897e-06,
+      "loss": 0.9344,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0102564102564105e-06,
+      "loss": 0.874,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.030769230769231e-06,
+      "loss": 0.7802,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 0.7505,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0717948717948717e-06,
+      "loss": 0.787,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0923076923076926e-06,
+      "loss": 0.8392,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.112820512820513e-06,
+      "loss": 0.7409,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 0.8521,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.153846153846154e-06,
+      "loss": 0.8243,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1743589743589746e-06,
+      "loss": 0.8227,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.194871794871795e-06,
+      "loss": 0.8086,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.215384615384616e-06,
+      "loss": 0.8339,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2358974358974362e-06,
+      "loss": 0.8639,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2564102564102566e-06,
+      "loss": 0.8028,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.276923076923077e-06,
+      "loss": 0.8096,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2974358974358975e-06,
+      "loss": 0.8874,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3179487179487183e-06,
+      "loss": 0.8884,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3384615384615387e-06,
+      "loss": 0.8519,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.358974358974359e-06,
+      "loss": 0.7744,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3794871794871795e-06,
+      "loss": 0.7634,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.7106,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4205128205128207e-06,
+      "loss": 0.82,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.441025641025641e-06,
+      "loss": 0.8137,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 0.8442,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4820512820512824e-06,
+      "loss": 0.7487,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5025641025641028e-06,
+      "loss": 0.8495,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.523076923076923e-06,
+      "loss": 0.8657,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.543589743589744e-06,
+      "loss": 0.868,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 0.8264,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.584615384615385e-06,
+      "loss": 0.7406,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.605128205128205e-06,
+      "loss": 0.7422,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.625641025641026e-06,
+      "loss": 0.9073,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6461538461538464e-06,
+      "loss": 0.7899,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.8369,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6871794871794877e-06,
+      "loss": 0.8533,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7076923076923076e-06,
+      "loss": 0.8895,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7282051282051285e-06,
+      "loss": 0.7419,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.748717948717949e-06,
+      "loss": 0.7537,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7692307692307697e-06,
+      "loss": 0.7513,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7897435897435897e-06,
+      "loss": 0.8537,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8102564102564105e-06,
+      "loss": 0.8094,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.830769230769231e-06,
+      "loss": 0.7643,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8512820512820517e-06,
+      "loss": 0.8747,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8717948717948717e-06,
+      "loss": 0.8702,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8923076923076925e-06,
+      "loss": 0.8042,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.912820512820513e-06,
+      "loss": 0.7789,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 0.8104,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.953846153846154e-06,
+      "loss": 0.7191,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9743589743589746e-06,
+      "loss": 0.8064,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9948717948717954e-06,
+      "loss": 0.7395,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0153846153846154e-06,
+      "loss": 0.8477,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.035897435897436e-06,
+      "loss": 0.7233,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0564102564102566e-06,
+      "loss": 0.7721,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.8101,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0974358974358974e-06,
+      "loss": 0.7348,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1179487179487182e-06,
+      "loss": 0.7795,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1384615384615386e-06,
+      "loss": 0.8062,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1589743589743595e-06,
+      "loss": 0.7572,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1794871794871795e-06,
+      "loss": 0.8075,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.7472,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2205128205128207e-06,
+      "loss": 0.8603,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2410256410256415e-06,
+      "loss": 0.751,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2615384615384615e-06,
+      "loss": 0.8344,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2820512820512823e-06,
+      "loss": 0.8599,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.302564102564103e-06,
+      "loss": 0.8371,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.323076923076923e-06,
+      "loss": 0.7772,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.343589743589744e-06,
+      "loss": 0.8298,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3641025641025644e-06,
+      "loss": 0.7678,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.384615384615385e-06,
+      "loss": 0.8191,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.405128205128205e-06,
+      "loss": 0.7022,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.425641025641026e-06,
+      "loss": 0.7794,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4461538461538464e-06,
+      "loss": 0.888,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 0.7652,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.487179487179487e-06,
+      "loss": 0.8145,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.507692307692308e-06,
+      "loss": 0.8955,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5282051282051284e-06,
+      "loss": 0.7853,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5487179487179493e-06,
+      "loss": 0.7298,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5692307692307692e-06,
+      "loss": 0.714,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 0.662,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.610256410256411e-06,
+      "loss": 0.8445,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.630769230769231e-06,
+      "loss": 0.7885,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6512820512820517e-06,
+      "loss": 0.784,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.671794871794872e-06,
+      "loss": 0.7569,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 0.811,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.712820512820513e-06,
+      "loss": 0.7827,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 0.8568,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.753846153846154e-06,
+      "loss": 0.8187,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.774358974358975e-06,
+      "loss": 0.7578,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.794871794871795e-06,
+      "loss": 0.903,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.815384615384616e-06,
+      "loss": 0.6702,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.835897435897436e-06,
+      "loss": 0.7279,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.856410256410257e-06,
+      "loss": 0.7572,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.876923076923077e-06,
+      "loss": 0.8597,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.897435897435898e-06,
+      "loss": 0.6962,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.917948717948718e-06,
+      "loss": 0.7496,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.938461538461539e-06,
+      "loss": 0.7749,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.95897435897436e-06,
+      "loss": 0.8117,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.979487179487179e-06,
+      "loss": 0.7882,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.8032,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.020512820512821e-06,
+      "loss": 0.8091,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0410256410256415e-06,
+      "loss": 0.7585,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.061538461538462e-06,
+      "loss": 0.7492,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.082051282051282e-06,
+      "loss": 0.8823,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.7358,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.123076923076924e-06,
+      "loss": 0.7649,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1435897435897435e-06,
+      "loss": 0.6531,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.164102564102565e-06,
+      "loss": 0.7902,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.184615384615385e-06,
+      "loss": 0.7063,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2051282051282055e-06,
+      "loss": 0.816,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.225641025641026e-06,
+      "loss": 0.8109,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.246153846153846e-06,
+      "loss": 0.8504,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 0.8036,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.287179487179487e-06,
+      "loss": 0.7618,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 0.842,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.328205128205129e-06,
+      "loss": 0.7105,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.348717948717949e-06,
+      "loss": 0.8942,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.36923076923077e-06,
+      "loss": 0.7183,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.38974358974359e-06,
+      "loss": 0.7217,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4102564102564104e-06,
+      "loss": 0.7926,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.430769230769232e-06,
+      "loss": 0.7247,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.451282051282051e-06,
+      "loss": 0.7731,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4717948717948725e-06,
+      "loss": 0.8425,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.492307692307693e-06,
+      "loss": 0.8524,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.512820512820513e-06,
+      "loss": 0.7169,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 0.7349,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.553846153846154e-06,
+      "loss": 0.7637,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5743589743589745e-06,
+      "loss": 0.6892,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.594871794871795e-06,
+      "loss": 0.6911,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.7698,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6358974358974366e-06,
+      "loss": 0.723,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.656410256410257e-06,
+      "loss": 0.7807,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.676923076923077e-06,
+      "loss": 0.7789,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.697435897435898e-06,
+      "loss": 0.8325,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.717948717948718e-06,
+      "loss": 0.7784,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.738461538461539e-06,
+      "loss": 0.7016,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.758974358974359e-06,
+      "loss": 0.7366,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77948717948718e-06,
+      "loss": 0.8953,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.7708,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.820512820512821e-06,
+      "loss": 0.8112,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8410256410256414e-06,
+      "loss": 0.785,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.861538461538462e-06,
+      "loss": 0.7643,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.882051282051282e-06,
+      "loss": 0.9118,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.902564102564103e-06,
+      "loss": 0.7724,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 0.7338,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943589743589744e-06,
+      "loss": 0.8379,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.964102564102565e-06,
+      "loss": 0.6678,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984615384615385e-06,
+      "loss": 0.721,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.0051282051282055e-06,
+      "loss": 0.6766,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.025641025641026e-06,
+      "loss": 0.709,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.046153846153846e-06,
+      "loss": 0.7339,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 0.685,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.087179487179488e-06,
+      "loss": 0.6321,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.1076923076923075e-06,
+      "loss": 0.7674,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.7267,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.148717948717949e-06,
+      "loss": 0.7162,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.16923076923077e-06,
+      "loss": 0.7891,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.18974358974359e-06,
+      "loss": 0.733,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.21025641025641e-06,
+      "loss": 0.8123,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.230769230769232e-06,
+      "loss": 0.7801,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.251282051282052e-06,
+      "loss": 0.7233,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.271794871794872e-06,
+      "loss": 0.8277,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.292307692307693e-06,
+      "loss": 0.719,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.312820512820513e-06,
+      "loss": 0.846,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.6652,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.353846153846154e-06,
+      "loss": 0.7973,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.374358974358975e-06,
+      "loss": 0.7711,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.394871794871796e-06,
+      "loss": 0.8153,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.415384615384615e-06,
+      "loss": 0.7481,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.435897435897436e-06,
+      "loss": 0.7107,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.456410256410257e-06,
+      "loss": 0.7788,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.476923076923077e-06,
+      "loss": 0.7239,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.497435897435898e-06,
+      "loss": 0.7744,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.517948717948718e-06,
+      "loss": 0.7321,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 0.712,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.55897435897436e-06,
+      "loss": 0.6946,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.579487179487179e-06,
+      "loss": 0.7985,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.6611,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.620512820512821e-06,
+      "loss": 0.7926,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 0.6823,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.661538461538462e-06,
+      "loss": 0.878,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.682051282051283e-06,
+      "loss": 0.652,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7025641025641035e-06,
+      "loss": 0.6777,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.723076923076923e-06,
+      "loss": 0.6616,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.743589743589743e-06,
+      "loss": 0.7432,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.764102564102565e-06,
+      "loss": 0.7555,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.784615384615385e-06,
+      "loss": 0.7092,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8051282051282055e-06,
+      "loss": 0.7599,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.825641025641026e-06,
+      "loss": 0.7708,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.846153846153847e-06,
+      "loss": 0.7587,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 0.7088,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.887179487179487e-06,
+      "loss": 0.8174,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.907692307692308e-06,
+      "loss": 0.7953,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.928205128205129e-06,
+      "loss": 0.8036,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.948717948717949e-06,
+      "loss": 0.7137,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9692307692307695e-06,
+      "loss": 0.7583,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.989743589743591e-06,
+      "loss": 0.688,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.010256410256411e-06,
+      "loss": 0.8034,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.030769230769231e-06,
+      "loss": 0.7452,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.051282051282051e-06,
+      "loss": 0.7457,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.071794871794872e-06,
+      "loss": 0.8002,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.092307692307693e-06,
+      "loss": 0.7527,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.112820512820513e-06,
+      "loss": 0.7722,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 0.6915,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.7305,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.174358974358975e-06,
+      "loss": 0.8126,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.194871794871795e-06,
+      "loss": 0.6434,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.215384615384615e-06,
+      "loss": 0.7571,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2358974358974365e-06,
+      "loss": 0.7341,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.256410256410257e-06,
+      "loss": 0.7379,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.276923076923077e-06,
+      "loss": 0.6766,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2974358974358985e-06,
+      "loss": 0.6293,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.317948717948719e-06,
+      "loss": 0.6968,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3384615384615385e-06,
+      "loss": 0.708,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.358974358974359e-06,
+      "loss": 0.6001,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.37948717948718e-06,
+      "loss": 0.6963,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.7314,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.420512820512821e-06,
+      "loss": 0.7544,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.441025641025641e-06,
+      "loss": 0.8277,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.461538461538463e-06,
+      "loss": 0.7588,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.482051282051283e-06,
+      "loss": 0.7183,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5025641025641026e-06,
+      "loss": 0.7183,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.523076923076923e-06,
+      "loss": 0.664,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.543589743589744e-06,
+      "loss": 0.8708,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.564102564102565e-06,
+      "loss": 0.7067,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.584615384615385e-06,
+      "loss": 0.7735,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.605128205128206e-06,
+      "loss": 0.7525,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.625641025641027e-06,
+      "loss": 0.7429,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.646153846153846e-06,
+      "loss": 0.6721,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7452,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.687179487179488e-06,
+      "loss": 0.767,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.707692307692308e-06,
+      "loss": 0.7238,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.728205128205129e-06,
+      "loss": 0.7269,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.748717948717949e-06,
+      "loss": 0.7979,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 0.7224,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.789743589743591e-06,
+      "loss": 0.6774,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.81025641025641e-06,
+      "loss": 0.735,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.830769230769231e-06,
+      "loss": 0.8133,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.851282051282052e-06,
+      "loss": 0.7314,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.871794871794872e-06,
+      "loss": 0.7336,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.892307692307693e-06,
+      "loss": 0.7264,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.912820512820514e-06,
+      "loss": 0.7808,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 0.7563,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.953846153846154e-06,
+      "loss": 0.7596,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.974358974358974e-06,
+      "loss": 0.7906,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.994871794871796e-06,
+      "loss": 0.6288,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.015384615384616e-06,
+      "loss": 0.8034,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.0358974358974364e-06,
+      "loss": 0.6853,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.056410256410257e-06,
+      "loss": 0.766,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.076923076923078e-06,
+      "loss": 0.8603,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.0974358974358985e-06,
+      "loss": 0.7967,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.117948717948718e-06,
+      "loss": 0.7187,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.1384615384615385e-06,
+      "loss": 0.6909,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.15897435897436e-06,
+      "loss": 0.764,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.724,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.8503,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.220512820512822e-06,
+      "loss": 0.6496,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.241025641025642e-06,
+      "loss": 0.781,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.261538461538462e-06,
+      "loss": 0.7539,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.282051282051282e-06,
+      "loss": 0.7784,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.302564102564103e-06,
+      "loss": 0.6887,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.323076923076924e-06,
+      "loss": 0.7064,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.343589743589744e-06,
+      "loss": 0.7841,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.364102564102565e-06,
+      "loss": 0.7456,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 0.7072,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.405128205128206e-06,
+      "loss": 0.6761,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.425641025641026e-06,
+      "loss": 0.7134,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.446153846153846e-06,
+      "loss": 0.5944,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.6926,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.487179487179488e-06,
+      "loss": 0.6765,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.507692307692308e-06,
+      "loss": 0.7,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.528205128205129e-06,
+      "loss": 0.7151,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.54871794871795e-06,
+      "loss": 0.7341,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5692307692307695e-06,
+      "loss": 0.7376,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.58974358974359e-06,
+      "loss": 0.7988,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.610256410256411e-06,
+      "loss": 0.7162,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.630769230769232e-06,
+      "loss": 0.7325,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.651282051282051e-06,
+      "loss": 0.7018,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.671794871794871e-06,
+      "loss": 0.6515,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.5577,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.712820512820514e-06,
+      "loss": 0.6396,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 0.7666,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.753846153846155e-06,
+      "loss": 0.7462,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.774358974358975e-06,
+      "loss": 0.6613,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.794871794871796e-06,
+      "loss": 0.7571,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.815384615384616e-06,
+      "loss": 0.6669,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.835897435897436e-06,
+      "loss": 0.676,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.856410256410257e-06,
+      "loss": 0.6595,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.876923076923077e-06,
+      "loss": 0.7501,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.897435897435898e-06,
+      "loss": 0.6315,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.91794871794872e-06,
+      "loss": 0.6784,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.93846153846154e-06,
+      "loss": 0.7335,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.958974358974359e-06,
+      "loss": 0.7318,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.97948717948718e-06,
+      "loss": 0.7613,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7109,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.020512820512822e-06,
+      "loss": 0.8099,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.041025641025642e-06,
+      "loss": 0.8523,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.061538461538463e-06,
+      "loss": 0.7574,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.082051282051283e-06,
+      "loss": 0.8356,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.102564102564103e-06,
+      "loss": 0.6929,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.123076923076924e-06,
+      "loss": 0.6417,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.143589743589744e-06,
+      "loss": 0.8469,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.164102564102565e-06,
+      "loss": 0.7656,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.184615384615385e-06,
+      "loss": 0.7181,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.6814,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.225641025641027e-06,
+      "loss": 0.7574,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.246153846153848e-06,
+      "loss": 0.8518,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 0.6451,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.287179487179487e-06,
+      "loss": 0.7733,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.307692307692309e-06,
+      "loss": 0.757,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.32820512820513e-06,
+      "loss": 0.6975,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.34871794871795e-06,
+      "loss": 0.78,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.36923076923077e-06,
+      "loss": 0.6824,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.38974358974359e-06,
+      "loss": 0.7517,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.410256410256411e-06,
+      "loss": 0.7558,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.430769230769231e-06,
+      "loss": 0.7758,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.451282051282052e-06,
+      "loss": 0.6326,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.471794871794872e-06,
+      "loss": 0.6788,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.492307692307693e-06,
+      "loss": 0.7094,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.512820512820513e-06,
+      "loss": 0.7731,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 0.6687,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.553846153846156e-06,
+      "loss": 0.7417,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.574358974358974e-06,
+      "loss": 0.7989,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.594871794871795e-06,
+      "loss": 0.7123,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 0.6607,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.635897435897437e-06,
+      "loss": 0.6818,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.656410256410258e-06,
+      "loss": 0.6991,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.676923076923078e-06,
+      "loss": 0.7781,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.697435897435898e-06,
+      "loss": 0.6444,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 0.7389,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.73846153846154e-06,
+      "loss": 0.6659,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.75897435897436e-06,
+      "loss": 0.7765,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.77948717948718e-06,
+      "loss": 0.7578,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.8e-06,
+      "loss": 0.7768,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.820512820512821e-06,
+      "loss": 0.7212,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.841025641025641e-06,
+      "loss": 0.6862,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.861538461538463e-06,
+      "loss": 0.7711,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.882051282051282e-06,
+      "loss": 0.7678,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.902564102564102e-06,
+      "loss": 0.7492,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.923076923076925e-06,
+      "loss": 0.754,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.943589743589745e-06,
+      "loss": 0.7414,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.964102564102565e-06,
+      "loss": 0.7374,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.984615384615386e-06,
+      "loss": 0.758,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.005128205128206e-06,
+      "loss": 0.7488,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.025641025641027e-06,
+      "loss": 0.8115,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.046153846153847e-06,
+      "loss": 0.7117,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.7362,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.087179487179488e-06,
+      "loss": 0.6696,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.107692307692308e-06,
+      "loss": 0.6638,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.128205128205129e-06,
+      "loss": 0.7435,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.148717948717949e-06,
+      "loss": 0.7085,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.169230769230771e-06,
+      "loss": 0.7049,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.18974358974359e-06,
+      "loss": 0.7257,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.21025641025641e-06,
+      "loss": 0.6819,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.6933,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.251282051282053e-06,
+      "loss": 0.7498,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.271794871794873e-06,
+      "loss": 0.6928,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.292307692307694e-06,
+      "loss": 0.7646,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.312820512820514e-06,
+      "loss": 0.6916,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.6714,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.353846153846155e-06,
+      "loss": 0.7689,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.374358974358975e-06,
+      "loss": 0.7613,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.394871794871796e-06,
+      "loss": 0.6298,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.415384615384616e-06,
+      "loss": 0.6846,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.435897435897436e-06,
+      "loss": 0.6838,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.456410256410257e-06,
+      "loss": 0.7023,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.476923076923079e-06,
+      "loss": 0.7782,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.497435897435898e-06,
+      "loss": 0.6903,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.517948717948718e-06,
+      "loss": 0.6965,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.53846153846154e-06,
+      "loss": 0.779,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.55897435897436e-06,
+      "loss": 0.8156,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.579487179487181e-06,
+      "loss": 0.7498,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.7471,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.620512820512822e-06,
+      "loss": 0.7144,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.641025641025642e-06,
+      "loss": 0.7379,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.661538461538462e-06,
+      "loss": 0.8281,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.682051282051283e-06,
+      "loss": 0.6976,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.702564102564103e-06,
+      "loss": 0.7778,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.723076923076924e-06,
+      "loss": 0.6803,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.7161,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.764102564102564e-06,
+      "loss": 0.7494,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.784615384615387e-06,
+      "loss": 0.6606,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.805128205128205e-06,
+      "loss": 0.6509,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.825641025641026e-06,
+      "loss": 0.7081,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 0.7234,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 0.7261,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887179487179489e-06,
+      "loss": 0.7689,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.907692307692309e-06,
+      "loss": 0.7332,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.92820512820513e-06,
+      "loss": 0.8202,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94871794871795e-06,
+      "loss": 0.7799,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96923076923077e-06,
+      "loss": 0.7083,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98974358974359e-06,
+      "loss": 0.7644,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0010256410256411e-05,
+      "loss": 0.7386,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0030769230769231e-05,
+      "loss": 0.6471,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0051282051282052e-05,
+      "loss": 0.6879,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0071794871794874e-05,
+      "loss": 0.7003,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0092307692307693e-05,
+      "loss": 0.7287,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0112820512820513e-05,
+      "loss": 0.7345,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 0.7862,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0153846153846154e-05,
+      "loss": 0.7108,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0174358974358976e-05,
+      "loss": 0.7034,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0194871794871796e-05,
+      "loss": 0.8047,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0215384615384615e-05,
+      "loss": 0.788,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0235897435897437e-05,
+      "loss": 0.738,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.7434,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0276923076923078e-05,
+      "loss": 0.7564,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0297435897435898e-05,
+      "loss": 0.6534,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0317948717948719e-05,
+      "loss": 0.6265,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.033846153846154e-05,
+      "loss": 0.7376,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0358974358974361e-05,
+      "loss": 0.6626,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.037948717948718e-05,
+      "loss": 0.8322,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.04e-05,
+      "loss": 0.8114,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.042051282051282e-05,
+      "loss": 0.7996,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0441025641025641e-05,
+      "loss": 0.7156,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 0.6311,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0482051282051282e-05,
+      "loss": 0.7216,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0502564102564104e-05,
+      "loss": 0.7924,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0523076923076924e-05,
+      "loss": 0.7401,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0543589743589743e-05,
+      "loss": 0.7253,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0564102564102565e-05,
+      "loss": 0.7603,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0584615384615386e-05,
+      "loss": 0.7027,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0605128205128206e-05,
+      "loss": 0.6893,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0625641025641027e-05,
+      "loss": 0.8435,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0646153846153845e-05,
+      "loss": 0.7215,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.6666,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.068717948717949e-05,
+      "loss": 0.7328,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0707692307692308e-05,
+      "loss": 0.7725,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0728205128205129e-05,
+      "loss": 0.6882,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.074871794871795e-05,
+      "loss": 0.7196,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.8277,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0789743589743591e-05,
+      "loss": 0.6787,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.081025641025641e-05,
+      "loss": 0.6739,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.083076923076923e-05,
+      "loss": 0.6829,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0851282051282053e-05,
+      "loss": 0.6599,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0871794871794871e-05,
+      "loss": 0.664,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0892307692307693e-05,
+      "loss": 0.7458,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0912820512820514e-05,
+      "loss": 0.7175,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 0.7383,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0953846153846155e-05,
+      "loss": 0.787,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0974358974358977e-05,
+      "loss": 0.7965,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0994871794871795e-05,
+      "loss": 0.6714,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1015384615384616e-05,
+      "loss": 0.7028,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1035897435897436e-05,
+      "loss": 0.6535,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1056410256410257e-05,
+      "loss": 0.7566,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 0.7882,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1097435897435897e-05,
+      "loss": 0.706,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.111794871794872e-05,
+      "loss": 0.724,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.113846153846154e-05,
+      "loss": 0.7396,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1158974358974359e-05,
+      "loss": 0.7004,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.117948717948718e-05,
+      "loss": 0.6533,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.6738,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1220512820512822e-05,
+      "loss": 0.7861,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1241025641025642e-05,
+      "loss": 0.6539,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.126153846153846e-05,
+      "loss": 0.7354,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.6756,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1302564102564105e-05,
+      "loss": 0.7233,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1323076923076924e-05,
+      "loss": 0.626,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1343589743589744e-05,
+      "loss": 0.6682,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1364102564102566e-05,
+      "loss": 0.7424,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1384615384615385e-05,
+      "loss": 0.7163,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1405128205128207e-05,
+      "loss": 0.708,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1425641025641026e-05,
+      "loss": 0.7524,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1446153846153846e-05,
+      "loss": 0.7047,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 0.804,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1487179487179487e-05,
+      "loss": 0.6301,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1507692307692309e-05,
+      "loss": 0.7165,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.152820512820513e-05,
+      "loss": 0.6834,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.154871794871795e-05,
+      "loss": 0.6772,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.156923076923077e-05,
+      "loss": 0.745,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1589743589743592e-05,
+      "loss": 0.7516,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1610256410256411e-05,
+      "loss": 0.8398,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1630769230769231e-05,
+      "loss": 0.7281,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1651282051282052e-05,
+      "loss": 0.6486,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1671794871794872e-05,
+      "loss": 0.7041,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 0.6782,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1712820512820513e-05,
+      "loss": 0.6348,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 0.7873,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1753846153846155e-05,
+      "loss": 0.7507,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1774358974358974e-05,
+      "loss": 0.6748,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 0.6766,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1815384615384617e-05,
+      "loss": 0.7507,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1835897435897437e-05,
+      "loss": 0.7335,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1856410256410257e-05,
+      "loss": 0.6174,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1876923076923076e-05,
+      "loss": 0.7269,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1897435897435898e-05,
+      "loss": 0.6964,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.191794871794872e-05,
+      "loss": 0.7223,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1938461538461539e-05,
+      "loss": 0.6708,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.195897435897436e-05,
+      "loss": 0.5673,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1979487179487182e-05,
+      "loss": 0.7501,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6679,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2020512820512822e-05,
+      "loss": 0.7828,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2041025641025641e-05,
+      "loss": 0.7564,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2061538461538462e-05,
+      "loss": 0.7523,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2082051282051284e-05,
+      "loss": 0.7174,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2102564102564102e-05,
+      "loss": 0.7202,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2123076923076924e-05,
+      "loss": 0.7385,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2143589743589745e-05,
+      "loss": 0.6996,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2164102564102565e-05,
+      "loss": 0.6646,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2184615384615386e-05,
+      "loss": 0.7108,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2205128205128208e-05,
+      "loss": 0.6535,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2225641025641026e-05,
+      "loss": 0.6978,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2246153846153847e-05,
+      "loss": 0.7843,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 0.6585,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2287179487179488e-05,
+      "loss": 0.7202,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.6662,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2328205128205128e-05,
+      "loss": 0.7774,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.234871794871795e-05,
+      "loss": 0.7699,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2369230769230771e-05,
+      "loss": 0.7268,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.238974358974359e-05,
+      "loss": 0.7353,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2410256410256412e-05,
+      "loss": 0.7629,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.243076923076923e-05,
+      "loss": 0.7116,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2451282051282053e-05,
+      "loss": 0.6878,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2471794871794873e-05,
+      "loss": 0.7358,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2492307692307692e-05,
+      "loss": 0.654,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2512820512820514e-05,
+      "loss": 0.5858,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 0.7628,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2553846153846155e-05,
+      "loss": 0.7631,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2574358974358975e-05,
+      "loss": 0.7801,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2594871794871797e-05,
+      "loss": 0.7292,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2615384615384616e-05,
+      "loss": 0.7477,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2635897435897438e-05,
+      "loss": 0.6181,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2656410256410257e-05,
+      "loss": 0.7996,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2676923076923077e-05,
+      "loss": 0.661,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2697435897435899e-05,
+      "loss": 0.7854,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2717948717948718e-05,
+      "loss": 0.7276,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.273846153846154e-05,
+      "loss": 0.6228,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.275897435897436e-05,
+      "loss": 0.6659,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.277948717948718e-05,
+      "loss": 0.6617,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.7241,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 0.7533,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2841025641025642e-05,
+      "loss": 0.804,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2861538461538462e-05,
+      "loss": 0.6391,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2882051282051283e-05,
+      "loss": 0.5952,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2902564102564103e-05,
+      "loss": 0.8256,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 0.6544,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2943589743589744e-05,
+      "loss": 0.6749,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2964102564102566e-05,
+      "loss": 0.7227,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2984615384615386e-05,
+      "loss": 0.654,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3005128205128205e-05,
+      "loss": 0.696,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3025641025641027e-05,
+      "loss": 0.6578,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3046153846153846e-05,
+      "loss": 0.7449,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 0.6933,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3087179487179488e-05,
+      "loss": 0.7915,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3107692307692307e-05,
+      "loss": 0.6744,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.312820512820513e-05,
+      "loss": 0.6323,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3148717948717951e-05,
+      "loss": 0.7355,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.316923076923077e-05,
+      "loss": 0.6977,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.318974358974359e-05,
+      "loss": 0.6497,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3210256410256413e-05,
+      "loss": 0.6112,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3230769230769231e-05,
+      "loss": 0.7682,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3251282051282053e-05,
+      "loss": 0.5594,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3271794871794872e-05,
+      "loss": 0.7315,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3292307692307692e-05,
+      "loss": 0.8313,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3312820512820515e-05,
+      "loss": 0.7828,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.6552,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3353846153846155e-05,
+      "loss": 0.7177,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3374358974358976e-05,
+      "loss": 0.721,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3394871794871796e-05,
+      "loss": 0.6739,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3415384615384617e-05,
+      "loss": 0.7059,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3435897435897435e-05,
+      "loss": 0.7018,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3456410256410257e-05,
+      "loss": 0.6738,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3476923076923078e-05,
+      "loss": 0.6833,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3497435897435898e-05,
+      "loss": 0.6975,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3517948717948719e-05,
+      "loss": 0.6497,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.6728,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.355897435897436e-05,
+      "loss": 0.7057,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3579487179487182e-05,
+      "loss": 0.7089,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.7295,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.362051282051282e-05,
+      "loss": 0.6633,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3641025641025643e-05,
+      "loss": 0.6111,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3661538461538461e-05,
+      "loss": 0.6263,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3682051282051284e-05,
+      "loss": 0.6559,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3702564102564104e-05,
+      "loss": 0.7045,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3723076923076923e-05,
+      "loss": 0.7727,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3743589743589745e-05,
+      "loss": 0.6893,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3764102564102567e-05,
+      "loss": 0.7302,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3784615384615386e-05,
+      "loss": 0.6476,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3805128205128206e-05,
+      "loss": 0.6096,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3825641025641028e-05,
+      "loss": 0.7439,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.77,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 0.7272,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3887179487179488e-05,
+      "loss": 0.6471,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3907692307692308e-05,
+      "loss": 0.6845,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.392820512820513e-05,
+      "loss": 0.6363,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3948717948717949e-05,
+      "loss": 0.6982,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3969230769230771e-05,
+      "loss": 0.6576,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3989743589743591e-05,
+      "loss": 0.7072,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4010256410256412e-05,
+      "loss": 0.7675,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4030769230769232e-05,
+      "loss": 0.694,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.405128205128205e-05,
+      "loss": 0.7384,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4071794871794873e-05,
+      "loss": 0.7985,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4092307692307693e-05,
+      "loss": 0.734,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4112820512820514e-05,
+      "loss": 0.7428,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 0.6261,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.6974,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4174358974358975e-05,
+      "loss": 0.7175,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4194871794871797e-05,
+      "loss": 0.725,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4215384615384617e-05,
+      "loss": 0.6739,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4235897435897436e-05,
+      "loss": 0.7176,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4256410256410258e-05,
+      "loss": 0.6691,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4276923076923077e-05,
+      "loss": 0.7236,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4297435897435899e-05,
+      "loss": 0.8044,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.431794871794872e-05,
+      "loss": 0.7474,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4338461538461538e-05,
+      "loss": 0.6471,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.7482,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4379487179487182e-05,
+      "loss": 0.7706,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.6723,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4420512820512821e-05,
+      "loss": 0.66,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4441025641025644e-05,
+      "loss": 0.6571,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 0.7063,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4482051282051284e-05,
+      "loss": 0.7178,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4502564102564103e-05,
+      "loss": 0.6906,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4523076923076923e-05,
+      "loss": 0.7051,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4543589743589746e-05,
+      "loss": 0.6884,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4564102564102564e-05,
+      "loss": 0.6598,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4584615384615386e-05,
+      "loss": 0.7632,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4605128205128207e-05,
+      "loss": 0.6662,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4625641025641027e-05,
+      "loss": 0.7657,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4646153846153848e-05,
+      "loss": 0.6634,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.7109,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4687179487179488e-05,
+      "loss": 0.7174,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4707692307692309e-05,
+      "loss": 0.7231,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.472820512820513e-05,
+      "loss": 0.6116,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.474871794871795e-05,
+      "loss": 0.6774,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.6524,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.478974358974359e-05,
+      "loss": 0.7362,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4810256410256412e-05,
+      "loss": 0.6732,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4830769230769233e-05,
+      "loss": 0.6528,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4851282051282052e-05,
+      "loss": 0.7529,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 0.6483,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4892307692307692e-05,
+      "loss": 0.7392,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4912820512820515e-05,
+      "loss": 0.5597,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 0.7516,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4953846153846154e-05,
+      "loss": 0.6738,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4974358974358976e-05,
+      "loss": 0.7066,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4994871794871798e-05,
+      "loss": 0.6656,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5015384615384617e-05,
+      "loss": 0.6618,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5035897435897437e-05,
+      "loss": 0.7091,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5056410256410257e-05,
+      "loss": 0.5794,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 0.7436,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.50974358974359e-05,
+      "loss": 0.7052,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5117948717948719e-05,
+      "loss": 0.7071,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5138461538461539e-05,
+      "loss": 0.7394,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5158974358974361e-05,
+      "loss": 0.8079,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.517948717948718e-05,
+      "loss": 0.7036,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.7428,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5220512820512822e-05,
+      "loss": 0.5938,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5241025641025643e-05,
+      "loss": 0.6479,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5261538461538465e-05,
+      "loss": 0.8086,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5282051282051282e-05,
+      "loss": 0.745,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5302564102564102e-05,
+      "loss": 0.7235,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5323076923076926e-05,
+      "loss": 0.6577,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5343589743589743e-05,
+      "loss": 0.6686,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5364102564102567e-05,
+      "loss": 0.7134,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.5624,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5405128205128204e-05,
+      "loss": 0.7372,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5425641025641028e-05,
+      "loss": 0.7735,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.544615384615385e-05,
+      "loss": 0.6869,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 0.6157,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.548717948717949e-05,
+      "loss": 0.587,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.550769230769231e-05,
+      "loss": 0.6802,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.552820512820513e-05,
+      "loss": 0.7701,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.554871794871795e-05,
+      "loss": 0.7776,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.556923076923077e-05,
+      "loss": 0.7499,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.558974358974359e-05,
+      "loss": 0.7224,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.561025641025641e-05,
+      "loss": 0.6055,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5630769230769232e-05,
+      "loss": 0.7593,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5651282051282052e-05,
+      "loss": 0.6497,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5671794871794873e-05,
+      "loss": 0.6364,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 0.7105,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5712820512820514e-05,
+      "loss": 0.7063,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 0.7107,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5753846153846154e-05,
+      "loss": 0.6792,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5774358974358975e-05,
+      "loss": 0.6937,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5794871794871795e-05,
+      "loss": 0.7284,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5815384615384616e-05,
+      "loss": 0.7102,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.583589743589744e-05,
+      "loss": 0.8373,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5856410256410256e-05,
+      "loss": 0.6752,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.587692307692308e-05,
+      "loss": 0.7171,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.5699,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5917948717948718e-05,
+      "loss": 0.7517,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.593846153846154e-05,
+      "loss": 0.6354,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.595897435897436e-05,
+      "loss": 0.6084,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5979487179487182e-05,
+      "loss": 0.7348,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5726,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.602051282051282e-05,
+      "loss": 0.6394,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6041025641025643e-05,
+      "loss": 0.7957,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.606153846153846e-05,
+      "loss": 0.6154,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6082051282051284e-05,
+      "loss": 0.6426,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6102564102564105e-05,
+      "loss": 0.6417,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6123076923076925e-05,
+      "loss": 0.6842,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6143589743589745e-05,
+      "loss": 0.759,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6164102564102566e-05,
+      "loss": 0.7072,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6184615384615386e-05,
+      "loss": 0.6327,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6205128205128207e-05,
+      "loss": 0.7324,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6225641025641027e-05,
+      "loss": 0.7175,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6246153846153848e-05,
+      "loss": 0.6963,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 0.6907,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.628717948717949e-05,
+      "loss": 0.6937,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 0.7604,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.632820512820513e-05,
+      "loss": 0.6385,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.634871794871795e-05,
+      "loss": 0.6665,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.636923076923077e-05,
+      "loss": 0.6467,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.638974358974359e-05,
+      "loss": 0.7826,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.6309,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.643076923076923e-05,
+      "loss": 0.7784,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6451282051282055e-05,
+      "loss": 0.7117,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6471794871794872e-05,
+      "loss": 0.6584,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6492307692307696e-05,
+      "loss": 0.5833,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6512820512820513e-05,
+      "loss": 0.6818,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 0.7095,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6553846153846157e-05,
+      "loss": 0.6996,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6574358974358974e-05,
+      "loss": 0.5918,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6594871794871798e-05,
+      "loss": 0.7089,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 0.8042,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6635897435897435e-05,
+      "loss": 0.6811,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.665641025641026e-05,
+      "loss": 0.7185,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6676923076923076e-05,
+      "loss": 0.7811,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.66974358974359e-05,
+      "loss": 0.6524,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.671794871794872e-05,
+      "loss": 0.6633,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.673846153846154e-05,
+      "loss": 0.7803,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.675897435897436e-05,
+      "loss": 0.6068,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.677948717948718e-05,
+      "loss": 0.8092,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.7516,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6820512820512822e-05,
+      "loss": 0.6856,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6841025641025643e-05,
+      "loss": 0.6667,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6861538461538463e-05,
+      "loss": 0.6302,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6882051282051283e-05,
+      "loss": 0.6707,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6902564102564104e-05,
+      "loss": 0.6302,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.6225,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6943589743589745e-05,
+      "loss": 0.7178,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6964102564102565e-05,
+      "loss": 0.7333,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6984615384615385e-05,
+      "loss": 0.6847,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7005128205128206e-05,
+      "loss": 0.7394,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7025641025641026e-05,
+      "loss": 0.5808,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7046153846153847e-05,
+      "loss": 0.729,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 0.7079,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7087179487179487e-05,
+      "loss": 0.7355,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.710769230769231e-05,
+      "loss": 0.6653,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7128205128205128e-05,
+      "loss": 0.7057,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.714871794871795e-05,
+      "loss": 0.7503,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7169230769230772e-05,
+      "loss": 0.6633,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.718974358974359e-05,
+      "loss": 0.839,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7210256410256413e-05,
+      "loss": 0.8463,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 0.7556,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.725128205128205e-05,
+      "loss": 0.6757,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7271794871794874e-05,
+      "loss": 0.6297,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.729230769230769e-05,
+      "loss": 0.7222,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7312820512820515e-05,
+      "loss": 0.6481,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.7151,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7353846153846156e-05,
+      "loss": 0.7135,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7374358974358976e-05,
+      "loss": 0.7487,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7394871794871797e-05,
+      "loss": 0.6877,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7415384615384617e-05,
+      "loss": 0.7064,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 0.6471,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7456410256410258e-05,
+      "loss": 0.6778,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.747692307692308e-05,
+      "loss": 0.5813,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.74974358974359e-05,
+      "loss": 0.6979,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.751794871794872e-05,
+      "loss": 0.6725,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 0.7503,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.755897435897436e-05,
+      "loss": 0.6779,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.757948717948718e-05,
+      "loss": 0.7156,
+      "step": 857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.76e-05,
+      "loss": 0.6455,
+      "step": 858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.762051282051282e-05,
+      "loss": 0.6632,
+      "step": 859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7641025641025642e-05,
+      "loss": 0.6711,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7661538461538462e-05,
+      "loss": 0.7508,
+      "step": 861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7682051282051283e-05,
+      "loss": 0.7081,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7702564102564103e-05,
+      "loss": 0.6267,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7723076923076927e-05,
+      "loss": 0.739,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7743589743589744e-05,
+      "loss": 0.7006,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7764102564102564e-05,
+      "loss": 0.7201,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7784615384615388e-05,
+      "loss": 0.7167,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7805128205128205e-05,
+      "loss": 0.6661,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.782564102564103e-05,
+      "loss": 0.6221,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 0.7405,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 0.6379,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.788717948717949e-05,
+      "loss": 0.7667,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7907692307692307e-05,
+      "loss": 0.6472,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.792820512820513e-05,
+      "loss": 0.6789,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.6588,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.796923076923077e-05,
+      "loss": 0.7116,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7989743589743592e-05,
+      "loss": 0.5679,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8010256410256412e-05,
+      "loss": 0.7863,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8030769230769233e-05,
+      "loss": 0.7517,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8051282051282053e-05,
+      "loss": 0.624,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8071794871794874e-05,
+      "loss": 0.7218,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8092307692307694e-05,
+      "loss": 0.6234,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8112820512820514e-05,
+      "loss": 0.5534,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 0.626,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 0.7226,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8174358974358976e-05,
+      "loss": 0.7323,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8194871794871796e-05,
+      "loss": 0.6435,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8215384615384616e-05,
+      "loss": 0.6527,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8235897435897437e-05,
+      "loss": 0.6939,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8256410256410257e-05,
+      "loss": 0.6585,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8276923076923078e-05,
+      "loss": 0.6915,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8297435897435898e-05,
+      "loss": 0.7417,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.831794871794872e-05,
+      "loss": 0.68,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8338461538461542e-05,
+      "loss": 0.6547,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.835897435897436e-05,
+      "loss": 0.6791,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.837948717948718e-05,
+      "loss": 0.6852,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.6637,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.842051282051282e-05,
+      "loss": 0.7446,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8441025641025644e-05,
+      "loss": 0.6119,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.6379,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.848205128205128e-05,
+      "loss": 0.6704,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8502564102564105e-05,
+      "loss": 0.7318,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8523076923076922e-05,
+      "loss": 0.6909,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8543589743589746e-05,
+      "loss": 0.6042,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8564102564102567e-05,
+      "loss": 0.6853,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8584615384615387e-05,
+      "loss": 0.6942,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8605128205128207e-05,
+      "loss": 0.7501,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8625641025641028e-05,
+      "loss": 0.7317,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8646153846153848e-05,
+      "loss": 0.7307,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.688,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.868717948717949e-05,
+      "loss": 0.7774,
+      "step": 911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.870769230769231e-05,
+      "loss": 0.7475,
+      "step": 912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.872820512820513e-05,
+      "loss": 0.6484,
+      "step": 913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.874871794871795e-05,
+      "loss": 0.697,
+      "step": 914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 0.6364,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.878974358974359e-05,
+      "loss": 0.7051,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.881025641025641e-05,
+      "loss": 0.6496,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8830769230769232e-05,
+      "loss": 0.7203,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8851282051282052e-05,
+      "loss": 0.7362,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8871794871794873e-05,
+      "loss": 0.7528,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8892307692307693e-05,
+      "loss": 0.8081,
+      "step": 921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8912820512820513e-05,
+      "loss": 0.686,
+      "step": 922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 0.7836,
+      "step": 923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8953846153846158e-05,
+      "loss": 0.7045,
+      "step": 924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.7464,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8994871794871795e-05,
+      "loss": 0.7759,
+      "step": 926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901538461538462e-05,
+      "loss": 0.7002,
+      "step": 927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9035897435897436e-05,
+      "loss": 0.7103,
+      "step": 928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905641025641026e-05,
+      "loss": 0.7029,
+      "step": 929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 0.7067,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9097435897435897e-05,
+      "loss": 0.8451,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911794871794872e-05,
+      "loss": 0.6528,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9138461538461538e-05,
+      "loss": 0.6847,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9158974358974362e-05,
+      "loss": 0.6182,
+      "step": 934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9179487179487182e-05,
+      "loss": 0.699,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.6096,
+      "step": 936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9220512820512823e-05,
+      "loss": 0.6809,
+      "step": 937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9241025641025643e-05,
+      "loss": 0.632,
+      "step": 938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9261538461538464e-05,
+      "loss": 0.5565,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9282051282051284e-05,
+      "loss": 0.7441,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9302564102564105e-05,
+      "loss": 0.6564,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323076923076925e-05,
+      "loss": 0.7795,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9343589743589745e-05,
+      "loss": 0.6993,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9364102564102566e-05,
+      "loss": 0.6403,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 0.7891,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9405128205128207e-05,
+      "loss": 0.668,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9425641025641027e-05,
+      "loss": 0.7121,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9446153846153847e-05,
+      "loss": 0.7029,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 0.5905,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.7593,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.950769230769231e-05,
+      "loss": 0.8032,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.952820512820513e-05,
+      "loss": 0.622,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.954871794871795e-05,
+      "loss": 0.6664,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9569230769230773e-05,
+      "loss": 0.7714,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.958974358974359e-05,
+      "loss": 0.6634,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961025641025641e-05,
+      "loss": 0.6864,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9630769230769234e-05,
+      "loss": 0.635,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.965128205128205e-05,
+      "loss": 0.7415,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9671794871794875e-05,
+      "loss": 0.6448,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 0.562,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9712820512820513e-05,
+      "loss": 0.7069,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 0.7521,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9753846153846153e-05,
+      "loss": 0.6846,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9774358974358977e-05,
+      "loss": 0.7085,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9794871794871798e-05,
+      "loss": 0.7193,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9815384615384618e-05,
+      "loss": 0.6303,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.983589743589744e-05,
+      "loss": 0.7408,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.985641025641026e-05,
+      "loss": 0.7667,
+      "step": 968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.987692307692308e-05,
+      "loss": 0.6857,
+      "step": 969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98974358974359e-05,
+      "loss": 0.791,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.991794871794872e-05,
+      "loss": 0.6738,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.993846153846154e-05,
+      "loss": 0.6386,
+      "step": 972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995897435897436e-05,
+      "loss": 0.6941,
+      "step": 973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997948717948718e-05,
+      "loss": 0.6722,
+      "step": 974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.734,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999995032964e-05,
+      "loss": 0.6673,
+      "step": 976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999801318554e-05,
+      "loss": 0.6836,
+      "step": 977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999955296674e-05,
+      "loss": 0.7034,
+      "step": 978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999920527421e-05,
+      "loss": 0.7169,
+      "step": 979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999998758240956e-05,
+      "loss": 0.5749,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999998211866995e-05,
+      "loss": 0.7012,
+      "step": 981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997566152323e-05,
+      "loss": 0.729,
+      "step": 982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999996821096948e-05,
+      "loss": 0.6881,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999995976700886e-05,
+      "loss": 0.6924,
+      "step": 984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999995032964135e-05,
+      "loss": 0.8047,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999993989886706e-05,
+      "loss": 0.6968,
+      "step": 986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999992847468613e-05,
+      "loss": 0.685,
+      "step": 987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999991605709863e-05,
+      "loss": 0.6352,
+      "step": 988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999990264610475e-05,
+      "loss": 0.6454,
+      "step": 989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999988824170456e-05,
+      "loss": 0.8039,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999998728438982e-05,
+      "loss": 0.6372,
+      "step": 991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999998564526859e-05,
+      "loss": 0.8576,
+      "step": 992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999983906806773e-05,
+      "loss": 0.7553,
+      "step": 993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999982069004392e-05,
+      "loss": 0.6149,
+      "step": 994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999998013186146e-05,
+      "loss": 0.689,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997809537801e-05,
+      "loss": 0.5812,
+      "step": 996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999975959554044e-05,
+      "loss": 0.7332,
+      "step": 997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997372438959e-05,
+      "loss": 0.665,
+      "step": 998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999971389884677e-05,
+      "loss": 0.6757,
+      "step": 999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999996895603932e-05,
+      "loss": 0.698,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999966422853544e-05,
+      "loss": 0.6872,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999996379032738e-05,
+      "loss": 0.6628,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999961058460847e-05,
+      "loss": 0.7817,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999958227253976e-05,
+      "loss": 0.7176,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999955296706797e-05,
+      "loss": 0.7494,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999952266819333e-05,
+      "loss": 0.5908,
+      "step": 1006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999949137591617e-05,
+      "loss": 0.7363,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999945909023685e-05,
+      "loss": 0.6721,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999942581115562e-05,
+      "loss": 0.699,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999939153867287e-05,
+      "loss": 0.7742,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999993562727889e-05,
+      "loss": 0.6146,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999932001350403e-05,
+      "loss": 0.7472,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999992827608187e-05,
+      "loss": 0.7125,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999992445147332e-05,
+      "loss": 0.5204,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99999205275248e-05,
+      "loss": 0.626,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999991650423634e-05,
+      "loss": 0.6928,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999912381607988e-05,
+      "loss": 0.7149,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999908159639778e-05,
+      "loss": 0.6276,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999903838331757e-05,
+      "loss": 0.6037,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999899417683963e-05,
+      "loss": 0.736,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999894897696443e-05,
+      "loss": 0.6778,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999890278369244e-05,
+      "loss": 0.647,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999885559702408e-05,
+      "loss": 0.7315,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999880741695984e-05,
+      "loss": 0.6944,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999875824350018e-05,
+      "loss": 0.5862,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999987080766456e-05,
+      "loss": 0.673,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999986569163966e-05,
+      "loss": 0.5673,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999986047627537e-05,
+      "loss": 0.6458,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999855161571736e-05,
+      "loss": 0.6872,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999984974752882e-05,
+      "loss": 0.6648,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999984423414667e-05,
+      "loss": 0.778,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999838621425344e-05,
+      "loss": 0.6729,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999832909364894e-05,
+      "loss": 0.6685,
+      "step": 1033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999982709796538e-05,
+      "loss": 0.6737,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999821187226853e-05,
+      "loss": 0.6739,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999981517714938e-05,
+      "loss": 0.7221,
+      "step": 1036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999980906773302e-05,
+      "loss": 0.7327,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999980285897783e-05,
+      "loss": 0.7822,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999979655088387e-05,
+      "loss": 0.6343,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999790143451207e-05,
+      "loss": 0.7407,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999783636679902e-05,
+      "loss": 0.6384,
+      "step": 1041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999777030570024e-05,
+      "loss": 0.5656,
+      "step": 1042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999770325121632e-05,
+      "loss": 0.6613,
+      "step": 1043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99997635203348e-05,
+      "loss": 0.6127,
+      "step": 1044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999975661620959e-05,
+      "loss": 0.6534,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999749612746072e-05,
+      "loss": 0.7233,
+      "step": 1046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999742509944316e-05,
+      "loss": 0.6699,
+      "step": 1047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999973530780439e-05,
+      "loss": 0.6568,
+      "step": 1048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999728006326375e-05,
+      "loss": 0.6829,
+      "step": 1049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999720605510327e-05,
+      "loss": 0.661,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999713105356336e-05,
+      "loss": 0.7134,
+      "step": 1051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999705505864466e-05,
+      "loss": 0.6918,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99996978070348e-05,
+      "loss": 0.7228,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999969000886741e-05,
+      "loss": 0.6836,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999682111362368e-05,
+      "loss": 0.732,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999674114519762e-05,
+      "loss": 0.7245,
+      "step": 1056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999966601833967e-05,
+      "loss": 0.6664,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999657822822167e-05,
+      "loss": 0.7225,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999964952796734e-05,
+      "loss": 0.6824,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999641133775267e-05,
+      "loss": 0.6485,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999632640246033e-05,
+      "loss": 0.6962,
+      "step": 1061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999624047379726e-05,
+      "loss": 0.6862,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999615355176427e-05,
+      "loss": 0.6668,
+      "step": 1063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999606563636225e-05,
+      "loss": 0.6804,
+      "step": 1064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999597672759204e-05,
+      "loss": 0.6882,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999588682545455e-05,
+      "loss": 0.6518,
+      "step": 1066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999579592995064e-05,
+      "loss": 0.6917,
+      "step": 1067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999957040410813e-05,
+      "loss": 0.749,
+      "step": 1068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999956111588473e-05,
+      "loss": 0.688,
+      "step": 1069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999955172832497e-05,
+      "loss": 0.7041,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999542241428936e-05,
+      "loss": 0.5989,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999532655196723e-05,
+      "loss": 0.6461,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999522969628428e-05,
+      "loss": 0.639,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999951318472415e-05,
+      "loss": 0.7268,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999503300483978e-05,
+      "loss": 0.7108,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999493316908017e-05,
+      "loss": 0.6773,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999483233996362e-05,
+      "loss": 0.6741,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999473051749115e-05,
+      "loss": 0.6765,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999462770166378e-05,
+      "loss": 0.6267,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999452389248254e-05,
+      "loss": 0.6976,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999441908994843e-05,
+      "loss": 0.7651,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999943132940625e-05,
+      "loss": 0.7683,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999420650482582e-05,
+      "loss": 0.7161,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999409872223943e-05,
+      "loss": 0.6889,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999398994630442e-05,
+      "loss": 0.7031,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999388017702186e-05,
+      "loss": 0.6483,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999937694143928e-05,
+      "loss": 0.6477,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999365765841842e-05,
+      "loss": 0.5601,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999354490909977e-05,
+      "loss": 0.5931,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99993431166438e-05,
+      "loss": 0.7704,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999331643043424e-05,
+      "loss": 0.6987,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999320070108963e-05,
+      "loss": 0.6667,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999308397840525e-05,
+      "loss": 0.7103,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999296626238237e-05,
+      "loss": 0.7851,
+      "step": 1094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999284755302213e-05,
+      "loss": 0.5959,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999272785032565e-05,
+      "loss": 0.7184,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999260715429415e-05,
+      "loss": 0.6006,
+      "step": 1097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999248546492886e-05,
+      "loss": 0.6951,
+      "step": 1098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999236278223094e-05,
+      "loss": 0.6683,
+      "step": 1099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999223910620164e-05,
+      "loss": 0.7098,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999921144368422e-05,
+      "loss": 0.7148,
+      "step": 1101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999198877415383e-05,
+      "loss": 0.6621,
+      "step": 1102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999918621181378e-05,
+      "loss": 0.7222,
+      "step": 1103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999173446879536e-05,
+      "loss": 0.7325,
+      "step": 1104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999160582612775e-05,
+      "loss": 0.7145,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999914761901363e-05,
+      "loss": 0.7064,
+      "step": 1106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999134556082225e-05,
+      "loss": 0.7807,
+      "step": 1107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999121393818693e-05,
+      "loss": 0.695,
+      "step": 1108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999108132223163e-05,
+      "loss": 0.6332,
+      "step": 1109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999094771295767e-05,
+      "loss": 0.6698,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999908131103664e-05,
+      "loss": 0.6348,
+      "step": 1111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999067751445912e-05,
+      "loss": 0.6758,
+      "step": 1112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999905409252372e-05,
+      "loss": 0.617,
+      "step": 1113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999040334270198e-05,
+      "loss": 0.6209,
+      "step": 1114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999026476685484e-05,
+      "loss": 0.6607,
+      "step": 1115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999901251976972e-05,
+      "loss": 0.6575,
+      "step": 1116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998998463523036e-05,
+      "loss": 0.6983,
+      "step": 1117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998984307945576e-05,
+      "loss": 0.7281,
+      "step": 1118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999897005303748e-05,
+      "loss": 0.6209,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999895569879889e-05,
+      "loss": 0.5997,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998941245229955e-05,
+      "loss": 0.7371,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998926692330806e-05,
+      "loss": 0.6318,
+      "step": 1122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998912040101595e-05,
+      "loss": 0.6734,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998897288542466e-05,
+      "loss": 0.7069,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998882437653565e-05,
+      "loss": 0.6406,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998867487435042e-05,
+      "loss": 0.6973,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998852437887043e-05,
+      "loss": 0.697,
+      "step": 1127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999883728900972e-05,
+      "loss": 0.7221,
+      "step": 1128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999882204080322e-05,
+      "loss": 0.6885,
+      "step": 1129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998806693267697e-05,
+      "loss": 0.6579,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998791246403305e-05,
+      "loss": 0.6141,
+      "step": 1131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998775700210195e-05,
+      "loss": 0.7442,
+      "step": 1132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999876005468852e-05,
+      "loss": 0.6686,
+      "step": 1133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998744309838437e-05,
+      "loss": 0.6745,
+      "step": 1134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998728465660105e-05,
+      "loss": 0.7663,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998712522153675e-05,
+      "loss": 0.6149,
+      "step": 1136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998696479319314e-05,
+      "loss": 0.5896,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998680337157176e-05,
+      "loss": 0.5928,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998664095667418e-05,
+      "loss": 0.6651,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999864775485021e-05,
+      "loss": 0.5919,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998631314705706e-05,
+      "loss": 0.7141,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998614775234078e-05,
+      "loss": 0.6372,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999859813643548e-05,
+      "loss": 0.6553,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998581398310088e-05,
+      "loss": 0.6159,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999856456085806e-05,
+      "loss": 0.6715,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998547624079564e-05,
+      "loss": 0.6119,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998530587974775e-05,
+      "loss": 0.6781,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998513452543853e-05,
+      "loss": 0.694,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998496217786978e-05,
+      "loss": 0.6546,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998478883704312e-05,
+      "loss": 0.616,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998461450296035e-05,
+      "loss": 0.659,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998443917562314e-05,
+      "loss": 0.6911,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998426285503327e-05,
+      "loss": 0.69,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998408554119247e-05,
+      "loss": 0.7242,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999839072341025e-05,
+      "loss": 0.6853,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999837279337652e-05,
+      "loss": 0.7704,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998354764018222e-05,
+      "loss": 0.7451,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998336635335547e-05,
+      "loss": 0.66,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998318407328668e-05,
+      "loss": 0.6585,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998300079997767e-05,
+      "loss": 0.7679,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998281653343034e-05,
+      "loss": 0.6871,
+      "step": 1161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998263127364638e-05,
+      "loss": 0.7801,
+      "step": 1162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998244502062775e-05,
+      "loss": 0.6025,
+      "step": 1163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998225777437626e-05,
+      "loss": 0.6848,
+      "step": 1164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998206953489374e-05,
+      "loss": 0.8307,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999818803021821e-05,
+      "loss": 0.5929,
+      "step": 1166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998169007624323e-05,
+      "loss": 0.6747,
+      "step": 1167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99981498857079e-05,
+      "loss": 0.547,
+      "step": 1168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998130664469127e-05,
+      "loss": 0.6386,
+      "step": 1169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99981113439082e-05,
+      "loss": 0.6876,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999809192402531e-05,
+      "loss": 0.7651,
+      "step": 1171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998072404820648e-05,
+      "loss": 0.7488,
+      "step": 1172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998052786294413e-05,
+      "loss": 0.5844,
+      "step": 1173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998033068446795e-05,
+      "loss": 0.7121,
+      "step": 1174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998013251277988e-05,
+      "loss": 0.6443,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997993334788195e-05,
+      "loss": 0.8277,
+      "step": 1176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999797331897761e-05,
+      "loss": 0.5711,
+      "step": 1177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997953203846433e-05,
+      "loss": 0.6774,
+      "step": 1178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997932989394864e-05,
+      "loss": 0.6721,
+      "step": 1179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997912675623104e-05,
+      "loss": 0.7132,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997892262531352e-05,
+      "loss": 0.7386,
+      "step": 1181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999787175011981e-05,
+      "loss": 0.6315,
+      "step": 1182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999785113838869e-05,
+      "loss": 0.5873,
+      "step": 1183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999783042733819e-05,
+      "loss": 0.6495,
+      "step": 1184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997809616968514e-05,
+      "loss": 0.6005,
+      "step": 1185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997788707279877e-05,
+      "loss": 0.6566,
+      "step": 1186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999776769827248e-05,
+      "loss": 0.575,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999774658994653e-05,
+      "loss": 0.6552,
+      "step": 1188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997725382302248e-05,
+      "loss": 0.6366,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999770407533983e-05,
+      "loss": 0.6369,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997682669059494e-05,
+      "loss": 0.62,
+      "step": 1191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997661163461458e-05,
+      "loss": 0.6539,
+      "step": 1192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997639558545926e-05,
+      "loss": 0.6098,
+      "step": 1193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999761785431312e-05,
+      "loss": 0.6352,
+      "step": 1194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997596050763253e-05,
+      "loss": 0.7546,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999757414789654e-05,
+      "loss": 0.7557,
+      "step": 1196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99975521457132e-05,
+      "loss": 0.6319,
+      "step": 1197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999753004421345e-05,
+      "loss": 0.6229,
+      "step": 1198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997507843397514e-05,
+      "loss": 0.6948,
+      "step": 1199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997485543265613e-05,
+      "loss": 0.6628,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999746314381796e-05,
+      "loss": 0.6003,
+      "step": 1201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997440645054782e-05,
+      "loss": 0.7224,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997418046976307e-05,
+      "loss": 0.6689,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997395349582755e-05,
+      "loss": 0.6365,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999737255287435e-05,
+      "loss": 0.6203,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997349656851323e-05,
+      "loss": 0.686,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997326661513897e-05,
+      "loss": 0.8245,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997303566862306e-05,
+      "loss": 0.6899,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997280372896776e-05,
+      "loss": 0.6167,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997257079617534e-05,
+      "loss": 0.6798,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997233687024814e-05,
+      "loss": 0.6559,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997210195118854e-05,
+      "loss": 0.7744,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999718660389988e-05,
+      "loss": 0.6055,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997162913368127e-05,
+      "loss": 0.7338,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999713912352384e-05,
+      "loss": 0.7292,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999711523436724e-05,
+      "loss": 0.5956,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997091245898575e-05,
+      "loss": 0.7052,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997067158118078e-05,
+      "loss": 0.6462,
+      "step": 1218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997042971025994e-05,
+      "loss": 0.6223,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997018684622556e-05,
+      "loss": 0.6984,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996994298908012e-05,
+      "loss": 0.7874,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996969813882602e-05,
+      "loss": 0.6351,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996945229546564e-05,
+      "loss": 0.7085,
+      "step": 1223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996920545900154e-05,
+      "loss": 0.6369,
+      "step": 1224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996895762943604e-05,
+      "loss": 0.6703,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999687088067717e-05,
+      "loss": 0.652,
+      "step": 1226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999684589910109e-05,
+      "loss": 0.67,
+      "step": 1227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996820818215622e-05,
+      "loss": 0.5958,
+      "step": 1228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996795638021013e-05,
+      "loss": 0.6064,
+      "step": 1229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996770358517506e-05,
+      "loss": 0.6556,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999674497970536e-05,
+      "loss": 0.7048,
+      "step": 1231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999671950158482e-05,
+      "loss": 0.7282,
+      "step": 1232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996693924156146e-05,
+      "loss": 0.7,
+      "step": 1233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999666824741959e-05,
+      "loss": 0.6462,
+      "step": 1234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996642471375406e-05,
+      "loss": 0.6062,
+      "step": 1235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999661659602385e-05,
+      "loss": 0.7652,
+      "step": 1236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999659062136518e-05,
+      "loss": 0.7022,
+      "step": 1237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996564547399653e-05,
+      "loss": 0.6654,
+      "step": 1238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999653837412753e-05,
+      "loss": 0.7834,
+      "step": 1239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996512101549066e-05,
+      "loss": 0.7036,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996485729664528e-05,
+      "loss": 0.629,
+      "step": 1241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996459258474176e-05,
+      "loss": 0.7525,
+      "step": 1242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999643268797827e-05,
+      "loss": 0.6671,
+      "step": 1243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999640601817708e-05,
+      "loss": 0.6283,
+      "step": 1244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996379249070865e-05,
+      "loss": 0.6406,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996352380659892e-05,
+      "loss": 0.7669,
+      "step": 1246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996325412944434e-05,
+      "loss": 0.6657,
+      "step": 1247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999629834592475e-05,
+      "loss": 0.6487,
+      "step": 1248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996271179601113e-05,
+      "loss": 0.6349,
+      "step": 1249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996243913973793e-05,
+      "loss": 0.6599,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996216549043064e-05,
+      "loss": 0.6813,
+      "step": 1251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999618908480919e-05,
+      "loss": 0.6847,
+      "step": 1252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999616152127245e-05,
+      "loss": 0.6994,
+      "step": 1253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999613385843312e-05,
+      "loss": 0.6005,
+      "step": 1254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999610609629147e-05,
+      "loss": 0.6152,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996078234847772e-05,
+      "loss": 0.6791,
+      "step": 1256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999605027410231e-05,
+      "loss": 0.6103,
+      "step": 1257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996022214055363e-05,
+      "loss": 0.7107,
+      "step": 1258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99959940547072e-05,
+      "loss": 0.72,
+      "step": 1259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999596579605811e-05,
+      "loss": 0.6826,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995937438108372e-05,
+      "loss": 0.6375,
+      "step": 1261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995908980858262e-05,
+      "loss": 0.6816,
+      "step": 1262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999588042430807e-05,
+      "loss": 0.6815,
+      "step": 1263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995851768458077e-05,
+      "loss": 0.6915,
+      "step": 1264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995823013308567e-05,
+      "loss": 0.5677,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995794158859826e-05,
+      "loss": 0.7082,
+      "step": 1266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995765205112142e-05,
+      "loss": 0.7314,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99957361520658e-05,
+      "loss": 0.6181,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995706999721087e-05,
+      "loss": 0.6485,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995677748078295e-05,
+      "loss": 0.7155,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999564839713772e-05,
+      "loss": 0.7243,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995618946899648e-05,
+      "loss": 0.5756,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999558939736437e-05,
+      "loss": 0.5903,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995559748532184e-05,
+      "loss": 0.6042,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995530000403378e-05,
+      "loss": 0.6159,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995500152978255e-05,
+      "loss": 0.6983,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995470206257107e-05,
+      "loss": 0.7138,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995440160240234e-05,
+      "loss": 0.7306,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995410014927935e-05,
+      "loss": 0.7253,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995379770320505e-05,
+      "loss": 0.704,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995349426418248e-05,
+      "loss": 0.7131,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995318983221463e-05,
+      "loss": 0.7362,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995288440730455e-05,
+      "loss": 0.6916,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999525779894553e-05,
+      "loss": 0.6078,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995227057866985e-05,
+      "loss": 0.6628,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995196217495127e-05,
+      "loss": 0.6827,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999516527783027e-05,
+      "loss": 0.651,
+      "step": 1287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995134238872714e-05,
+      "loss": 0.6172,
+      "step": 1288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995103100622767e-05,
+      "loss": 0.6323,
+      "step": 1289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995071863080744e-05,
+      "loss": 0.6536,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999504052624695e-05,
+      "loss": 0.7762,
+      "step": 1291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99950090901217e-05,
+      "loss": 0.7223,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994977554705304e-05,
+      "loss": 0.7737,
+      "step": 1293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994945919998073e-05,
+      "loss": 0.6711,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999491418600033e-05,
+      "loss": 0.659,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994882352712378e-05,
+      "loss": 0.6848,
+      "step": 1296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994850420134543e-05,
+      "loss": 0.6393,
+      "step": 1297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999481838826714e-05,
+      "loss": 0.6442,
+      "step": 1298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994786257110484e-05,
+      "loss": 0.6243,
+      "step": 1299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994754026664897e-05,
+      "loss": 0.7009,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994721696930694e-05,
+      "loss": 0.6604,
+      "step": 1301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994689267908203e-05,
+      "loss": 0.6914,
+      "step": 1302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994656739597748e-05,
+      "loss": 0.6432,
+      "step": 1303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994624111999642e-05,
+      "loss": 0.6353,
+      "step": 1304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994591385114218e-05,
+      "loss": 0.6709,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99945585589418e-05,
+      "loss": 0.6872,
+      "step": 1306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994525633482712e-05,
+      "loss": 0.6833,
+      "step": 1307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994492608737278e-05,
+      "loss": 0.5914,
+      "step": 1308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999445948470583e-05,
+      "loss": 0.6836,
+      "step": 1309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994426261388696e-05,
+      "loss": 0.5985,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999439293878621e-05,
+      "loss": 0.7226,
+      "step": 1311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994359516898692e-05,
+      "loss": 0.6766,
+      "step": 1312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994325995726492e-05,
+      "loss": 0.7195,
+      "step": 1313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994292375269925e-05,
+      "loss": 0.7024,
+      "step": 1314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994258655529336e-05,
+      "loss": 0.6592,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999422483650505e-05,
+      "loss": 0.6886,
+      "step": 1316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994190918197415e-05,
+      "loss": 0.626,
+      "step": 1317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994156900606763e-05,
+      "loss": 0.8449,
+      "step": 1318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999412278373343e-05,
+      "loss": 0.6685,
+      "step": 1319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994088567577753e-05,
+      "loss": 0.6769,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999405425214008e-05,
+      "loss": 0.6681,
+      "step": 1321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994019837420743e-05,
+      "loss": 0.7167,
+      "step": 1322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999398532342009e-05,
+      "loss": 0.6966,
+      "step": 1323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993950710138462e-05,
+      "loss": 0.7044,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993915997576204e-05,
+      "loss": 0.6432,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993881185733658e-05,
+      "loss": 0.6335,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993846274611172e-05,
+      "loss": 0.6134,
+      "step": 1327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999381126420909e-05,
+      "loss": 0.6079,
+      "step": 1328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999377615452776e-05,
+      "loss": 0.7209,
+      "step": 1329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993740945567535e-05,
+      "loss": 0.6619,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993705637328764e-05,
+      "loss": 0.6515,
+      "step": 1331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993670229811798e-05,
+      "loss": 0.7406,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999363472301698e-05,
+      "loss": 0.5947,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993599116944674e-05,
+      "loss": 0.6096,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999356341159523e-05,
+      "loss": 0.6425,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993527606969e-05,
+      "loss": 0.6721,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993491703066345e-05,
+      "loss": 0.7163,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999345569988761e-05,
+      "loss": 0.6557,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999341959743317e-05,
+      "loss": 0.738,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993383395703372e-05,
+      "loss": 0.6041,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993347094698577e-05,
+      "loss": 0.6799,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993310694419146e-05,
+      "loss": 0.654,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993274194865445e-05,
+      "loss": 0.655,
+      "step": 1343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993237596037832e-05,
+      "loss": 0.6426,
+      "step": 1344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993200897936673e-05,
+      "loss": 0.6619,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999316410056233e-05,
+      "loss": 0.6543,
+      "step": 1346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999312720391517e-05,
+      "loss": 0.6819,
+      "step": 1347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993090207995557e-05,
+      "loss": 0.6462,
+      "step": 1348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999305311280386e-05,
+      "loss": 0.6847,
+      "step": 1349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993015918340454e-05,
+      "loss": 0.5951,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99929786246057e-05,
+      "loss": 0.6908,
+      "step": 1351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999294123159997e-05,
+      "loss": 0.6842,
+      "step": 1352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992903739323636e-05,
+      "loss": 0.5708,
+      "step": 1353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992866147777078e-05,
+      "loss": 0.6522,
+      "step": 1354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992828456960654e-05,
+      "loss": 0.6779,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992790666874754e-05,
+      "loss": 0.5845,
+      "step": 1356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992752777519743e-05,
+      "loss": 0.7276,
+      "step": 1357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992714788895998e-05,
+      "loss": 0.6267,
+      "step": 1358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992676701003905e-05,
+      "loss": 0.7098,
+      "step": 1359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992638513843836e-05,
+      "loss": 0.6246,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999260022741617e-05,
+      "loss": 0.6666,
+      "step": 1361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992561841721287e-05,
+      "loss": 0.647,
+      "step": 1362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999252335675957e-05,
+      "loss": 0.6826,
+      "step": 1363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99924847725314e-05,
+      "loss": 0.5976,
+      "step": 1364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992446089037164e-05,
+      "loss": 0.607,
+      "step": 1365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992407306277245e-05,
+      "loss": 0.5853,
+      "step": 1366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992368424252025e-05,
+      "loss": 0.6307,
+      "step": 1367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999232944296189e-05,
+      "loss": 0.6102,
+      "step": 1368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992290362407232e-05,
+      "loss": 0.6502,
+      "step": 1369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992251182588432e-05,
+      "loss": 0.6347,
+      "step": 1370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992211903505892e-05,
+      "loss": 0.5612,
+      "step": 1371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999217252515999e-05,
+      "loss": 0.6007,
+      "step": 1372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992133047551117e-05,
+      "loss": 0.7263,
+      "step": 1373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992093470679673e-05,
+      "loss": 0.6519,
+      "step": 1374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999205379454605e-05,
+      "loss": 0.6461,
+      "step": 1375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992014019150635e-05,
+      "loss": 0.6844,
+      "step": 1376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991974144493832e-05,
+      "loss": 0.7329,
+      "step": 1377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991934170576027e-05,
+      "loss": 0.5832,
+      "step": 1378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999189409739763e-05,
+      "loss": 0.6578,
+      "step": 1379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999185392495903e-05,
+      "loss": 0.6507,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991813653260624e-05,
+      "loss": 0.7002,
+      "step": 1381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991773282302824e-05,
+      "loss": 0.6941,
+      "step": 1382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999173281208602e-05,
+      "loss": 0.5982,
+      "step": 1383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999169224261061e-05,
+      "loss": 0.6889,
+      "step": 1384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991651573877012e-05,
+      "loss": 0.5672,
+      "step": 1385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991610805885622e-05,
+      "loss": 0.5872,
+      "step": 1386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991569938636845e-05,
+      "loss": 0.6195,
+      "step": 1387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991528972131084e-05,
+      "loss": 0.5754,
+      "step": 1388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991487906368755e-05,
+      "loss": 0.643,
+      "step": 1389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991446741350255e-05,
+      "loss": 0.6471,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991405477076e-05,
+      "loss": 0.7287,
+      "step": 1391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991364113546398e-05,
+      "loss": 0.6369,
+      "step": 1392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991322650761864e-05,
+      "loss": 0.7043,
+      "step": 1393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99912810887228e-05,
+      "loss": 0.8212,
+      "step": 1394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999123942742963e-05,
+      "loss": 0.7209,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999119766688276e-05,
+      "loss": 0.6364,
+      "step": 1396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991155807082604e-05,
+      "loss": 0.6097,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991113848029588e-05,
+      "loss": 0.5881,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991071789724117e-05,
+      "loss": 0.6706,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991029632166616e-05,
+      "loss": 0.7258,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990987375357504e-05,
+      "loss": 0.657,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990945019297194e-05,
+      "loss": 0.5613,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990902563986114e-05,
+      "loss": 0.6509,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999086000942468e-05,
+      "loss": 0.686,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990817355613323e-05,
+      "loss": 0.6044,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999077460255246e-05,
+      "loss": 0.5891,
+      "step": 1406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990731750242516e-05,
+      "loss": 0.6066,
+      "step": 1407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999068879868392e-05,
+      "loss": 0.626,
+      "step": 1408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990645747877093e-05,
+      "loss": 0.7298,
+      "step": 1409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990602597822468e-05,
+      "loss": 0.7793,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990559348520474e-05,
+      "loss": 0.7517,
+      "step": 1411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990515999971534e-05,
+      "loss": 0.6688,
+      "step": 1412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990472552176087e-05,
+      "loss": 0.6511,
+      "step": 1413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999042900513456e-05,
+      "loss": 0.6245,
+      "step": 1414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990385358847385e-05,
+      "loss": 0.6845,
+      "step": 1415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990341613314998e-05,
+      "loss": 0.6261,
+      "step": 1416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990297768537834e-05,
+      "loss": 0.7022,
+      "step": 1417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990253824516327e-05,
+      "loss": 0.6429,
+      "step": 1418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990209781250913e-05,
+      "loss": 0.5557,
+      "step": 1419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999016563874203e-05,
+      "loss": 0.6388,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990121396990117e-05,
+      "loss": 0.6416,
+      "step": 1421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990077055995614e-05,
+      "loss": 0.6795,
+      "step": 1422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990032615758957e-05,
+      "loss": 0.6535,
+      "step": 1423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989988076280596e-05,
+      "loss": 0.6821,
+      "step": 1424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998994343756097e-05,
+      "loss": 0.6588,
+      "step": 1425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989898699600518e-05,
+      "loss": 0.6909,
+      "step": 1426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989853862399685e-05,
+      "loss": 0.7831,
+      "step": 1427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989808925958923e-05,
+      "loss": 0.6339,
+      "step": 1428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989763890278674e-05,
+      "loss": 0.6876,
+      "step": 1429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989718755359383e-05,
+      "loss": 0.6593,
+      "step": 1430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989673521201503e-05,
+      "loss": 0.6442,
+      "step": 1431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989628187805482e-05,
+      "loss": 0.6467,
+      "step": 1432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989582755171767e-05,
+      "loss": 0.6908,
+      "step": 1433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989537223300816e-05,
+      "loss": 0.6806,
+      "step": 1434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989491592193074e-05,
+      "loss": 0.6325,
+      "step": 1435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989445861849e-05,
+      "loss": 0.6553,
+      "step": 1436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989400032269044e-05,
+      "loss": 0.6549,
+      "step": 1437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989354103453666e-05,
+      "loss": 0.6932,
+      "step": 1438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989308075403317e-05,
+      "loss": 0.6111,
+      "step": 1439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998926194811846e-05,
+      "loss": 0.6242,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989215721599547e-05,
+      "loss": 0.7608,
+      "step": 1441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989169395847045e-05,
+      "loss": 0.6593,
+      "step": 1442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989122970861407e-05,
+      "loss": 0.6376,
+      "step": 1443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998907644664309e-05,
+      "loss": 0.6602,
+      "step": 1444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989029823192575e-05,
+      "loss": 0.6555,
+      "step": 1445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988983100510303e-05,
+      "loss": 0.5724,
+      "step": 1446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988936278596752e-05,
+      "loss": 0.6691,
+      "step": 1447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988889357452384e-05,
+      "loss": 0.6528,
+      "step": 1448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988842337077665e-05,
+      "loss": 0.6909,
+      "step": 1449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998879521747306e-05,
+      "loss": 0.6576,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998874799863904e-05,
+      "loss": 0.7816,
+      "step": 1451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998870068057607e-05,
+      "loss": 0.6451,
+      "step": 1452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988653263284625e-05,
+      "loss": 0.6285,
+      "step": 1453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988605746765172e-05,
+      "loss": 0.6846,
+      "step": 1454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988558131018188e-05,
+      "loss": 0.652,
+      "step": 1455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998851041604414e-05,
+      "loss": 0.7029,
+      "step": 1456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988462601843505e-05,
+      "loss": 0.6565,
+      "step": 1457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998841468841676e-05,
+      "loss": 0.7572,
+      "step": 1458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988366675764376e-05,
+      "loss": 0.6256,
+      "step": 1459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988318563886834e-05,
+      "loss": 0.699,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988270352784612e-05,
+      "loss": 0.5625,
+      "step": 1461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988222042458187e-05,
+      "loss": 0.7685,
+      "step": 1462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998817363290804e-05,
+      "loss": 0.5807,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988125124134648e-05,
+      "loss": 0.6224,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99880765161385e-05,
+      "loss": 0.6641,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988027808920073e-05,
+      "loss": 0.6636,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987979002479857e-05,
+      "loss": 0.6343,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998793009681833e-05,
+      "loss": 0.6946,
+      "step": 1468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998788109193598e-05,
+      "loss": 0.6976,
+      "step": 1469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987831987833296e-05,
+      "loss": 0.7083,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998778278451076e-05,
+      "loss": 0.6157,
+      "step": 1471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987733481968874e-05,
+      "loss": 0.7158,
+      "step": 1472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987684080208113e-05,
+      "loss": 0.6276,
+      "step": 1473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987634579228978e-05,
+      "loss": 0.6658,
+      "step": 1474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987584979031954e-05,
+      "loss": 0.676,
+      "step": 1475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987535279617533e-05,
+      "loss": 0.6038,
+      "step": 1476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987485480986216e-05,
+      "loss": 0.6483,
+      "step": 1477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998743558313849e-05,
+      "loss": 0.6618,
+      "step": 1478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987385586074856e-05,
+      "loss": 0.7389,
+      "step": 1479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987335489795812e-05,
+      "loss": 0.5515,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998728529430185e-05,
+      "loss": 0.6221,
+      "step": 1481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998723499959347e-05,
+      "loss": 0.6427,
+      "step": 1482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987184605671177e-05,
+      "loss": 0.6294,
+      "step": 1483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987134112535466e-05,
+      "loss": 0.6488,
+      "step": 1484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998708352018684e-05,
+      "loss": 0.6126,
+      "step": 1485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99870328286258e-05,
+      "loss": 0.6602,
+      "step": 1486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998698203785285e-05,
+      "loss": 0.71,
+      "step": 1487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99869311478685e-05,
+      "loss": 0.7068,
+      "step": 1488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998688015867325e-05,
+      "loss": 0.5388,
+      "step": 1489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986829070267606e-05,
+      "loss": 0.6271,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986777882652082e-05,
+      "loss": 0.6061,
+      "step": 1491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986726595827177e-05,
+      "loss": 0.695,
+      "step": 1492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998667520979341e-05,
+      "loss": 0.6803,
+      "step": 1493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998662372455128e-05,
+      "loss": 0.6799,
+      "step": 1494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986572140101312e-05,
+      "loss": 0.6914,
+      "step": 1495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998652045644401e-05,
+      "loss": 0.7298,
+      "step": 1496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998646867357989e-05,
+      "loss": 0.7065,
+      "step": 1497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986416791509463e-05,
+      "loss": 0.6653,
+      "step": 1498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998636481023325e-05,
+      "loss": 0.5773,
+      "step": 1499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998631272975176e-05,
+      "loss": 0.6527,
+      "step": 1500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986260550065523e-05,
+      "loss": 0.6245,
+      "step": 1501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986208271175044e-05,
+      "loss": 0.6702,
+      "step": 1502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986155893080845e-05,
+      "loss": 0.5844,
+      "step": 1503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986103415783452e-05,
+      "loss": 0.6461,
+      "step": 1504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986050839283383e-05,
+      "loss": 0.6454,
+      "step": 1505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998599816358116e-05,
+      "loss": 0.6681,
+      "step": 1506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985945388677308e-05,
+      "loss": 0.6437,
+      "step": 1507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985892514572345e-05,
+      "loss": 0.6627,
+      "step": 1508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998583954126681e-05,
+      "loss": 0.728,
+      "step": 1509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985786468761216e-05,
+      "loss": 0.6745,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985733297056092e-05,
+      "loss": 0.6871,
+      "step": 1511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985680026151972e-05,
+      "loss": 0.6167,
+      "step": 1512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998562665604938e-05,
+      "loss": 0.7501,
+      "step": 1513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985573186748854e-05,
+      "loss": 0.6694,
+      "step": 1514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985519618250912e-05,
+      "loss": 0.7209,
+      "step": 1515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99854659505561e-05,
+      "loss": 0.6451,
+      "step": 1516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985412183664942e-05,
+      "loss": 0.6674,
+      "step": 1517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985358317577973e-05,
+      "loss": 0.6192,
+      "step": 1518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985304352295732e-05,
+      "loss": 0.6995,
+      "step": 1519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985250287818756e-05,
+      "loss": 0.6287,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985196124147575e-05,
+      "loss": 0.6517,
+      "step": 1521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998514186128273e-05,
+      "loss": 0.6689,
+      "step": 1522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985087499224764e-05,
+      "loss": 0.6896,
+      "step": 1523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985033037974214e-05,
+      "loss": 0.5857,
+      "step": 1524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984978477531625e-05,
+      "loss": 0.6036,
+      "step": 1525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998492381789753e-05,
+      "loss": 0.6659,
+      "step": 1526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984869059072482e-05,
+      "loss": 0.7029,
+      "step": 1527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984814201057016e-05,
+      "loss": 0.6183,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984759243851685e-05,
+      "loss": 0.5858,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998470418745703e-05,
+      "loss": 0.6281,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984649031873595e-05,
+      "loss": 0.6414,
+      "step": 1531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984593777101937e-05,
+      "loss": 0.6536,
+      "step": 1532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99845384231426e-05,
+      "loss": 0.7621,
+      "step": 1533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984482969996133e-05,
+      "loss": 0.6426,
+      "step": 1534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984427417663085e-05,
+      "loss": 0.6066,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984371766144017e-05,
+      "loss": 0.5734,
+      "step": 1536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984316015439467e-05,
+      "loss": 0.6555,
+      "step": 1537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984260165550006e-05,
+      "loss": 0.7877,
+      "step": 1538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998420421647617e-05,
+      "loss": 0.61,
+      "step": 1539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984148168218536e-05,
+      "loss": 0.6668,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984092020777646e-05,
+      "loss": 0.7148,
+      "step": 1541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998403577415406e-05,
+      "loss": 0.655,
+      "step": 1542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983979428348335e-05,
+      "loss": 0.7189,
+      "step": 1543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998392298336104e-05,
+      "loss": 0.6812,
+      "step": 1544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983866439192726e-05,
+      "loss": 0.7373,
+      "step": 1545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998380979584396e-05,
+      "loss": 0.7066,
+      "step": 1546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983753053315304e-05,
+      "loss": 0.6743,
+      "step": 1547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998369621160732e-05,
+      "loss": 0.6155,
+      "step": 1548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983639270720573e-05,
+      "loss": 0.5414,
+      "step": 1549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998358223065563e-05,
+      "loss": 0.7283,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983525091413057e-05,
+      "loss": 0.6948,
+      "step": 1551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983467852993422e-05,
+      "loss": 0.6497,
+      "step": 1552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998341051539729e-05,
+      "loss": 0.5762,
+      "step": 1553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983353078625238e-05,
+      "loss": 0.7309,
+      "step": 1554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998329554267783e-05,
+      "loss": 0.677,
+      "step": 1555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998323790755564e-05,
+      "loss": 0.6874,
+      "step": 1556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983180173259238e-05,
+      "loss": 0.6439,
+      "step": 1557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983122339789203e-05,
+      "loss": 0.6031,
+      "step": 1558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99830644071461e-05,
+      "loss": 0.542,
+      "step": 1559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998300637533052e-05,
+      "loss": 0.7326,
+      "step": 1560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982948244343024e-05,
+      "loss": 0.61,
+      "step": 1561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99828900141842e-05,
+      "loss": 0.5926,
+      "step": 1562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998283168485462e-05,
+      "loss": 0.642,
+      "step": 1563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982773256354864e-05,
+      "loss": 0.636,
+      "step": 1564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998271472868552e-05,
+      "loss": 0.6242,
+      "step": 1565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998265610184716e-05,
+      "loss": 0.7336,
+      "step": 1566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982597375840373e-05,
+      "loss": 0.6217,
+      "step": 1567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982538550665732e-05,
+      "loss": 0.6665,
+      "step": 1568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982479626323835e-05,
+      "loss": 0.6815,
+      "step": 1569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998242060281526e-05,
+      "loss": 0.6586,
+      "step": 1570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982361480140596e-05,
+      "loss": 0.6074,
+      "step": 1571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998230225830043e-05,
+      "loss": 0.7143,
+      "step": 1572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998224293729534e-05,
+      "loss": 0.6389,
+      "step": 1573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982183517125934e-05,
+      "loss": 0.6584,
+      "step": 1574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998212399779279e-05,
+      "loss": 0.6353,
+      "step": 1575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982064379296503e-05,
+      "loss": 0.6065,
+      "step": 1576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998200466163766e-05,
+      "loss": 0.6358,
+      "step": 1577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998194484481686e-05,
+      "loss": 0.6186,
+      "step": 1578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99818849288347e-05,
+      "loss": 0.6999,
+      "step": 1579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981824913691765e-05,
+      "loss": 0.6353,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981764799388663e-05,
+      "loss": 0.686,
+      "step": 1581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998170458592598e-05,
+      "loss": 0.6107,
+      "step": 1582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998164427330432e-05,
+      "loss": 0.5868,
+      "step": 1583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981583861524284e-05,
+      "loss": 0.6902,
+      "step": 1584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998152335058647e-05,
+      "loss": 0.6746,
+      "step": 1585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981462740491477e-05,
+      "loss": 0.6478,
+      "step": 1586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998140203123991e-05,
+      "loss": 0.6149,
+      "step": 1587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981341222832373e-05,
+      "loss": 0.5931,
+      "step": 1588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981280315269464e-05,
+      "loss": 0.6338,
+      "step": 1589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981219308551795e-05,
+      "loss": 0.667,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998115820267997e-05,
+      "loss": 0.6287,
+      "step": 1591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981096997654596e-05,
+      "loss": 0.5801,
+      "step": 1592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9981035693476275e-05,
+      "loss": 0.6899,
+      "step": 1593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980974290145626e-05,
+      "loss": 0.6419,
+      "step": 1594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980912787663253e-05,
+      "loss": 0.5709,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980851186029767e-05,
+      "loss": 0.6302,
+      "step": 1596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980789485245782e-05,
+      "loss": 0.7214,
+      "step": 1597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998072768531191e-05,
+      "loss": 0.5349,
+      "step": 1598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980665786228766e-05,
+      "loss": 0.7005,
+      "step": 1599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980603787996965e-05,
+      "loss": 0.544,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980541690617117e-05,
+      "loss": 0.9018,
+      "step": 1601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998047949408985e-05,
+      "loss": 0.6044,
+      "step": 1602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980417198415773e-05,
+      "loss": 0.6526,
+      "step": 1603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980354803595505e-05,
+      "loss": 0.6608,
+      "step": 1604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998029230962967e-05,
+      "loss": 0.6223,
+      "step": 1605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998022971651889e-05,
+      "loss": 0.6241,
+      "step": 1606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998016702426378e-05,
+      "loss": 0.6251,
+      "step": 1607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998010423286497e-05,
+      "loss": 0.6048,
+      "step": 1608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980041342323077e-05,
+      "loss": 0.6069,
+      "step": 1609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997997835263873e-05,
+      "loss": 0.6224,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979915263812554e-05,
+      "loss": 0.6807,
+      "step": 1611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997985207584518e-05,
+      "loss": 0.6138,
+      "step": 1612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979788788737228e-05,
+      "loss": 0.6407,
+      "step": 1613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997972540248933e-05,
+      "loss": 0.6434,
+      "step": 1614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979661917102116e-05,
+      "loss": 0.6966,
+      "step": 1615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979598332576217e-05,
+      "loss": 0.5972,
+      "step": 1616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979534648912263e-05,
+      "loss": 0.6013,
+      "step": 1617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997947086611089e-05,
+      "loss": 0.7624,
+      "step": 1618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979406984172728e-05,
+      "loss": 0.6435,
+      "step": 1619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997934300309841e-05,
+      "loss": 0.6456,
+      "step": 1620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997927892288858e-05,
+      "loss": 0.5571,
+      "step": 1621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979214743543863e-05,
+      "loss": 0.5905,
+      "step": 1622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997915046506491e-05,
+      "loss": 0.7226,
+      "step": 1623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979086087452345e-05,
+      "loss": 0.6848,
+      "step": 1624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979021610706822e-05,
+      "loss": 0.591,
+      "step": 1625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997895703482897e-05,
+      "loss": 0.6104,
+      "step": 1626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978892359819438e-05,
+      "loss": 0.6618,
+      "step": 1627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997882758567886e-05,
+      "loss": 0.7156,
+      "step": 1628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978762712407887e-05,
+      "loss": 0.6094,
+      "step": 1629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978697740007164e-05,
+      "loss": 0.5835,
+      "step": 1630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997863266847733e-05,
+      "loss": 0.6076,
+      "step": 1631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997856749781904e-05,
+      "loss": 0.5748,
+      "step": 1632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978502228032932e-05,
+      "loss": 0.5687,
+      "step": 1633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997843685911966e-05,
+      "loss": 0.6577,
+      "step": 1634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978371391079874e-05,
+      "loss": 0.6667,
+      "step": 1635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997830582391422e-05,
+      "loss": 0.7848,
+      "step": 1636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978240157623356e-05,
+      "loss": 0.5902,
+      "step": 1637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978174392207927e-05,
+      "loss": 0.6692,
+      "step": 1638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997810852766859e-05,
+      "loss": 0.6818,
+      "step": 1639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978042564006e-05,
+      "loss": 0.646,
+      "step": 1640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997797650122081e-05,
+      "loss": 0.6451,
+      "step": 1641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997791033931368e-05,
+      "loss": 0.597,
+      "step": 1642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997784407828526e-05,
+      "loss": 0.6235,
+      "step": 1643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977777718136218e-05,
+      "loss": 0.6776,
+      "step": 1644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977711258867204e-05,
+      "loss": 0.6161,
+      "step": 1645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977644700478884e-05,
+      "loss": 0.601,
+      "step": 1646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977578042971923e-05,
+      "loss": 0.6936,
+      "step": 1647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997751128634697e-05,
+      "loss": 0.6239,
+      "step": 1648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99774444306047e-05,
+      "loss": 0.6267,
+      "step": 1649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977377475745772e-05,
+      "loss": 0.6694,
+      "step": 1650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977310421770854e-05,
+      "loss": 0.6964,
+      "step": 1651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997724326868061e-05,
+      "loss": 0.5806,
+      "step": 1652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977176016475702e-05,
+      "loss": 0.6905,
+      "step": 1653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997710866515681e-05,
+      "loss": 0.5935,
+      "step": 1654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977041214724594e-05,
+      "loss": 0.6846,
+      "step": 1655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9976973665179732e-05,
+      "loss": 0.6563,
+      "step": 1656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997690601652288e-05,
+      "loss": 0.6188,
+      "step": 1657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976838268754727e-05,
+      "loss": 0.6377,
+      "step": 1658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997677042187594e-05,
+      "loss": 0.6491,
+      "step": 1659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997670247588719e-05,
+      "loss": 0.6446,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997663443078915e-05,
+      "loss": 0.6228,
+      "step": 1661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976566286582503e-05,
+      "loss": 0.5528,
+      "step": 1662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997649804326792e-05,
+      "loss": 0.5331,
+      "step": 1663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976429700846086e-05,
+      "loss": 0.6868,
+      "step": 1664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976361259317674e-05,
+      "loss": 0.6317,
+      "step": 1665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976292718683365e-05,
+      "loss": 0.7327,
+      "step": 1666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997622407894384e-05,
+      "loss": 0.5052,
+      "step": 1667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997615534009978e-05,
+      "loss": 0.6643,
+      "step": 1668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997608650215187e-05,
+      "loss": 0.6216,
+      "step": 1669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976017565100795e-05,
+      "loss": 0.6402,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975948528947235e-05,
+      "loss": 0.5763,
+      "step": 1671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997587939369188e-05,
+      "loss": 0.6352,
+      "step": 1672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975810159335415e-05,
+      "loss": 0.7417,
+      "step": 1673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997574082587853e-05,
+      "loss": 0.6469,
+      "step": 1674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997567139332191e-05,
+      "loss": 0.5799,
+      "step": 1675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975601861666246e-05,
+      "loss": 0.6638,
+      "step": 1676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997553223091223e-05,
+      "loss": 0.6389,
+      "step": 1677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975462501060553e-05,
+      "loss": 0.6379,
+      "step": 1678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997539267211191e-05,
+      "loss": 0.6482,
+      "step": 1679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975322744066987e-05,
+      "loss": 0.6507,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975252716926488e-05,
+      "loss": 0.6748,
+      "step": 1681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975182590691103e-05,
+      "loss": 0.5922,
+      "step": 1682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975112365361533e-05,
+      "loss": 0.6141,
+      "step": 1683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975042040938474e-05,
+      "loss": 0.6083,
+      "step": 1684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974971617422622e-05,
+      "loss": 0.6587,
+      "step": 1685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974901094814676e-05,
+      "loss": 0.6412,
+      "step": 1686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997483047311534e-05,
+      "loss": 0.5721,
+      "step": 1687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974759752325314e-05,
+      "loss": 0.6888,
+      "step": 1688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974688932445304e-05,
+      "loss": 0.6451,
+      "step": 1689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997461801347601e-05,
+      "loss": 0.6199,
+      "step": 1690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974546995418133e-05,
+      "loss": 0.6367,
+      "step": 1691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974475878272385e-05,
+      "loss": 0.5398,
+      "step": 1692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997440466203947e-05,
+      "loss": 0.6647,
+      "step": 1693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974333346720096e-05,
+      "loss": 0.6941,
+      "step": 1694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997426193231497e-05,
+      "loss": 0.6343,
+      "step": 1695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99741904188248e-05,
+      "loss": 0.5711,
+      "step": 1696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974118806250305e-05,
+      "loss": 0.6769,
+      "step": 1697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974047094592185e-05,
+      "loss": 0.5803,
+      "step": 1698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973975283851163e-05,
+      "loss": 0.6203,
+      "step": 1699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997390337402794e-05,
+      "loss": 0.6212,
+      "step": 1700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997383136512324e-05,
+      "loss": 0.7591,
+      "step": 1701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973759257137776e-05,
+      "loss": 0.6824,
+      "step": 1702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973687050072264e-05,
+      "loss": 0.6544,
+      "step": 1703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973614743927424e-05,
+      "loss": 0.6776,
+      "step": 1704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973542338703965e-05,
+      "loss": 0.6408,
+      "step": 1705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997346983440262e-05,
+      "loss": 0.6508,
+      "step": 1706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973397231024095e-05,
+      "loss": 0.681,
+      "step": 1707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997332452856912e-05,
+      "loss": 0.6035,
+      "step": 1708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997325172703842e-05,
+      "loss": 0.6371,
+      "step": 1709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973178826432713e-05,
+      "loss": 0.6547,
+      "step": 1710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973105826752723e-05,
+      "loss": 0.6721,
+      "step": 1711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973032727999174e-05,
+      "loss": 0.5948,
+      "step": 1712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972959530172797e-05,
+      "loss": 0.6874,
+      "step": 1713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972886233274312e-05,
+      "loss": 0.6327,
+      "step": 1714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972812837304454e-05,
+      "loss": 0.7362,
+      "step": 1715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972739342263953e-05,
+      "loss": 0.6594,
+      "step": 1716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972665748153533e-05,
+      "loss": 0.7403,
+      "step": 1717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972592054973926e-05,
+      "loss": 0.6598,
+      "step": 1718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997251826272587e-05,
+      "loss": 0.5585,
+      "step": 1719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972444371410092e-05,
+      "loss": 0.6147,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997237038102733e-05,
+      "loss": 0.5869,
+      "step": 1721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972296291578316e-05,
+      "loss": 0.6406,
+      "step": 1722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972222103063786e-05,
+      "loss": 0.5928,
+      "step": 1723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997214781548448e-05,
+      "loss": 0.6248,
+      "step": 1724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972073428841134e-05,
+      "loss": 0.617,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971998943134486e-05,
+      "loss": 0.6186,
+      "step": 1726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971924358365274e-05,
+      "loss": 0.6004,
+      "step": 1727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997184967453425e-05,
+      "loss": 0.5627,
+      "step": 1728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997177489164214e-05,
+      "loss": 0.6706,
+      "step": 1729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99717000096897e-05,
+      "loss": 0.6466,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971625028677665e-05,
+      "loss": 0.6091,
+      "step": 1731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971549948606784e-05,
+      "loss": 0.6783,
+      "step": 1732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971474769477807e-05,
+      "loss": 0.5123,
+      "step": 1733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997139949129147e-05,
+      "loss": 0.5636,
+      "step": 1734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971324114048533e-05,
+      "loss": 0.6998,
+      "step": 1735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971248637749735e-05,
+      "loss": 0.6359,
+      "step": 1736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971173062395827e-05,
+      "loss": 0.6823,
+      "step": 1737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971097387987566e-05,
+      "loss": 0.6544,
+      "step": 1738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99710216145257e-05,
+      "loss": 0.5951,
+      "step": 1739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997094574201098e-05,
+      "loss": 0.6539,
+      "step": 1740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970869770444165e-05,
+      "loss": 0.6759,
+      "step": 1741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970793699826003e-05,
+      "loss": 0.6054,
+      "step": 1742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970717530157256e-05,
+      "loss": 0.6263,
+      "step": 1743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970641261438674e-05,
+      "loss": 0.6476,
+      "step": 1744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997056489367102e-05,
+      "loss": 0.6987,
+      "step": 1745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970488426855052e-05,
+      "loss": 0.5982,
+      "step": 1746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997041186099153e-05,
+      "loss": 0.601,
+      "step": 1747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970335196081212e-05,
+      "loss": 0.5578,
+      "step": 1748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997025843212486e-05,
+      "loss": 0.5795,
+      "step": 1749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970181569123235e-05,
+      "loss": 0.6455,
+      "step": 1750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970104607077107e-05,
+      "loss": 0.5437,
+      "step": 1751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970027545987234e-05,
+      "loss": 0.5539,
+      "step": 1752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969950385854387e-05,
+      "loss": 0.6347,
+      "step": 1753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969873126679323e-05,
+      "loss": 0.7185,
+      "step": 1754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969795768462824e-05,
+      "loss": 0.5602,
+      "step": 1755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969718311205647e-05,
+      "loss": 0.6687,
+      "step": 1756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969640754908565e-05,
+      "loss": 0.5932,
+      "step": 1757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969563099572346e-05,
+      "loss": 0.5957,
+      "step": 1758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996948534519777e-05,
+      "loss": 0.6344,
+      "step": 1759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969407491785596e-05,
+      "loss": 0.6765,
+      "step": 1760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996932953933661e-05,
+      "loss": 0.667,
+      "step": 1761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996925148785158e-05,
+      "loss": 0.5717,
+      "step": 1762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969173337331283e-05,
+      "loss": 0.6333,
+      "step": 1763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996909508777649e-05,
+      "loss": 0.5697,
+      "step": 1764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996901673918799e-05,
+      "loss": 0.6026,
+      "step": 1765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996893829156655e-05,
+      "loss": 0.6671,
+      "step": 1766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968859744912957e-05,
+      "loss": 0.5982,
+      "step": 1767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968781099227984e-05,
+      "loss": 0.7275,
+      "step": 1768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996870235451242e-05,
+      "loss": 0.6317,
+      "step": 1769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996862351076704e-05,
+      "loss": 0.6813,
+      "step": 1770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968544567992634e-05,
+      "loss": 0.5886,
+      "step": 1771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996846552618998e-05,
+      "loss": 0.6076,
+      "step": 1772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968386385359867e-05,
+      "loss": 0.6387,
+      "step": 1773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968307145503085e-05,
+      "loss": 0.8268,
+      "step": 1774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968227806620413e-05,
+      "loss": 0.7063,
+      "step": 1775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996814836871264e-05,
+      "loss": 0.5804,
+      "step": 1776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968068831780562e-05,
+      "loss": 0.6299,
+      "step": 1777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967989195824967e-05,
+      "loss": 0.5778,
+      "step": 1778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996790946084664e-05,
+      "loss": 0.5525,
+      "step": 1779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996782962684638e-05,
+      "loss": 0.6094,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967749693824975e-05,
+      "loss": 0.6391,
+      "step": 1781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996766966178322e-05,
+      "loss": 0.6016,
+      "step": 1782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967589530721915e-05,
+      "loss": 0.5335,
+      "step": 1783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996750930064185e-05,
+      "loss": 0.6443,
+      "step": 1784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967428971543824e-05,
+      "loss": 0.5469,
+      "step": 1785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967348543428637e-05,
+      "loss": 0.6213,
+      "step": 1786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967268016297087e-05,
+      "loss": 0.6006,
+      "step": 1787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967187390149974e-05,
+      "loss": 0.5989,
+      "step": 1788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967106664988095e-05,
+      "loss": 0.6989,
+      "step": 1789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967025840812258e-05,
+      "loss": 0.5834,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966944917623264e-05,
+      "loss": 0.6443,
+      "step": 1791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966863895421912e-05,
+      "loss": 0.6886,
+      "step": 1792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966782774209013e-05,
+      "loss": 0.6209,
+      "step": 1793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996670155398537e-05,
+      "loss": 0.5047,
+      "step": 1794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966620234751794e-05,
+      "loss": 0.6099,
+      "step": 1795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996653881650909e-05,
+      "loss": 0.6477,
+      "step": 1796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966457299258063e-05,
+      "loss": 0.5727,
+      "step": 1797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966375682999522e-05,
+      "loss": 0.5852,
+      "step": 1798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966293967734288e-05,
+      "loss": 0.6566,
+      "step": 1799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966212153463168e-05,
+      "loss": 0.6815,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996613024018697e-05,
+      "loss": 0.6849,
+      "step": 1801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966048227906513e-05,
+      "loss": 0.6979,
+      "step": 1802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996596611662261e-05,
+      "loss": 0.5735,
+      "step": 1803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965883906336073e-05,
+      "loss": 0.731,
+      "step": 1804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965801597047724e-05,
+      "loss": 0.6973,
+      "step": 1805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996571918875838e-05,
+      "loss": 0.6507,
+      "step": 1806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996563668146886e-05,
+      "loss": 0.5604,
+      "step": 1807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996555407517998e-05,
+      "loss": 0.6481,
+      "step": 1808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965471369892564e-05,
+      "loss": 0.5531,
+      "step": 1809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996538856560743e-05,
+      "loss": 0.614,
+      "step": 1810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965305662325406e-05,
+      "loss": 0.5832,
+      "step": 1811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996522266004731e-05,
+      "loss": 0.6195,
+      "step": 1812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996513955877397e-05,
+      "loss": 0.56,
+      "step": 1813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965056358506213e-05,
+      "loss": 0.712,
+      "step": 1814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996497305924486e-05,
+      "loss": 0.6969,
+      "step": 1815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964889660990745e-05,
+      "loss": 0.5684,
+      "step": 1816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996480616374469e-05,
+      "loss": 0.6487,
+      "step": 1817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996472256750753e-05,
+      "loss": 0.6405,
+      "step": 1818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964638872280088e-05,
+      "loss": 0.6647,
+      "step": 1819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964555078063206e-05,
+      "loss": 0.5785,
+      "step": 1820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996447118485771e-05,
+      "loss": 0.6296,
+      "step": 1821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964387192664433e-05,
+      "loss": 0.6671,
+      "step": 1822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996430310148421e-05,
+      "loss": 0.6258,
+      "step": 1823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996421891131788e-05,
+      "loss": 0.5821,
+      "step": 1824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964134622166273e-05,
+      "loss": 0.6614,
+      "step": 1825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964050234030234e-05,
+      "loss": 0.6286,
+      "step": 1826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996396574691059e-05,
+      "loss": 0.5625,
+      "step": 1827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963881160808194e-05,
+      "loss": 0.618,
+      "step": 1828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963796475723877e-05,
+      "loss": 0.5658,
+      "step": 1829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963711691658485e-05,
+      "loss": 0.6737,
+      "step": 1830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996362680861286e-05,
+      "loss": 0.6624,
+      "step": 1831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963541826587837e-05,
+      "loss": 0.5576,
+      "step": 1832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963456745584268e-05,
+      "loss": 0.5842,
+      "step": 1833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963371565603e-05,
+      "loss": 0.6841,
+      "step": 1834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963286286644875e-05,
+      "loss": 0.6041,
+      "step": 1835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963200908710738e-05,
+      "loss": 0.7903,
+      "step": 1836
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963115431801445e-05,
+      "loss": 0.6669,
+      "step": 1837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996302985591784e-05,
+      "loss": 0.6163,
+      "step": 1838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962944181060772e-05,
+      "loss": 0.6201,
+      "step": 1839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962858407231093e-05,
+      "loss": 0.6262,
+      "step": 1840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962772534429656e-05,
+      "loss": 0.6061,
+      "step": 1841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962686562657315e-05,
+      "loss": 0.706,
+      "step": 1842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962600491914924e-05,
+      "loss": 0.6989,
+      "step": 1843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962514322203336e-05,
+      "loss": 0.801,
+      "step": 1844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962428053523406e-05,
+      "loss": 0.6354,
+      "step": 1845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962341685875994e-05,
+      "loss": 0.6539,
+      "step": 1846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962255219261957e-05,
+      "loss": 0.6507,
+      "step": 1847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962168653682155e-05,
+      "loss": 0.6091,
+      "step": 1848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962081989137443e-05,
+      "loss": 0.5679,
+      "step": 1849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996199522562869e-05,
+      "loss": 0.6157,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996190836315675e-05,
+      "loss": 0.6492,
+      "step": 1851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9961821401722492e-05,
+      "loss": 0.5953,
+      "step": 1852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961734341326778e-05,
+      "loss": 0.7207,
+      "step": 1853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996164718197047e-05,
+      "loss": 0.6894,
+      "step": 1854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996155992365444e-05,
+      "loss": 0.6653,
+      "step": 1855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996147256637955e-05,
+      "loss": 0.695,
+      "step": 1856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961385110146668e-05,
+      "loss": 0.6886,
+      "step": 1857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996129755495666e-05,
+      "loss": 0.589,
+      "step": 1858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961209900810404e-05,
+      "loss": 0.6233,
+      "step": 1859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961122147708764e-05,
+      "loss": 0.6067,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961034295652616e-05,
+      "loss": 0.6603,
+      "step": 1861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996094634464283e-05,
+      "loss": 0.6449,
+      "step": 1862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996085829468028e-05,
+      "loss": 0.5989,
+      "step": 1863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996077014576584e-05,
+      "loss": 0.6048,
+      "step": 1864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960681897900388e-05,
+      "loss": 0.5519,
+      "step": 1865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99605935510848e-05,
+      "loss": 0.6196,
+      "step": 1866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960505105319948e-05,
+      "loss": 0.6653,
+      "step": 1867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996041656060672e-05,
+      "loss": 0.6505,
+      "step": 1868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960327916945993e-05,
+      "loss": 0.6619,
+      "step": 1869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996023917433864e-05,
+      "loss": 0.7462,
+      "step": 1870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960150332785556e-05,
+      "loss": 0.6143,
+      "step": 1871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996006139228761e-05,
+      "loss": 0.5926,
+      "step": 1872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959972352845695e-05,
+      "loss": 0.632,
+      "step": 1873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995988321446069e-05,
+      "loss": 0.7471,
+      "step": 1874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959793977133483e-05,
+      "loss": 0.6341,
+      "step": 1875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959704640864964e-05,
+      "loss": 0.7206,
+      "step": 1876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995961520565601e-05,
+      "loss": 0.7264,
+      "step": 1877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959525671507516e-05,
+      "loss": 0.6065,
+      "step": 1878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959436038420377e-05,
+      "loss": 0.5386,
+      "step": 1879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959346306395476e-05,
+      "loss": 0.7042,
+      "step": 1880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959256475433705e-05,
+      "loss": 0.6534,
+      "step": 1881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959166545535958e-05,
+      "loss": 0.7242,
+      "step": 1882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995907651670313e-05,
+      "loss": 0.5806,
+      "step": 1883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958986388936108e-05,
+      "loss": 0.8286,
+      "step": 1884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958896162235794e-05,
+      "loss": 0.6561,
+      "step": 1885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958805836603088e-05,
+      "loss": 0.5583,
+      "step": 1886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995871541203888e-05,
+      "loss": 0.6414,
+      "step": 1887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958624888544068e-05,
+      "loss": 0.6954,
+      "step": 1888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958534266119556e-05,
+      "loss": 0.6034,
+      "step": 1889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995844354476624e-05,
+      "loss": 0.66,
+      "step": 1890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958352724485027e-05,
+      "loss": 0.6548,
+      "step": 1891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958261805276813e-05,
+      "loss": 0.6404,
+      "step": 1892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958170787142506e-05,
+      "loss": 0.6047,
+      "step": 1893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958079670083005e-05,
+      "loss": 0.7198,
+      "step": 1894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995798845409922e-05,
+      "loss": 0.6499,
+      "step": 1895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957897139192056e-05,
+      "loss": 0.5431,
+      "step": 1896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957805725362415e-05,
+      "loss": 0.5997,
+      "step": 1897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995771421261121e-05,
+      "loss": 0.5698,
+      "step": 1898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957622600939356e-05,
+      "loss": 0.6234,
+      "step": 1899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957530890347754e-05,
+      "loss": 0.6577,
+      "step": 1900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957439080837314e-05,
+      "loss": 0.6333,
+      "step": 1901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957347172408953e-05,
+      "loss": 0.628,
+      "step": 1902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957255165063584e-05,
+      "loss": 0.6833,
+      "step": 1903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957163058802123e-05,
+      "loss": 0.6863,
+      "step": 1904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957070853625476e-05,
+      "loss": 0.6702,
+      "step": 1905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956978549534568e-05,
+      "loss": 0.656,
+      "step": 1906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956886146530314e-05,
+      "loss": 0.5415,
+      "step": 1907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956793644613627e-05,
+      "loss": 0.6644,
+      "step": 1908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995670104378543e-05,
+      "loss": 0.6641,
+      "step": 1909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956608344046646e-05,
+      "loss": 0.7116,
+      "step": 1910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995651554539819e-05,
+      "loss": 0.625,
+      "step": 1911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956422647840984e-05,
+      "loss": 0.6032,
+      "step": 1912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995632965137596e-05,
+      "loss": 0.64,
+      "step": 1913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956236556004026e-05,
+      "loss": 0.5812,
+      "step": 1914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956143361726123e-05,
+      "loss": 0.5757,
+      "step": 1915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956050068543167e-05,
+      "loss": 0.6327,
+      "step": 1916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9955956676456087e-05,
+      "loss": 0.5825,
+      "step": 1917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955863185465808e-05,
+      "loss": 0.5914,
+      "step": 1918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955769595573265e-05,
+      "loss": 0.643,
+      "step": 1919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955675906779386e-05,
+      "loss": 0.6169,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955582119085097e-05,
+      "loss": 0.5613,
+      "step": 1921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955488232491333e-05,
+      "loss": 0.6678,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955394246999028e-05,
+      "loss": 0.6028,
+      "step": 1923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955300162609118e-05,
+      "loss": 0.6215,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955205979322528e-05,
+      "loss": 0.6977,
+      "step": 1925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955111697140203e-05,
+      "loss": 0.5875,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955017316063074e-05,
+      "loss": 0.6107,
+      "step": 1927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954922836092085e-05,
+      "loss": 0.6303,
+      "step": 1928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954828257228166e-05,
+      "loss": 0.5478,
+      "step": 1929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954733579472262e-05,
+      "loss": 0.6263,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995463880282531e-05,
+      "loss": 0.7451,
+      "step": 1931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954543927288256e-05,
+      "loss": 0.6173,
+      "step": 1932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995444895286204e-05,
+      "loss": 0.6724,
+      "step": 1933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995435387954761e-05,
+      "loss": 0.6254,
+      "step": 1934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954258707345903e-05,
+      "loss": 0.6187,
+      "step": 1935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954163436257867e-05,
+      "loss": 0.6188,
+      "step": 1936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995406806628445e-05,
+      "loss": 0.6007,
+      "step": 1937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953972597426597e-05,
+      "loss": 0.6903,
+      "step": 1938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995387702968526e-05,
+      "loss": 0.6936,
+      "step": 1939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953781363061388e-05,
+      "loss": 0.6213,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953685597555927e-05,
+      "loss": 0.5943,
+      "step": 1941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995358973316983e-05,
+      "loss": 0.6332,
+      "step": 1942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953493769904054e-05,
+      "loss": 0.6403,
+      "step": 1943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953397707759548e-05,
+      "loss": 0.7535,
+      "step": 1944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953301546737267e-05,
+      "loss": 0.7073,
+      "step": 1945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953205286838164e-05,
+      "loss": 0.7206,
+      "step": 1946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99531089280632e-05,
+      "loss": 0.668,
+      "step": 1947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995301247041333e-05,
+      "loss": 0.6401,
+      "step": 1948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952915913889513e-05,
+      "loss": 0.5528,
+      "step": 1949
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952819258492705e-05,
+      "loss": 0.705,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952722504223866e-05,
+      "loss": 0.6463,
+      "step": 1951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952625651083965e-05,
+      "loss": 0.6795,
+      "step": 1952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952528699073953e-05,
+      "loss": 0.6037,
+      "step": 1953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952431648194803e-05,
+      "loss": 0.6616,
+      "step": 1954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952334498447473e-05,
+      "loss": 0.6607,
+      "step": 1955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952237249832933e-05,
+      "loss": 0.674,
+      "step": 1956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952139902352142e-05,
+      "loss": 0.541,
+      "step": 1957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952042456006072e-05,
+      "loss": 0.5984,
+      "step": 1958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951944910795692e-05,
+      "loss": 0.6748,
+      "step": 1959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 0.6351,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951749523785874e-05,
+      "loss": 0.5817,
+      "step": 1961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951651681988378e-05,
+      "loss": 0.6138,
+      "step": 1962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995155374133045e-05,
+      "loss": 0.6389,
+      "step": 1963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951455701813062e-05,
+      "loss": 0.5542,
+      "step": 1964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951357563437198e-05,
+      "loss": 0.6854,
+      "step": 1965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951259326203824e-05,
+      "loss": 0.582,
+      "step": 1966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951160990113912e-05,
+      "loss": 0.7187,
+      "step": 1967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995106255516845e-05,
+      "loss": 0.6287,
+      "step": 1968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995096402136841e-05,
+      "loss": 0.5848,
+      "step": 1969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995086538871477e-05,
+      "loss": 0.6295,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950766657208516e-05,
+      "loss": 0.6361,
+      "step": 1971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950667826850617e-05,
+      "loss": 0.5884,
+      "step": 1972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995056889764207e-05,
+      "loss": 0.616,
+      "step": 1973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950469869583844e-05,
+      "loss": 0.561,
+      "step": 1974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995037074267693e-05,
+      "loss": 0.6132,
+      "step": 1975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950271516922312e-05,
+      "loss": 0.6186,
+      "step": 1976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950172192320973e-05,
+      "loss": 0.7015,
+      "step": 1977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99500727688739e-05,
+      "loss": 0.518,
+      "step": 1978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9949973246582085e-05,
+      "loss": 0.6006,
+      "step": 1979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9949873625446516e-05,
+      "loss": 0.5734,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9949773905468174e-05,
+      "loss": 0.7251,
+      "step": 1981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.994967408664806e-05,
+      "loss": 0.5456,
+      "step": 1982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949574168987164e-05,
+      "loss": 0.714,
+      "step": 1983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949474152486474e-05,
+      "loss": 0.5945,
+      "step": 1984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949374037146984e-05,
+      "loss": 0.6435,
+      "step": 1985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949273822969695e-05,
+      "loss": 0.6724,
+      "step": 1986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949173509955594e-05,
+      "loss": 0.6218,
+      "step": 1987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949073098105685e-05,
+      "loss": 0.5173,
+      "step": 1988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994897258742096e-05,
+      "loss": 0.6204,
+      "step": 1989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994887197790242e-05,
+      "loss": 0.6191,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948771269551063e-05,
+      "loss": 0.7234,
+      "step": 1991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994867046236789e-05,
+      "loss": 0.5733,
+      "step": 1992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948569556353906e-05,
+      "loss": 0.7069,
+      "step": 1993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948468551510107e-05,
+      "loss": 0.7628,
+      "step": 1994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948367447837503e-05,
+      "loss": 0.7157,
+      "step": 1995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994826624533709e-05,
+      "loss": 0.6165,
+      "step": 1996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994816494400988e-05,
+      "loss": 0.5832,
+      "step": 1997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948063543856882e-05,
+      "loss": 0.674,
+      "step": 1998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947962044879094e-05,
+      "loss": 0.616,
+      "step": 1999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947860447077532e-05,
+      "loss": 0.6789,
+      "step": 2000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.99477587504532e-05,
+      "loss": 0.7053,
+      "step": 2001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947656955007115e-05,
+      "loss": 0.5763,
+      "step": 2002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994755506074028e-05,
+      "loss": 0.6072,
+      "step": 2003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947453067653714e-05,
+      "loss": 0.5802,
+      "step": 2004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947350975748428e-05,
+      "loss": 0.6463,
+      "step": 2005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947248785025435e-05,
+      "loss": 0.6576,
+      "step": 2006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994714649548575e-05,
+      "loss": 0.6095,
+      "step": 2007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994704410713039e-05,
+      "loss": 0.6203,
+      "step": 2008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994694161996037e-05,
+      "loss": 0.5822,
+      "step": 2009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946839033976714e-05,
+      "loss": 0.5846,
+      "step": 2010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946736349180437e-05,
+      "loss": 0.5456,
+      "step": 2011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994663356557256e-05,
+      "loss": 0.5964,
+      "step": 2012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946530683154103e-05,
+      "loss": 0.6196,
+      "step": 2013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946427701926086e-05,
+      "loss": 0.623,
+      "step": 2014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946324621889538e-05,
+      "loss": 0.6317,
+      "step": 2015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994622144304548e-05,
+      "loss": 0.6373,
+      "step": 2016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994611816539493e-05,
+      "loss": 0.7316,
+      "step": 2017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946014788938925e-05,
+      "loss": 0.5824,
+      "step": 2018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994591131367849e-05,
+      "loss": 0.5626,
+      "step": 2019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994580773961465e-05,
+      "loss": 0.594,
+      "step": 2020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945704066748428e-05,
+      "loss": 0.5625,
+      "step": 2021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945600295080865e-05,
+      "loss": 0.6254,
+      "step": 2022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945496424612988e-05,
+      "loss": 0.5773,
+      "step": 2023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945392455345826e-05,
+      "loss": 0.6148,
+      "step": 2024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994528838728042e-05,
+      "loss": 0.5238,
+      "step": 2025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994518422041779e-05,
+      "loss": 0.6442,
+      "step": 2026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994507995475898e-05,
+      "loss": 0.622,
+      "step": 2027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944975590305024e-05,
+      "loss": 0.6981,
+      "step": 2028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944871127056958e-05,
+      "loss": 0.5299,
+      "step": 2029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944766565015824e-05,
+      "loss": 0.6306,
+      "step": 2030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944661904182656e-05,
+      "loss": 0.6462,
+      "step": 2031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944557144558496e-05,
+      "loss": 0.6672,
+      "step": 2032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944452286144383e-05,
+      "loss": 0.6613,
+      "step": 2033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994434732894136e-05,
+      "loss": 0.6443,
+      "step": 2034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944242272950468e-05,
+      "loss": 0.631,
+      "step": 2035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944137118172754e-05,
+      "loss": 0.6361,
+      "step": 2036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944031864609256e-05,
+      "loss": 0.5736,
+      "step": 2037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994392651226103e-05,
+      "loss": 0.6272,
+      "step": 2038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994382106112911e-05,
+      "loss": 0.4984,
+      "step": 2039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943715511214553e-05,
+      "loss": 0.6517,
+      "step": 2040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943609862518408e-05,
+      "loss": 0.5634,
+      "step": 2041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943504115041715e-05,
+      "loss": 0.5474,
+      "step": 2042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943398268785537e-05,
+      "loss": 0.4952,
+      "step": 2043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943292323750918e-05,
+      "loss": 0.6564,
+      "step": 2044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943186279938908e-05,
+      "loss": 0.5874,
+      "step": 2045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943080137350564e-05,
+      "loss": 0.6833,
+      "step": 2046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9942973895986947e-05,
+      "loss": 0.5653,
+      "step": 2047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99428675558491e-05,
+      "loss": 0.6377,
+      "step": 2048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942761116938087e-05,
+      "loss": 0.5743,
+      "step": 2049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942654579254962e-05,
+      "loss": 0.5955,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994254794280079e-05,
+      "loss": 0.6353,
+      "step": 2051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994244120757662e-05,
+      "loss": 0.6485,
+      "step": 2052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942334373583522e-05,
+      "loss": 0.5758,
+      "step": 2053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942227440822554e-05,
+      "loss": 0.6483,
+      "step": 2054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942120409294768e-05,
+      "loss": 0.702,
+      "step": 2055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942013279001244e-05,
+      "loss": 0.5119,
+      "step": 2056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941906049943037e-05,
+      "loss": 0.5588,
+      "step": 2057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941798722121216e-05,
+      "loss": 0.6321,
+      "step": 2058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941691295536847e-05,
+      "loss": 0.6167,
+      "step": 2059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994158377019099e-05,
+      "loss": 0.6297,
+      "step": 2060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941476146084723e-05,
+      "loss": 0.5627,
+      "step": 2061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994136842321911e-05,
+      "loss": 0.6556,
+      "step": 2062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994126060159522e-05,
+      "loss": 0.6363,
+      "step": 2063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941152681214125e-05,
+      "loss": 0.6921,
+      "step": 2064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941044662076898e-05,
+      "loss": 0.6815,
+      "step": 2065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940936544184614e-05,
+      "loss": 0.5396,
+      "step": 2066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940828327538344e-05,
+      "loss": 0.5905,
+      "step": 2067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940720012139166e-05,
+      "loss": 0.6208,
+      "step": 2068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994061159798815e-05,
+      "loss": 0.6165,
+      "step": 2069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940503085086382e-05,
+      "loss": 0.6349,
+      "step": 2070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940394473434934e-05,
+      "loss": 0.6024,
+      "step": 2071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940285763034887e-05,
+      "loss": 0.6239,
+      "step": 2072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940176953887316e-05,
+      "loss": 0.6233,
+      "step": 2073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994006804599331e-05,
+      "loss": 0.6958,
+      "step": 2074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939959039353947e-05,
+      "loss": 0.5712,
+      "step": 2075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939849933970307e-05,
+      "loss": 0.7164,
+      "step": 2076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939740729843477e-05,
+      "loss": 0.6435,
+      "step": 2077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939631426974543e-05,
+      "loss": 0.625,
+      "step": 2078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993952202536459e-05,
+      "loss": 0.6305,
+      "step": 2079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99394125250147e-05,
+      "loss": 0.6354,
+      "step": 2080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939302925925968e-05,
+      "loss": 0.5955,
+      "step": 2081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993919322809948e-05,
+      "loss": 0.5747,
+      "step": 2082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939083431536324e-05,
+      "loss": 0.6523,
+      "step": 2083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938973536237595e-05,
+      "loss": 0.7878,
+      "step": 2084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993886354220438e-05,
+      "loss": 0.6115,
+      "step": 2085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993875344943777e-05,
+      "loss": 0.6082,
+      "step": 2086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938643257938866e-05,
+      "loss": 0.6556,
+      "step": 2087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938532967708755e-05,
+      "loss": 0.6546,
+      "step": 2088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993842257874854e-05,
+      "loss": 0.6878,
+      "step": 2089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938312091059316e-05,
+      "loss": 0.583,
+      "step": 2090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938201504642177e-05,
+      "loss": 0.6796,
+      "step": 2091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938090819498218e-05,
+      "loss": 0.6354,
+      "step": 2092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993798003562855e-05,
+      "loss": 0.5997,
+      "step": 2093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937869153034265e-05,
+      "loss": 0.6326,
+      "step": 2094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937758171716468e-05,
+      "loss": 0.5633,
+      "step": 2095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937647091676262e-05,
+      "loss": 0.6809,
+      "step": 2096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937535912914746e-05,
+      "loss": 0.5962,
+      "step": 2097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937424635433027e-05,
+      "loss": 0.5647,
+      "step": 2098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937313259232213e-05,
+      "loss": 0.6075,
+      "step": 2099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937201784313406e-05,
+      "loss": 0.5671,
+      "step": 2100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937090210677716e-05,
+      "loss": 0.6229,
+      "step": 2101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993697853832625e-05,
+      "loss": 0.6377,
+      "step": 2102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993686676726012e-05,
+      "loss": 0.6357,
+      "step": 2103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936754897480433e-05,
+      "loss": 0.6268,
+      "step": 2104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936642928988304e-05,
+      "loss": 0.5871,
+      "step": 2105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936530861784843e-05,
+      "loss": 0.5874,
+      "step": 2106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993641869587116e-05,
+      "loss": 0.6474,
+      "step": 2107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936306431248374e-05,
+      "loss": 0.6884,
+      "step": 2108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936194067917603e-05,
+      "loss": 0.6866,
+      "step": 2109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936081605879955e-05,
+      "loss": 0.6742,
+      "step": 2110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9935969045136555e-05,
+      "loss": 0.6362,
+      "step": 2111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9935856385688517e-05,
+      "loss": 0.6731,
+      "step": 2112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993574362753696e-05,
+      "loss": 0.5579,
+      "step": 2113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935630770683005e-05,
+      "loss": 0.6747,
+      "step": 2114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993551781512777e-05,
+      "loss": 0.6343,
+      "step": 2115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935404760872383e-05,
+      "loss": 0.5397,
+      "step": 2116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935291607917966e-05,
+      "loss": 0.5556,
+      "step": 2117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935178356265637e-05,
+      "loss": 0.6508,
+      "step": 2118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993506500591653e-05,
+      "loss": 0.6039,
+      "step": 2119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934951556871766e-05,
+      "loss": 0.6439,
+      "step": 2120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993483800913247e-05,
+      "loss": 0.6599,
+      "step": 2121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934724362699772e-05,
+      "loss": 0.6914,
+      "step": 2122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99346106175748e-05,
+      "loss": 0.62,
+      "step": 2123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934496773758688e-05,
+      "loss": 0.6639,
+      "step": 2124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934382831252565e-05,
+      "loss": 0.675,
+      "step": 2125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934268790057563e-05,
+      "loss": 0.579,
+      "step": 2126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934154650174813e-05,
+      "loss": 0.5531,
+      "step": 2127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934040411605448e-05,
+      "loss": 0.6087,
+      "step": 2128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933926074350607e-05,
+      "loss": 0.6235,
+      "step": 2129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933811638411425e-05,
+      "loss": 0.6823,
+      "step": 2130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933697103789035e-05,
+      "loss": 0.6734,
+      "step": 2131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993358247048458e-05,
+      "loss": 0.6361,
+      "step": 2132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993346773849919e-05,
+      "loss": 0.5692,
+      "step": 2133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933352907834016e-05,
+      "loss": 0.5807,
+      "step": 2134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933237978490193e-05,
+      "loss": 0.641,
+      "step": 2135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993312295046886e-05,
+      "loss": 0.5964,
+      "step": 2136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993300782377117e-05,
+      "loss": 0.5878,
+      "step": 2137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932892598398252e-05,
+      "loss": 0.6336,
+      "step": 2138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993277727435126e-05,
+      "loss": 0.5801,
+      "step": 2139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932661851631343e-05,
+      "loss": 0.5971,
+      "step": 2140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993254633023964e-05,
+      "loss": 0.6602,
+      "step": 2141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932430710177296e-05,
+      "loss": 0.7373,
+      "step": 2142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932314991445472e-05,
+      "loss": 0.6326,
+      "step": 2143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932199174045307e-05,
+      "loss": 0.5618,
+      "step": 2144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932083257977954e-05,
+      "loss": 0.6901,
+      "step": 2145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931967243244567e-05,
+      "loss": 0.5574,
+      "step": 2146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931851129846296e-05,
+      "loss": 0.5868,
+      "step": 2147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931734917784295e-05,
+      "loss": 0.7244,
+      "step": 2148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993161860705972e-05,
+      "loss": 0.7179,
+      "step": 2149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931502197673725e-05,
+      "loss": 0.6,
+      "step": 2150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931385689627465e-05,
+      "loss": 0.5161,
+      "step": 2151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99312690829221e-05,
+      "loss": 0.5671,
+      "step": 2152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931152377558785e-05,
+      "loss": 0.6855,
+      "step": 2153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931035573538685e-05,
+      "loss": 0.6195,
+      "step": 2154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993091867086296e-05,
+      "loss": 0.568,
+      "step": 2155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930801669532763e-05,
+      "loss": 0.5953,
+      "step": 2156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930684569549265e-05,
+      "loss": 0.5441,
+      "step": 2157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930567370913625e-05,
+      "loss": 0.7792,
+      "step": 2158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993045007362701e-05,
+      "loss": 0.5527,
+      "step": 2159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993033267769058e-05,
+      "loss": 0.5905,
+      "step": 2160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930215183105503e-05,
+      "loss": 0.6611,
+      "step": 2161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993009758987295e-05,
+      "loss": 0.6125,
+      "step": 2162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992997989799409e-05,
+      "loss": 0.5743,
+      "step": 2163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929862107470087e-05,
+      "loss": 0.6088,
+      "step": 2164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929744218302116e-05,
+      "loss": 0.6048,
+      "step": 2165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992962623049134e-05,
+      "loss": 0.6354,
+      "step": 2166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929508144038943e-05,
+      "loss": 0.5832,
+      "step": 2167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929389958946087e-05,
+      "loss": 0.5849,
+      "step": 2168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929271675213954e-05,
+      "loss": 0.5884,
+      "step": 2169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929153292843715e-05,
+      "loss": 0.5001,
+      "step": 2170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929034811836544e-05,
+      "loss": 0.6463,
+      "step": 2171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928916232193624e-05,
+      "loss": 0.7039,
+      "step": 2172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992879755391613e-05,
+      "loss": 0.6094,
+      "step": 2173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928678777005238e-05,
+      "loss": 0.753,
+      "step": 2174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992855990146213e-05,
+      "loss": 0.6719,
+      "step": 2175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928440927287994e-05,
+      "loss": 0.6046,
+      "step": 2176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928321854484002e-05,
+      "loss": 0.6211,
+      "step": 2177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992820268305134e-05,
+      "loss": 0.7139,
+      "step": 2178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992808341299119e-05,
+      "loss": 0.6079,
+      "step": 2179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992796404430474e-05,
+      "loss": 0.6968,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927844576993175e-05,
+      "loss": 0.658,
+      "step": 2181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927725011057687e-05,
+      "loss": 0.5677,
+      "step": 2182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927605346499453e-05,
+      "loss": 0.6425,
+      "step": 2183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992748558331967e-05,
+      "loss": 0.569,
+      "step": 2184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992736572151952e-05,
+      "loss": 0.6888,
+      "step": 2185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927245761100202e-05,
+      "loss": 0.589,
+      "step": 2186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927125702062905e-05,
+      "loss": 0.5489,
+      "step": 2187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927005544408822e-05,
+      "loss": 0.6304,
+      "step": 2188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926885288139144e-05,
+      "loss": 0.5659,
+      "step": 2189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992676493325507e-05,
+      "loss": 0.6244,
+      "step": 2190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926644479757788e-05,
+      "loss": 0.5607,
+      "step": 2191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99265239276485e-05,
+      "loss": 0.6446,
+      "step": 2192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926403276928407e-05,
+      "loss": 0.6306,
+      "step": 2193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926282527598703e-05,
+      "loss": 0.6711,
+      "step": 2194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926161679660588e-05,
+      "loss": 0.6168,
+      "step": 2195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992604073311526e-05,
+      "loss": 0.5281,
+      "step": 2196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925919687963924e-05,
+      "loss": 0.6835,
+      "step": 2197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925798544207785e-05,
+      "loss": 0.6195,
+      "step": 2198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992567730184804e-05,
+      "loss": 0.6357,
+      "step": 2199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99255559608859e-05,
+      "loss": 0.6906,
+      "step": 2200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925434521322562e-05,
+      "loss": 0.529,
+      "step": 2201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925312983159243e-05,
+      "loss": 0.5894,
+      "step": 2202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925191346397138e-05,
+      "loss": 0.5967,
+      "step": 2203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925069611037468e-05,
+      "loss": 0.5897,
+      "step": 2204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924947777081434e-05,
+      "loss": 0.6193,
+      "step": 2205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924825844530247e-05,
+      "loss": 0.6547,
+      "step": 2206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924703813385123e-05,
+      "loss": 0.6415,
+      "step": 2207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992458168364727e-05,
+      "loss": 0.6138,
+      "step": 2208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924459455317905e-05,
+      "loss": 0.5769,
+      "step": 2209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924337128398234e-05,
+      "loss": 0.6223,
+      "step": 2210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992421470288948e-05,
+      "loss": 0.643,
+      "step": 2211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924092178792862e-05,
+      "loss": 0.5225,
+      "step": 2212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923969556109588e-05,
+      "loss": 0.633,
+      "step": 2213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992384683484088e-05,
+      "loss": 0.5436,
+      "step": 2214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923724014987963e-05,
+      "loss": 0.5567,
+      "step": 2215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923601096552045e-05,
+      "loss": 0.6027,
+      "step": 2216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923478079534354e-05,
+      "loss": 0.5661,
+      "step": 2217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923354963936115e-05,
+      "loss": 0.652,
+      "step": 2218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923231749758546e-05,
+      "loss": 0.6618,
+      "step": 2219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923108437002876e-05,
+      "loss": 0.658,
+      "step": 2220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922985025670324e-05,
+      "loss": 0.598,
+      "step": 2221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992286151576212e-05,
+      "loss": 0.6757,
+      "step": 2222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992273790727949e-05,
+      "loss": 0.6111,
+      "step": 2223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992261420022366e-05,
+      "loss": 0.5455,
+      "step": 2224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922490394595862e-05,
+      "loss": 0.5837,
+      "step": 2225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922366490397326e-05,
+      "loss": 0.6455,
+      "step": 2226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992224248762928e-05,
+      "loss": 0.6553,
+      "step": 2227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922118386292956e-05,
+      "loss": 0.6503,
+      "step": 2228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921994186389587e-05,
+      "loss": 0.5961,
+      "step": 2229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921869887920412e-05,
+      "loss": 0.6748,
+      "step": 2230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921745490886658e-05,
+      "loss": 0.563,
+      "step": 2231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992162099528957e-05,
+      "loss": 0.5924,
+      "step": 2232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921496401130373e-05,
+      "loss": 0.6317,
+      "step": 2233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921371708410313e-05,
+      "loss": 0.668,
+      "step": 2234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921246917130627e-05,
+      "loss": 0.5795,
+      "step": 2235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921122027292555e-05,
+      "loss": 0.556,
+      "step": 2236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920997038897335e-05,
+      "loss": 0.579,
+      "step": 2237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920871951946213e-05,
+      "loss": 0.6519,
+      "step": 2238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992074676644043e-05,
+      "loss": 0.6343,
+      "step": 2239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920621482381223e-05,
+      "loss": 0.5977,
+      "step": 2240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992049609976985e-05,
+      "loss": 0.5737,
+      "step": 2241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920370618607543e-05,
+      "loss": 0.5874,
+      "step": 2242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.992024503889556e-05,
+      "loss": 0.5375,
+      "step": 2243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920119360635137e-05,
+      "loss": 0.554,
+      "step": 2244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991999358382753e-05,
+      "loss": 0.6242,
+      "step": 2245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919867708473992e-05,
+      "loss": 0.6759,
+      "step": 2246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919741734575765e-05,
+      "loss": 0.7064,
+      "step": 2247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919615662134104e-05,
+      "loss": 0.5948,
+      "step": 2248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991948949115026e-05,
+      "loss": 0.6589,
+      "step": 2249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919363221625493e-05,
+      "loss": 0.6287,
+      "step": 2250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919236853561046e-05,
+      "loss": 0.5725,
+      "step": 2251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919110386958184e-05,
+      "loss": 0.6351,
+      "step": 2252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918983821818156e-05,
+      "loss": 0.6504,
+      "step": 2253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918857158142226e-05,
+      "loss": 0.5768,
+      "step": 2254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918730395931648e-05,
+      "loss": 0.5176,
+      "step": 2255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918603535187686e-05,
+      "loss": 0.6201,
+      "step": 2256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918476575911592e-05,
+      "loss": 0.617,
+      "step": 2257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918349518104635e-05,
+      "loss": 0.5241,
+      "step": 2258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918222361768075e-05,
+      "loss": 0.6304,
+      "step": 2259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918095106903172e-05,
+      "loss": 0.5679,
+      "step": 2260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917967753511197e-05,
+      "loss": 0.5451,
+      "step": 2261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917840301593406e-05,
+      "loss": 0.5643,
+      "step": 2262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991771275115107e-05,
+      "loss": 0.6335,
+      "step": 2263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991758510218546e-05,
+      "loss": 0.5635,
+      "step": 2264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917457354697838e-05,
+      "loss": 0.6277,
+      "step": 2265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917329508689474e-05,
+      "loss": 0.6149,
+      "step": 2266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991720156416164e-05,
+      "loss": 0.5905,
+      "step": 2267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991707352111561e-05,
+      "loss": 0.6094,
+      "step": 2268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916945379552647e-05,
+      "loss": 0.5588,
+      "step": 2269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991681713947403e-05,
+      "loss": 0.5809,
+      "step": 2270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991668880088103e-05,
+      "loss": 0.6055,
+      "step": 2271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991656036377493e-05,
+      "loss": 0.6197,
+      "step": 2272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916431828156994e-05,
+      "loss": 0.6151,
+      "step": 2273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916303194028512e-05,
+      "loss": 0.6446,
+      "step": 2274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991617446139075e-05,
+      "loss": 0.5509,
+      "step": 2275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991604563024499e-05,
+      "loss": 0.5672,
+      "step": 2276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915916700592515e-05,
+      "loss": 0.6068,
+      "step": 2277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915787672434603e-05,
+      "loss": 0.5461,
+      "step": 2278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915658545772535e-05,
+      "loss": 0.6344,
+      "step": 2279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99155293206076e-05,
+      "loss": 0.6269,
+      "step": 2280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915399996941075e-05,
+      "loss": 0.6363,
+      "step": 2281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915270574774247e-05,
+      "loss": 0.621,
+      "step": 2282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915141054108404e-05,
+      "loss": 0.522,
+      "step": 2283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991501143494483e-05,
+      "loss": 0.5515,
+      "step": 2284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914881717284812e-05,
+      "loss": 0.6429,
+      "step": 2285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914751901129642e-05,
+      "loss": 0.6284,
+      "step": 2286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914621986480603e-05,
+      "loss": 0.6459,
+      "step": 2287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914491973338993e-05,
+      "loss": 0.6222,
+      "step": 2288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914361861706096e-05,
+      "loss": 0.5993,
+      "step": 2289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914231651583216e-05,
+      "loss": 0.5257,
+      "step": 2290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914101342971635e-05,
+      "loss": 0.7273,
+      "step": 2291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913970935872654e-05,
+      "loss": 0.583,
+      "step": 2292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913840430287562e-05,
+      "loss": 0.5479,
+      "step": 2293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913709826217665e-05,
+      "loss": 0.6431,
+      "step": 2294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913579123664253e-05,
+      "loss": 0.6971,
+      "step": 2295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913448322628625e-05,
+      "loss": 0.6012,
+      "step": 2296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913317423112088e-05,
+      "loss": 0.5573,
+      "step": 2297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913186425115932e-05,
+      "loss": 0.5747,
+      "step": 2298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913055328641463e-05,
+      "loss": 0.6116,
+      "step": 2299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912924133689984e-05,
+      "loss": 0.6413,
+      "step": 2300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912792840262798e-05,
+      "loss": 0.6343,
+      "step": 2301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912661448361207e-05,
+      "loss": 0.63,
+      "step": 2302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991252995798652e-05,
+      "loss": 0.5701,
+      "step": 2303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991239836914004e-05,
+      "loss": 0.6253,
+      "step": 2304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912266681823077e-05,
+      "loss": 0.5325,
+      "step": 2305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912134896036936e-05,
+      "loss": 0.5969,
+      "step": 2306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912003011782926e-05,
+      "loss": 0.7221,
+      "step": 2307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911871029062363e-05,
+      "loss": 0.5635,
+      "step": 2308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911738947876555e-05,
+      "loss": 0.5768,
+      "step": 2309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991160676822681e-05,
+      "loss": 0.6552,
+      "step": 2310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911474490114445e-05,
+      "loss": 0.6161,
+      "step": 2311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911342113540774e-05,
+      "loss": 0.6374,
+      "step": 2312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911209638507108e-05,
+      "loss": 0.6049,
+      "step": 2313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991107706501477e-05,
+      "loss": 0.6552,
+      "step": 2314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910944393065073e-05,
+      "loss": 0.7075,
+      "step": 2315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910811622659333e-05,
+      "loss": 0.6179,
+      "step": 2316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910678753798874e-05,
+      "loss": 0.6067,
+      "step": 2317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991054578648501e-05,
+      "loss": 0.6348,
+      "step": 2318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991041272071907e-05,
+      "loss": 0.6502,
+      "step": 2319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910279556502368e-05,
+      "loss": 0.5751,
+      "step": 2320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991014629383623e-05,
+      "loss": 0.6277,
+      "step": 2321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910012932721982e-05,
+      "loss": 0.5808,
+      "step": 2322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909879473160942e-05,
+      "loss": 0.5971,
+      "step": 2323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909745915154444e-05,
+      "loss": 0.6888,
+      "step": 2324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990961225870381e-05,
+      "loss": 0.5465,
+      "step": 2325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990947850381037e-05,
+      "loss": 0.6383,
+      "step": 2326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909344650475448e-05,
+      "loss": 0.6096,
+      "step": 2327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990921069870038e-05,
+      "loss": 0.6582,
+      "step": 2328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909076648486496e-05,
+      "loss": 0.6971,
+      "step": 2329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990894249983512e-05,
+      "loss": 0.5754,
+      "step": 2330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908808252747594e-05,
+      "loss": 0.6418,
+      "step": 2331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908673907225247e-05,
+      "loss": 0.6206,
+      "step": 2332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908539463269418e-05,
+      "loss": 0.5603,
+      "step": 2333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908404920881435e-05,
+      "loss": 0.6513,
+      "step": 2334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908270280062643e-05,
+      "loss": 0.5691,
+      "step": 2335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990813554081437e-05,
+      "loss": 0.5675,
+      "step": 2336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908000703137963e-05,
+      "loss": 0.5398,
+      "step": 2337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990786576703476e-05,
+      "loss": 0.5997,
+      "step": 2338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907730732506098e-05,
+      "loss": 0.6079,
+      "step": 2339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907595599553317e-05,
+      "loss": 0.5757,
+      "step": 2340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990746036817776e-05,
+      "loss": 0.623,
+      "step": 2341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990732503838078e-05,
+      "loss": 0.6627,
+      "step": 2342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907189610163712e-05,
+      "loss": 0.6018,
+      "step": 2343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907054083527903e-05,
+      "loss": 0.6145,
+      "step": 2344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99069184584747e-05,
+      "loss": 0.5432,
+      "step": 2345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990678273500545e-05,
+      "loss": 0.717,
+      "step": 2346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906646913121503e-05,
+      "loss": 0.6026,
+      "step": 2347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906510992824205e-05,
+      "loss": 0.52,
+      "step": 2348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990637497411491e-05,
+      "loss": 0.5791,
+      "step": 2349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990623885699497e-05,
+      "loss": 0.6628,
+      "step": 2350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906102641465726e-05,
+      "loss": 0.7822,
+      "step": 2351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905966327528548e-05,
+      "loss": 0.5795,
+      "step": 2352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905829915184775e-05,
+      "loss": 0.5623,
+      "step": 2353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990569340443577e-05,
+      "loss": 0.5876,
+      "step": 2354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905556795282894e-05,
+      "loss": 0.5971,
+      "step": 2355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905420087727493e-05,
+      "loss": 0.6627,
+      "step": 2356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905283281770935e-05,
+      "loss": 0.601,
+      "step": 2357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905146377414567e-05,
+      "loss": 0.5921,
+      "step": 2358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905009374659764e-05,
+      "loss": 0.6511,
+      "step": 2359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904872273507877e-05,
+      "loss": 0.5706,
+      "step": 2360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904735073960266e-05,
+      "loss": 0.5534,
+      "step": 2361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904597776018302e-05,
+      "loss": 0.6426,
+      "step": 2362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904460379683345e-05,
+      "loss": 0.596,
+      "step": 2363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904322884956756e-05,
+      "loss": 0.5895,
+      "step": 2364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990418529183991e-05,
+      "loss": 0.5364,
+      "step": 2365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904047600334167e-05,
+      "loss": 0.6057,
+      "step": 2366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903909810440896e-05,
+      "loss": 0.5813,
+      "step": 2367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903771922161464e-05,
+      "loss": 0.6356,
+      "step": 2368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990363393549725e-05,
+      "loss": 0.6454,
+      "step": 2369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903495850449614e-05,
+      "loss": 0.703,
+      "step": 2370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990335766701993e-05,
+      "loss": 0.5914,
+      "step": 2371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903219385209573e-05,
+      "loss": 0.6673,
+      "step": 2372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9903081005019918e-05,
+      "loss": 0.5696,
+      "step": 2373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990294252645234e-05,
+      "loss": 0.6193,
+      "step": 2374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902803949508206e-05,
+      "loss": 0.6093,
+      "step": 2375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99026652741889e-05,
+      "loss": 0.5464,
+      "step": 2376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902526500495806e-05,
+      "loss": 0.7115,
+      "step": 2377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990238762843029e-05,
+      "loss": 0.6691,
+      "step": 2378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902248657993735e-05,
+      "loss": 0.619,
+      "step": 2379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902109589187528e-05,
+      "loss": 0.494,
+      "step": 2380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901970422013042e-05,
+      "loss": 0.592,
+      "step": 2381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901831156471666e-05,
+      "loss": 0.5549,
+      "step": 2382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990169179256478e-05,
+      "loss": 0.5983,
+      "step": 2383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990155233029377e-05,
+      "loss": 0.5775,
+      "step": 2384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901412769660023e-05,
+      "loss": 0.5477,
+      "step": 2385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990127311066492e-05,
+      "loss": 0.7162,
+      "step": 2386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901133353309848e-05,
+      "loss": 0.5742,
+      "step": 2387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900993497596205e-05,
+      "loss": 0.7064,
+      "step": 2388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900853543525375e-05,
+      "loss": 0.6779,
+      "step": 2389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900713491098745e-05,
+      "loss": 0.6645,
+      "step": 2390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900573340317706e-05,
+      "loss": 0.5982,
+      "step": 2391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900433091183656e-05,
+      "loss": 0.5539,
+      "step": 2392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990029274369799e-05,
+      "loss": 0.6254,
+      "step": 2393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990015229786209e-05,
+      "loss": 0.5703,
+      "step": 2394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900011753677363e-05,
+      "loss": 0.4367,
+      "step": 2395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98998711111452e-05,
+      "loss": 0.5893,
+      "step": 2396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899730370266996e-05,
+      "loss": 0.692,
+      "step": 2397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899589531044157e-05,
+      "loss": 0.5964,
+      "step": 2398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899448593478076e-05,
+      "loss": 0.6115,
+      "step": 2399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899307557570154e-05,
+      "loss": 0.6176,
+      "step": 2400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899166423321793e-05,
+      "loss": 0.5992,
+      "step": 2401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899025190734393e-05,
+      "loss": 0.6294,
+      "step": 2402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989888385980936e-05,
+      "loss": 0.6609,
+      "step": 2403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98987424305481e-05,
+      "loss": 0.5204,
+      "step": 2404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989860090295201e-05,
+      "loss": 0.5651,
+      "step": 2405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98984592770225e-05,
+      "loss": 0.6162,
+      "step": 2406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989831755276098e-05,
+      "loss": 0.6295,
+      "step": 2407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898175730168852e-05,
+      "loss": 0.5168,
+      "step": 2408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989803380924753e-05,
+      "loss": 0.5166,
+      "step": 2409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989789178999842e-05,
+      "loss": 0.6019,
+      "step": 2410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989774967242294e-05,
+      "loss": 0.5802,
+      "step": 2411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897607456522495e-05,
+      "loss": 0.5954,
+      "step": 2412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897465142298497e-05,
+      "loss": 0.6029,
+      "step": 2413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897322729752362e-05,
+      "loss": 0.6013,
+      "step": 2414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989718021888551e-05,
+      "loss": 0.5715,
+      "step": 2415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897037609699347e-05,
+      "loss": 0.6155,
+      "step": 2416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896894902195293e-05,
+      "loss": 0.7028,
+      "step": 2417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896752096374773e-05,
+      "loss": 0.5816,
+      "step": 2418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896609192239196e-05,
+      "loss": 0.6605,
+      "step": 2419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896466189789985e-05,
+      "loss": 0.6187,
+      "step": 2420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989632308902856e-05,
+      "loss": 0.5969,
+      "step": 2421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896179889956345e-05,
+      "loss": 0.6717,
+      "step": 2422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896036592574762e-05,
+      "loss": 0.6012,
+      "step": 2423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989589319688523e-05,
+      "loss": 0.5899,
+      "step": 2424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989574970288918e-05,
+      "loss": 0.6865,
+      "step": 2425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895606110588035e-05,
+      "loss": 0.543,
+      "step": 2426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895462419983217e-05,
+      "loss": 0.5903,
+      "step": 2427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895318631076163e-05,
+      "loss": 0.5512,
+      "step": 2428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895174743868293e-05,
+      "loss": 0.5498,
+      "step": 2429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989503075836104e-05,
+      "loss": 0.5264,
+      "step": 2430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894886674555833e-05,
+      "loss": 0.6434,
+      "step": 2431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894742492454103e-05,
+      "loss": 0.6458,
+      "step": 2432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894598212057282e-05,
+      "loss": 0.5804,
+      "step": 2433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894453833366808e-05,
+      "loss": 0.6924,
+      "step": 2434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989430935638411e-05,
+      "loss": 0.6815,
+      "step": 2435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894164781110624e-05,
+      "loss": 0.611,
+      "step": 2436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989402010754779e-05,
+      "loss": 0.5633,
+      "step": 2437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989387533569704e-05,
+      "loss": 0.5793,
+      "step": 2438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893730465559816e-05,
+      "loss": 0.5691,
+      "step": 2439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893585497137556e-05,
+      "loss": 0.6487,
+      "step": 2440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.98934404304317e-05,
+      "loss": 0.5678,
+      "step": 2441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893295265443688e-05,
+      "loss": 0.5201,
+      "step": 2442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893150002174965e-05,
+      "loss": 0.6389,
+      "step": 2443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989300464062697e-05,
+      "loss": 0.5718,
+      "step": 2444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892859180801146e-05,
+      "loss": 0.6958,
+      "step": 2445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892713622698947e-05,
+      "loss": 0.518,
+      "step": 2446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989256796632181e-05,
+      "loss": 0.5899,
+      "step": 2447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892422211671185e-05,
+      "loss": 0.6717,
+      "step": 2448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989227635874852e-05,
+      "loss": 0.6842,
+      "step": 2449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892130407555267e-05,
+      "loss": 0.5694,
+      "step": 2450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989198435809287e-05,
+      "loss": 0.5776,
+      "step": 2451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891838210362783e-05,
+      "loss": 0.5796,
+      "step": 2452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989169196436646e-05,
+      "loss": 0.6328,
+      "step": 2453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891545620105347e-05,
+      "loss": 0.6317,
+      "step": 2454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989139917758091e-05,
+      "loss": 0.6256,
+      "step": 2455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891252636794588e-05,
+      "loss": 0.5975,
+      "step": 2456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891105997747848e-05,
+      "loss": 0.7087,
+      "step": 2457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989095926044214e-05,
+      "loss": 0.5422,
+      "step": 2458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890812424878932e-05,
+      "loss": 0.5909,
+      "step": 2459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890665491059667e-05,
+      "loss": 0.5392,
+      "step": 2460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989051845898582e-05,
+      "loss": 0.528,
+      "step": 2461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890371328658845e-05,
+      "loss": 0.55,
+      "step": 2462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.98902241000802e-05,
+      "loss": 0.6956,
+      "step": 2463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890076773251352e-05,
+      "loss": 0.6364,
+      "step": 2464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889929348173767e-05,
+      "loss": 0.7171,
+      "step": 2465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889781824848902e-05,
+      "loss": 0.6814,
+      "step": 2466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988963420327823e-05,
+      "loss": 0.6243,
+      "step": 2467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889486483463213e-05,
+      "loss": 0.5729,
+      "step": 2468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889338665405322e-05,
+      "loss": 0.6214,
+      "step": 2469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988919074910602e-05,
+      "loss": 0.7466,
+      "step": 2470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889042734566783e-05,
+      "loss": 0.62,
+      "step": 2471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888894621789077e-05,
+      "loss": 0.5717,
+      "step": 2472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988874641077437e-05,
+      "loss": 0.597,
+      "step": 2473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888598101524145e-05,
+      "loss": 0.6232,
+      "step": 2474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888449694039866e-05,
+      "loss": 0.6584,
+      "step": 2475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988830118832301e-05,
+      "loss": 0.5707,
+      "step": 2476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888152584375053e-05,
+      "loss": 0.5708,
+      "step": 2477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988800388219747e-05,
+      "loss": 0.7008,
+      "step": 2478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988785508179174e-05,
+      "loss": 0.554,
+      "step": 2479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887706183159344e-05,
+      "loss": 0.5945,
+      "step": 2480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988755718630175e-05,
+      "loss": 0.6407,
+      "step": 2481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988740809122045e-05,
+      "loss": 0.6423,
+      "step": 2482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988725889791692e-05,
+      "loss": 0.5997,
+      "step": 2483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887109606392645e-05,
+      "loss": 0.493,
+      "step": 2484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886960216649102e-05,
+      "loss": 0.5631,
+      "step": 2485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988681072868778e-05,
+      "loss": 0.5612,
+      "step": 2486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886661142510163e-05,
+      "loss": 0.6181,
+      "step": 2487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988651145811774e-05,
+      "loss": 0.6649,
+      "step": 2488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988636167551199e-05,
+      "loss": 0.6417,
+      "step": 2489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988621179469441e-05,
+      "loss": 0.6423,
+      "step": 2490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886061815666483e-05,
+      "loss": 0.6785,
+      "step": 2491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.98859117384297e-05,
+      "loss": 0.5924,
+      "step": 2492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885761562985553e-05,
+      "loss": 0.5159,
+      "step": 2493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885611289335536e-05,
+      "loss": 0.6213,
+      "step": 2494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885460917481137e-05,
+      "loss": 0.6186,
+      "step": 2495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988531044742385e-05,
+      "loss": 0.606,
+      "step": 2496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988515987916517e-05,
+      "loss": 0.6199,
+      "step": 2497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.98850092127066e-05,
+      "loss": 0.5888,
+      "step": 2498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884858448049633e-05,
+      "loss": 0.594,
+      "step": 2499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884707585195763e-05,
+      "loss": 0.5767,
+      "step": 2500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988455662414649e-05,
+      "loss": 0.5743,
+      "step": 2501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988440556490331e-05,
+      "loss": 0.6308,
+      "step": 2502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884254407467733e-05,
+      "loss": 0.5493,
+      "step": 2503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884103151841254e-05,
+      "loss": 0.5449,
+      "step": 2504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883951798025377e-05,
+      "loss": 0.6131,
+      "step": 2505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883800346021606e-05,
+      "loss": 0.5925,
+      "step": 2506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883648795831447e-05,
+      "loss": 0.5902,
+      "step": 2507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883497147456404e-05,
+      "loss": 0.5539,
+      "step": 2508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988334540089798e-05,
+      "loss": 0.6252,
+      "step": 2509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883193556157687e-05,
+      "loss": 0.6006,
+      "step": 2510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883041613237032e-05,
+      "loss": 0.6098,
+      "step": 2511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882889572137527e-05,
+      "loss": 0.6607,
+      "step": 2512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988273743286068e-05,
+      "loss": 0.6143,
+      "step": 2513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882585195408e-05,
+      "loss": 0.6143,
+      "step": 2514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882432859781002e-05,
+      "loss": 0.5829,
+      "step": 2515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882280425981197e-05,
+      "loss": 0.6612,
+      "step": 2516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882127894010104e-05,
+      "loss": 0.6167,
+      "step": 2517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881975263869234e-05,
+      "loss": 0.6914,
+      "step": 2518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881822535560108e-05,
+      "loss": 0.6278,
+      "step": 2519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881669709084238e-05,
+      "loss": 0.6571,
+      "step": 2520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881516784443145e-05,
+      "loss": 0.5784,
+      "step": 2521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881363761638346e-05,
+      "loss": 0.5485,
+      "step": 2522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881210640671363e-05,
+      "loss": 0.6303,
+      "step": 2523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881057421543718e-05,
+      "loss": 0.605,
+      "step": 2524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880904104256927e-05,
+      "loss": 0.5673,
+      "step": 2525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880750688812523e-05,
+      "loss": 0.5618,
+      "step": 2526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880597175212024e-05,
+      "loss": 0.6266,
+      "step": 2527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880443563456953e-05,
+      "loss": 0.6329,
+      "step": 2528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988028985354884e-05,
+      "loss": 0.6368,
+      "step": 2529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880136045489215e-05,
+      "loss": 0.6439,
+      "step": 2530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879982139279596e-05,
+      "loss": 0.6074,
+      "step": 2531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987982813492152e-05,
+      "loss": 0.5913,
+      "step": 2532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879674032416514e-05,
+      "loss": 0.5598,
+      "step": 2533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879519831766107e-05,
+      "loss": 0.6718,
+      "step": 2534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987936553297184e-05,
+      "loss": 0.5646,
+      "step": 2535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879211136035234e-05,
+      "loss": 0.5445,
+      "step": 2536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879056640957828e-05,
+      "loss": 0.5807,
+      "step": 2537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878902047741158e-05,
+      "loss": 0.5371,
+      "step": 2538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878747356386756e-05,
+      "loss": 0.7058,
+      "step": 2539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987859256689616e-05,
+      "loss": 0.6575,
+      "step": 2540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878437679270913e-05,
+      "loss": 0.5723,
+      "step": 2541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987828269351255e-05,
+      "loss": 0.7124,
+      "step": 2542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878127609622608e-05,
+      "loss": 0.593,
+      "step": 2543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877972427602627e-05,
+      "loss": 0.5644,
+      "step": 2544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877817147454156e-05,
+      "loss": 0.6052,
+      "step": 2545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987766176917873e-05,
+      "loss": 0.6404,
+      "step": 2546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877506292777897e-05,
+      "loss": 0.5712,
+      "step": 2547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98773507182532e-05,
+      "loss": 0.5517,
+      "step": 2548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987719504560618e-05,
+      "loss": 0.5375,
+      "step": 2549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877039274838393e-05,
+      "loss": 0.5722,
+      "step": 2550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 0.6042,
+      "step": 2551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987672743894669e-05,
+      "loss": 0.6811,
+      "step": 2552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987657137382587e-05,
+      "loss": 0.7129,
+      "step": 2553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987641521059048e-05,
+      "loss": 0.5634,
+      "step": 2554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987625894924206e-05,
+      "loss": 0.5772,
+      "step": 2555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987610258978217e-05,
+      "loss": 0.5319,
+      "step": 2556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987594613221236e-05,
+      "loss": 0.6653,
+      "step": 2557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875789576534182e-05,
+      "loss": 0.5733,
+      "step": 2558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875632922749193e-05,
+      "loss": 0.5344,
+      "step": 2559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987547617085895e-05,
+      "loss": 0.6102,
+      "step": 2560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875319320865016e-05,
+      "loss": 0.5526,
+      "step": 2561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987516237276894e-05,
+      "loss": 0.6969,
+      "step": 2562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875005326572285e-05,
+      "loss": 0.5154,
+      "step": 2563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874848182276607e-05,
+      "loss": 0.5776,
+      "step": 2564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874690939883473e-05,
+      "loss": 0.741,
+      "step": 2565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874533599394447e-05,
+      "loss": 0.6257,
+      "step": 2566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874376160811084e-05,
+      "loss": 0.5959,
+      "step": 2567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874218624134953e-05,
+      "loss": 0.5757,
+      "step": 2568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874060989367616e-05,
+      "loss": 0.5925,
+      "step": 2569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987390325651064e-05,
+      "loss": 0.7227,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873745425565594e-05,
+      "loss": 0.5933,
+      "step": 2571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873587496534048e-05,
+      "loss": 0.5436,
+      "step": 2572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873429469417563e-05,
+      "loss": 0.5798,
+      "step": 2573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873271344217716e-05,
+      "loss": 0.619,
+      "step": 2574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873113120936074e-05,
+      "loss": 0.5957,
+      "step": 2575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987295479957421e-05,
+      "loss": 0.5831,
+      "step": 2576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98727963801337e-05,
+      "loss": 0.5653,
+      "step": 2577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987263786261611e-05,
+      "loss": 0.5475,
+      "step": 2578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987247924702302e-05,
+      "loss": 0.5767,
+      "step": 2579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872320533356004e-05,
+      "loss": 0.5592,
+      "step": 2580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872161721616644e-05,
+      "loss": 0.5379,
+      "step": 2581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872002811806508e-05,
+      "loss": 0.6332,
+      "step": 2582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871843803927182e-05,
+      "loss": 0.7571,
+      "step": 2583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871684697980242e-05,
+      "loss": 0.6542,
+      "step": 2584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871525493967273e-05,
+      "loss": 0.6631,
+      "step": 2585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987136619188985e-05,
+      "loss": 0.6514,
+      "step": 2586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871206791749563e-05,
+      "loss": 0.6694,
+      "step": 2587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987104729354799e-05,
+      "loss": 0.6397,
+      "step": 2588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870887697286715e-05,
+      "loss": 0.5272,
+      "step": 2589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987072800296733e-05,
+      "loss": 0.6492,
+      "step": 2590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870568210591413e-05,
+      "loss": 0.6367,
+      "step": 2591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987040832016056e-05,
+      "loss": 0.524,
+      "step": 2592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987024833167635e-05,
+      "loss": 0.6351,
+      "step": 2593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870088245140376e-05,
+      "loss": 0.5991,
+      "step": 2594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869928060554234e-05,
+      "loss": 0.6353,
+      "step": 2595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869767777919504e-05,
+      "loss": 0.571,
+      "step": 2596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986960739723779e-05,
+      "loss": 0.5498,
+      "step": 2597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986944691851068e-05,
+      "loss": 0.5248,
+      "step": 2598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869286341739767e-05,
+      "loss": 0.6606,
+      "step": 2599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986912566692665e-05,
+      "loss": 0.64,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868964894072916e-05,
+      "loss": 0.582,
+      "step": 2601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868804023180174e-05,
+      "loss": 0.639,
+      "step": 2602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868643054250017e-05,
+      "loss": 0.5554,
+      "step": 2603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868481987284042e-05,
+      "loss": 0.5371,
+      "step": 2604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868320822283853e-05,
+      "loss": 0.6355,
+      "step": 2605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868159559251045e-05,
+      "loss": 0.6629,
+      "step": 2606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986799819818723e-05,
+      "loss": 0.5786,
+      "step": 2607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867836739094e-05,
+      "loss": 0.6684,
+      "step": 2608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867675181972968e-05,
+      "loss": 0.4864,
+      "step": 2609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986751352682573e-05,
+      "loss": 0.6645,
+      "step": 2610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98673517736539e-05,
+      "loss": 0.6768,
+      "step": 2611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986718992245908e-05,
+      "loss": 0.6058,
+      "step": 2612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986702797324288e-05,
+      "loss": 0.6857,
+      "step": 2613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986686592600691e-05,
+      "loss": 0.6074,
+      "step": 2614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866703780752777e-05,
+      "loss": 0.603,
+      "step": 2615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866541537482093e-05,
+      "loss": 0.5226,
+      "step": 2616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986637919619647e-05,
+      "loss": 0.5862,
+      "step": 2617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986621675689752e-05,
+      "loss": 0.6226,
+      "step": 2618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866054219586854e-05,
+      "loss": 0.5639,
+      "step": 2619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986589158426609e-05,
+      "loss": 0.5172,
+      "step": 2620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865728850936846e-05,
+      "loss": 0.5999,
+      "step": 2621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865566019600734e-05,
+      "loss": 0.6211,
+      "step": 2622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865403090259376e-05,
+      "loss": 0.6034,
+      "step": 2623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865240062914385e-05,
+      "loss": 0.5661,
+      "step": 2624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865076937567384e-05,
+      "loss": 0.5873,
+      "step": 2625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864913714219996e-05,
+      "loss": 0.564,
+      "step": 2626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864750392873837e-05,
+      "loss": 0.5469,
+      "step": 2627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986458697353053e-05,
+      "loss": 0.629,
+      "step": 2628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864423456191703e-05,
+      "loss": 0.6187,
+      "step": 2629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986425984085898e-05,
+      "loss": 0.5856,
+      "step": 2630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864096127533977e-05,
+      "loss": 0.6322,
+      "step": 2631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9863932316218333e-05,
+      "loss": 0.623,
+      "step": 2632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986376840691367e-05,
+      "loss": 0.5624,
+      "step": 2633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863604399621615e-05,
+      "loss": 0.6575,
+      "step": 2634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863440294343793e-05,
+      "loss": 0.5715,
+      "step": 2635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863276091081847e-05,
+      "loss": 0.6263,
+      "step": 2636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863111789837396e-05,
+      "loss": 0.506,
+      "step": 2637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986294739061208e-05,
+      "loss": 0.599,
+      "step": 2638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986278289340753e-05,
+      "loss": 0.5846,
+      "step": 2639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862618298225377e-05,
+      "loss": 0.538,
+      "step": 2640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986245360506726e-05,
+      "loss": 0.5533,
+      "step": 2641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862288813934808e-05,
+      "loss": 0.6385,
+      "step": 2642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862123924829668e-05,
+      "loss": 0.5383,
+      "step": 2643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986195893775347e-05,
+      "loss": 0.6533,
+      "step": 2644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986179385270786e-05,
+      "loss": 0.6482,
+      "step": 2645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861628669694473e-05,
+      "loss": 0.5834,
+      "step": 2646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861463388714954e-05,
+      "loss": 0.6354,
+      "step": 2647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986129800977094e-05,
+      "loss": 0.5638,
+      "step": 2648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861132532864076e-05,
+      "loss": 0.6359,
+      "step": 2649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860966957996007e-05,
+      "loss": 0.6997,
+      "step": 2650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860801285168377e-05,
+      "loss": 0.6134,
+      "step": 2651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860635514382827e-05,
+      "loss": 0.5514,
+      "step": 2652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860469645641014e-05,
+      "loss": 0.6296,
+      "step": 2653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860303678944577e-05,
+      "loss": 0.6028,
+      "step": 2654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986013761429517e-05,
+      "loss": 0.6972,
+      "step": 2655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985997145169444e-05,
+      "loss": 0.6057,
+      "step": 2656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859805191144036e-05,
+      "loss": 0.5651,
+      "step": 2657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859638832645614e-05,
+      "loss": 0.5831,
+      "step": 2658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859472376200826e-05,
+      "loss": 0.5832,
+      "step": 2659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985930582181132e-05,
+      "loss": 0.6339,
+      "step": 2660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985913916947876e-05,
+      "loss": 0.6721,
+      "step": 2661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858972419204796e-05,
+      "loss": 0.6429,
+      "step": 2662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858805570991077e-05,
+      "loss": 0.6356,
+      "step": 2663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858638624839275e-05,
+      "loss": 0.625,
+      "step": 2664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985847158075104e-05,
+      "loss": 0.6685,
+      "step": 2665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985830443872803e-05,
+      "loss": 0.6227,
+      "step": 2666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858137198771914e-05,
+      "loss": 0.5788,
+      "step": 2667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857969860884345e-05,
+      "loss": 0.6871,
+      "step": 2668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985780242506699e-05,
+      "loss": 0.6029,
+      "step": 2669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985763489132151e-05,
+      "loss": 0.6292,
+      "step": 2670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857467259649565e-05,
+      "loss": 0.5158,
+      "step": 2671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985729953005283e-05,
+      "loss": 0.6154,
+      "step": 2672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857131702532967e-05,
+      "loss": 0.5554,
+      "step": 2673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985696377709164e-05,
+      "loss": 0.5246,
+      "step": 2674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985679575373052e-05,
+      "loss": 0.53,
+      "step": 2675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856627632451278e-05,
+      "loss": 0.552,
+      "step": 2676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985645941325558e-05,
+      "loss": 0.5618,
+      "step": 2677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98562910961451e-05,
+      "loss": 0.6112,
+      "step": 2678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856122681121505e-05,
+      "loss": 0.5883,
+      "step": 2679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855954168186478e-05,
+      "loss": 0.7,
+      "step": 2680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855785557341685e-05,
+      "loss": 0.622,
+      "step": 2681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855616848588805e-05,
+      "loss": 0.595,
+      "step": 2682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855448041929507e-05,
+      "loss": 0.5876,
+      "step": 2683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855279137365476e-05,
+      "loss": 0.6848,
+      "step": 2684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985511013489839e-05,
+      "loss": 0.6261,
+      "step": 2685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854941034529922e-05,
+      "loss": 0.7243,
+      "step": 2686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854771836261757e-05,
+      "loss": 0.5163,
+      "step": 2687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985460254009557e-05,
+      "loss": 0.6142,
+      "step": 2688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985443314603305e-05,
+      "loss": 0.5989,
+      "step": 2689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854263654075873e-05,
+      "loss": 0.5043,
+      "step": 2690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854094064225728e-05,
+      "loss": 0.6404,
+      "step": 2691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853924376484297e-05,
+      "loss": 0.6818,
+      "step": 2692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853754590853267e-05,
+      "loss": 0.6723,
+      "step": 2693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853584707334323e-05,
+      "loss": 0.5828,
+      "step": 2694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853414725929157e-05,
+      "loss": 0.6316,
+      "step": 2695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853244646639452e-05,
+      "loss": 0.6672,
+      "step": 2696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853074469466897e-05,
+      "loss": 0.5948,
+      "step": 2697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852904194413186e-05,
+      "loss": 0.5915,
+      "step": 2698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852733821480014e-05,
+      "loss": 0.7841,
+      "step": 2699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852563350669064e-05,
+      "loss": 0.5531,
+      "step": 2700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852392781982038e-05,
+      "loss": 0.5795,
+      "step": 2701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985222211542063e-05,
+      "loss": 0.6686,
+      "step": 2702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852051350986528e-05,
+      "loss": 0.6014,
+      "step": 2703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851880488681436e-05,
+      "loss": 0.6611,
+      "step": 2704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851709528507047e-05,
+      "loss": 0.5795,
+      "step": 2705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851538470465062e-05,
+      "loss": 0.6471,
+      "step": 2706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851367314557178e-05,
+      "loss": 0.63,
+      "step": 2707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98511960607851e-05,
+      "loss": 0.6987,
+      "step": 2708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851024709150518e-05,
+      "loss": 0.6234,
+      "step": 2709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850853259655146e-05,
+      "loss": 0.5449,
+      "step": 2710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850681712300686e-05,
+      "loss": 0.5999,
+      "step": 2711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850510067088834e-05,
+      "loss": 0.585,
+      "step": 2712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850338324021306e-05,
+      "loss": 0.6609,
+      "step": 2713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98501664830998e-05,
+      "loss": 0.6986,
+      "step": 2714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984999454432602e-05,
+      "loss": 0.5842,
+      "step": 2715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849822507701687e-05,
+      "loss": 0.6221,
+      "step": 2716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98496503732285e-05,
+      "loss": 0.6504,
+      "step": 2717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849478140908173e-05,
+      "loss": 0.6075,
+      "step": 2718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984930581074241e-05,
+      "loss": 0.5776,
+      "step": 2719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849133382732935e-05,
+      "loss": 0.5272,
+      "step": 2720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848960856881453e-05,
+      "loss": 0.5521,
+      "step": 2721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848788233189674e-05,
+      "loss": 0.5194,
+      "step": 2722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848615511659325e-05,
+      "loss": 0.5787,
+      "step": 2723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848442692292113e-05,
+      "loss": 0.5213,
+      "step": 2724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848269775089756e-05,
+      "loss": 0.5797,
+      "step": 2725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848096760053973e-05,
+      "loss": 0.5293,
+      "step": 2726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847923647186484e-05,
+      "loss": 0.6147,
+      "step": 2727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847750436489003e-05,
+      "loss": 0.5709,
+      "step": 2728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984757712796326e-05,
+      "loss": 0.6546,
+      "step": 2729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847403721610963e-05,
+      "loss": 0.5783,
+      "step": 2730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984723021743385e-05,
+      "loss": 0.5404,
+      "step": 2731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847056615433636e-05,
+      "loss": 0.57,
+      "step": 2732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846882915612047e-05,
+      "loss": 0.6357,
+      "step": 2733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984670911797081e-05,
+      "loss": 0.491,
+      "step": 2734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846535222511648e-05,
+      "loss": 0.6229,
+      "step": 2735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846361229236294e-05,
+      "loss": 0.5126,
+      "step": 2736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984618713814647e-05,
+      "loss": 0.6425,
+      "step": 2737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984601294924391e-05,
+      "loss": 0.5703,
+      "step": 2738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984583866253034e-05,
+      "loss": 0.5559,
+      "step": 2739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845664278007496e-05,
+      "loss": 0.5595,
+      "step": 2740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984548979567711e-05,
+      "loss": 0.6173,
+      "step": 2741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845315215540915e-05,
+      "loss": 0.5968,
+      "step": 2742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845140537600645e-05,
+      "loss": 0.4773,
+      "step": 2743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844965761858033e-05,
+      "loss": 0.5855,
+      "step": 2744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844790888314812e-05,
+      "loss": 0.6123,
+      "step": 2745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844615916972733e-05,
+      "loss": 0.6925,
+      "step": 2746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984444084783352e-05,
+      "loss": 0.5876,
+      "step": 2747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844265680898917e-05,
+      "loss": 0.6352,
+      "step": 2748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844090416170666e-05,
+      "loss": 0.5852,
+      "step": 2749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843915053650507e-05,
+      "loss": 0.5728,
+      "step": 2750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843739593340184e-05,
+      "loss": 0.6011,
+      "step": 2751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984356403524143e-05,
+      "loss": 0.6026,
+      "step": 2752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843388379356002e-05,
+      "loss": 0.6349,
+      "step": 2753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843212625685643e-05,
+      "loss": 0.6559,
+      "step": 2754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984303677423209e-05,
+      "loss": 0.564,
+      "step": 2755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98428608249971e-05,
+      "loss": 0.5068,
+      "step": 2756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842684777982413e-05,
+      "loss": 0.5006,
+      "step": 2757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984250863318978e-05,
+      "loss": 0.6675,
+      "step": 2758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984233239062096e-05,
+      "loss": 0.6435,
+      "step": 2759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842156050277688e-05,
+      "loss": 0.5888,
+      "step": 2760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9841979612161726e-05,
+      "loss": 0.5194,
+      "step": 2761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9841803076274824e-05,
+      "loss": 0.582,
+      "step": 2762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841626442618738e-05,
+      "loss": 0.562,
+      "step": 2763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984144971119522e-05,
+      "loss": 0.6468,
+      "step": 2764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841272882006026e-05,
+      "loss": 0.573,
+      "step": 2765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841095955052913e-05,
+      "loss": 0.5848,
+      "step": 2766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984091893033764e-05,
+      "loss": 0.6348,
+      "step": 2767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840741807861963e-05,
+      "loss": 0.5945,
+      "step": 2768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840564587627643e-05,
+      "loss": 0.5,
+      "step": 2769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984038726963644e-05,
+      "loss": 0.6099,
+      "step": 2770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984020985389012e-05,
+      "loss": 0.5331,
+      "step": 2771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984003234039044e-05,
+      "loss": 0.6517,
+      "step": 2772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983985472913916e-05,
+      "loss": 0.5765,
+      "step": 2773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839677020138052e-05,
+      "loss": 0.6986,
+      "step": 2774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983949921338888e-05,
+      "loss": 0.5423,
+      "step": 2775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983932130889341e-05,
+      "loss": 0.5782,
+      "step": 2776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839143306653407e-05,
+      "loss": 0.5192,
+      "step": 2777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983896520667064e-05,
+      "loss": 0.61,
+      "step": 2778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983878700894688e-05,
+      "loss": 0.6173,
+      "step": 2779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983860871348389e-05,
+      "loss": 0.6094,
+      "step": 2780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838430320283454e-05,
+      "loss": 0.5323,
+      "step": 2781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983825182934734e-05,
+      "loss": 0.6181,
+      "step": 2782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838073240677312e-05,
+      "loss": 0.5695,
+      "step": 2783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983789455427515e-05,
+      "loss": 0.5732,
+      "step": 2784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837715770142633e-05,
+      "loss": 0.625,
+      "step": 2785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837536888281534e-05,
+      "loss": 0.5565,
+      "step": 2786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983735790869363e-05,
+      "loss": 0.5817,
+      "step": 2787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.98371788313807e-05,
+      "loss": 0.5746,
+      "step": 2788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836999656344517e-05,
+      "loss": 0.5483,
+      "step": 2789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983682038358687e-05,
+      "loss": 0.619,
+      "step": 2790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836641013109534e-05,
+      "loss": 0.6395,
+      "step": 2791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983646154491429e-05,
+      "loss": 0.7086,
+      "step": 2792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836281979002927e-05,
+      "loss": 0.667,
+      "step": 2793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836102315377228e-05,
+      "loss": 0.5058,
+      "step": 2794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983592255403897e-05,
+      "loss": 0.6393,
+      "step": 2795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835742694989942e-05,
+      "loss": 0.6093,
+      "step": 2796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835562738231938e-05,
+      "loss": 0.5541,
+      "step": 2797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835382683766737e-05,
+      "loss": 0.6055,
+      "step": 2798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983520253159613e-05,
+      "loss": 0.5432,
+      "step": 2799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983502228172191e-05,
+      "loss": 0.5035,
+      "step": 2800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834841934145865e-05,
+      "loss": 0.6053,
+      "step": 2801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834661488869787e-05,
+      "loss": 0.5608,
+      "step": 2802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834480945895468e-05,
+      "loss": 0.7356,
+      "step": 2803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834300305224698e-05,
+      "loss": 0.5197,
+      "step": 2804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983411956685928e-05,
+      "loss": 0.6146,
+      "step": 2805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833938730801003e-05,
+      "loss": 0.5721,
+      "step": 2806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833757797051666e-05,
+      "loss": 0.6494,
+      "step": 2807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833576765613064e-05,
+      "loss": 0.5595,
+      "step": 2808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833395636487e-05,
+      "loss": 0.5765,
+      "step": 2809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833214409675264e-05,
+      "loss": 0.5597,
+      "step": 2810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833033085179668e-05,
+      "loss": 0.65,
+      "step": 2811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983285166300201e-05,
+      "loss": 0.627,
+      "step": 2812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832670143144088e-05,
+      "loss": 0.6157,
+      "step": 2813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832488525607702e-05,
+      "loss": 0.5636,
+      "step": 2814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832306810394665e-05,
+      "loss": 0.6212,
+      "step": 2815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832124997506783e-05,
+      "loss": 0.646,
+      "step": 2816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983194308694585e-05,
+      "loss": 0.6932,
+      "step": 2817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983176107871369e-05,
+      "loss": 0.5825,
+      "step": 2818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831578972812097e-05,
+      "loss": 0.6023,
+      "step": 2819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983139676924289e-05,
+      "loss": 0.6701,
+      "step": 2820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831214468007868e-05,
+      "loss": 0.6873,
+      "step": 2821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983103206910885e-05,
+      "loss": 0.6166,
+      "step": 2822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983084957254765e-05,
+      "loss": 0.5393,
+      "step": 2823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983066697832607e-05,
+      "loss": 0.645,
+      "step": 2824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830484286445935e-05,
+      "loss": 0.6246,
+      "step": 2825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830301496909058e-05,
+      "loss": 0.6524,
+      "step": 2826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983011860971725e-05,
+      "loss": 0.6265,
+      "step": 2827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982993562487233e-05,
+      "loss": 0.5555,
+      "step": 2828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829752542376118e-05,
+      "loss": 0.5598,
+      "step": 2829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982956936223043e-05,
+      "loss": 0.5065,
+      "step": 2830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829386084437088e-05,
+      "loss": 0.6303,
+      "step": 2831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829202708997912e-05,
+      "loss": 0.616,
+      "step": 2832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982901923591472e-05,
+      "loss": 0.5452,
+      "step": 2833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982883566518934e-05,
+      "loss": 0.648,
+      "step": 2834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828651996823593e-05,
+      "loss": 0.5837,
+      "step": 2835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828468230819308e-05,
+      "loss": 0.5601,
+      "step": 2836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98282843671783e-05,
+      "loss": 0.6103,
+      "step": 2837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828100405902404e-05,
+      "loss": 0.6356,
+      "step": 2838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827916346993448e-05,
+      "loss": 0.4935,
+      "step": 2839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827732190453255e-05,
+      "loss": 0.6281,
+      "step": 2840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982754793628366e-05,
+      "loss": 0.6565,
+      "step": 2841
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982736358448649e-05,
+      "loss": 0.504,
+      "step": 2842
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827179135063576e-05,
+      "loss": 0.5609,
+      "step": 2843
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826994588016752e-05,
+      "loss": 0.4897,
+      "step": 2844
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982680994334785e-05,
+      "loss": 0.6391,
+      "step": 2845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982662520105871e-05,
+      "loss": 0.5416,
+      "step": 2846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826440361151155e-05,
+      "loss": 0.582,
+      "step": 2847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826255423627036e-05,
+      "loss": 0.6278,
+      "step": 2848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826070388488176e-05,
+      "loss": 0.5336,
+      "step": 2849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825885255736423e-05,
+      "loss": 0.5668,
+      "step": 2850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825700025373613e-05,
+      "loss": 0.5682,
+      "step": 2851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825514697401586e-05,
+      "loss": 0.5756,
+      "step": 2852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982532927182218e-05,
+      "loss": 0.5879,
+      "step": 2853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982514374863725e-05,
+      "loss": 0.4893,
+      "step": 2854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824958127848618e-05,
+      "loss": 0.5483,
+      "step": 2855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824772409458143e-05,
+      "loss": 0.6825,
+      "step": 2856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982458659346767e-05,
+      "loss": 0.6343,
+      "step": 2857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824400679879035e-05,
+      "loss": 0.556,
+      "step": 2858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824214668694095e-05,
+      "loss": 0.6299,
+      "step": 2859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982402855991469e-05,
+      "loss": 0.5487,
+      "step": 2860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823842353542675e-05,
+      "loss": 0.5608,
+      "step": 2861
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98236560495799e-05,
+      "loss": 0.6156,
+      "step": 2862
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823469648028207e-05,
+      "loss": 0.6511,
+      "step": 2863
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982328314888946e-05,
+      "loss": 0.6558,
+      "step": 2864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98230965521655e-05,
+      "loss": 0.5973,
+      "step": 2865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822909857858192e-05,
+      "loss": 0.6844,
+      "step": 2866
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822723065969378e-05,
+      "loss": 0.5986,
+      "step": 2867
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822536176500925e-05,
+      "loss": 0.5155,
+      "step": 2868
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822349189454683e-05,
+      "loss": 0.5004,
+      "step": 2869
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822162104832515e-05,
+      "loss": 0.5443,
+      "step": 2870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821974922636272e-05,
+      "loss": 0.5716,
+      "step": 2871
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982178764286782e-05,
+      "loss": 0.5933,
+      "step": 2872
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821600265529017e-05,
+      "loss": 0.5737,
+      "step": 2873
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982141279062172e-05,
+      "loss": 0.509,
+      "step": 2874
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98212252181478e-05,
+      "loss": 0.4328,
+      "step": 2875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821037548109118e-05,
+      "loss": 0.5987,
+      "step": 2876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820849780507534e-05,
+      "loss": 0.569,
+      "step": 2877
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820661915344916e-05,
+      "loss": 0.5822,
+      "step": 2878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982047395262313e-05,
+      "loss": 0.5285,
+      "step": 2879
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820285892344043e-05,
+      "loss": 0.6282,
+      "step": 2880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982009773450952e-05,
+      "loss": 0.5482,
+      "step": 2881
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819909479121437e-05,
+      "loss": 0.5971,
+      "step": 2882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819721126181665e-05,
+      "loss": 0.5535,
+      "step": 2883
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819532675692063e-05,
+      "loss": 0.5972,
+      "step": 2884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819344127654515e-05,
+      "loss": 0.5559,
+      "step": 2885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981915548207089e-05,
+      "loss": 0.5444,
+      "step": 2886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818966738943064e-05,
+      "loss": 0.5713,
+      "step": 2887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981877789827291e-05,
+      "loss": 0.6096,
+      "step": 2888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98185889600623e-05,
+      "loss": 0.6346,
+      "step": 2889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818399924313117e-05,
+      "loss": 0.631,
+      "step": 2890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818210791027237e-05,
+      "loss": 0.5585,
+      "step": 2891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818021560206537e-05,
+      "loss": 0.6378,
+      "step": 2892
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98178322318529e-05,
+      "loss": 0.7234,
+      "step": 2893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817642805968207e-05,
+      "loss": 0.5534,
+      "step": 2894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817453282554334e-05,
+      "loss": 0.5413,
+      "step": 2895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817263661613172e-05,
+      "loss": 0.6422,
+      "step": 2896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98170739431466e-05,
+      "loss": 0.5523,
+      "step": 2897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98168841271565e-05,
+      "loss": 0.6117,
+      "step": 2898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816694213644765e-05,
+      "loss": 0.6327,
+      "step": 2899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816504202613273e-05,
+      "loss": 0.5367,
+      "step": 2900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816314094063917e-05,
+      "loss": 0.5546,
+      "step": 2901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981612388799859e-05,
+      "loss": 0.5245,
+      "step": 2902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981593358441917e-05,
+      "loss": 0.525,
+      "step": 2903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815743183327557e-05,
+      "loss": 0.6051,
+      "step": 2904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815552684725635e-05,
+      "loss": 0.5911,
+      "step": 2905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815362088615306e-05,
+      "loss": 0.5592,
+      "step": 2906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815171394998453e-05,
+      "loss": 0.6309,
+      "step": 2907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981498060387698e-05,
+      "loss": 0.5848,
+      "step": 2908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814789715252773e-05,
+      "loss": 0.7162,
+      "step": 2909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814598729127736e-05,
+      "loss": 0.5839,
+      "step": 2910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814407645503762e-05,
+      "loss": 0.5867,
+      "step": 2911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814216464382752e-05,
+      "loss": 0.5883,
+      "step": 2912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814025185766603e-05,
+      "loss": 0.6293,
+      "step": 2913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813833809657215e-05,
+      "loss": 0.6695,
+      "step": 2914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813642336056488e-05,
+      "loss": 0.5055,
+      "step": 2915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981345076496633e-05,
+      "loss": 0.6352,
+      "step": 2916
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813259096388636e-05,
+      "loss": 0.6596,
+      "step": 2917
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813067330325317e-05,
+      "loss": 0.5301,
+      "step": 2918
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812875466778274e-05,
+      "loss": 0.6615,
+      "step": 2919
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812683505749415e-05,
+      "loss": 0.6845,
+      "step": 2920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812491447240646e-05,
+      "loss": 0.5125,
+      "step": 2921
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812299291253872e-05,
+      "loss": 0.5718,
+      "step": 2922
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981210703779101e-05,
+      "loss": 0.618,
+      "step": 2923
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811914686853963e-05,
+      "loss": 0.5694,
+      "step": 2924
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811722238444642e-05,
+      "loss": 0.5185,
+      "step": 2925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811529692564964e-05,
+      "loss": 0.5294,
+      "step": 2926
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811337049216838e-05,
+      "loss": 0.6757,
+      "step": 2927
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811144308402173e-05,
+      "loss": 0.5619,
+      "step": 2928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810951470122893e-05,
+      "loss": 0.5734,
+      "step": 2929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810758534380908e-05,
+      "loss": 0.5457,
+      "step": 2930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810565501178133e-05,
+      "loss": 0.5909,
+      "step": 2931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810372370516493e-05,
+      "loss": 0.6024,
+      "step": 2932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810179142397902e-05,
+      "loss": 0.5835,
+      "step": 2933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809985816824277e-05,
+      "loss": 0.5956,
+      "step": 2934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809792393797543e-05,
+      "loss": 0.5662,
+      "step": 2935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809598873319617e-05,
+      "loss": 0.6453,
+      "step": 2936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980940525539243e-05,
+      "loss": 0.5755,
+      "step": 2937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809211540017894e-05,
+      "loss": 0.675,
+      "step": 2938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809017727197937e-05,
+      "loss": 0.6563,
+      "step": 2939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808823816934494e-05,
+      "loss": 0.5208,
+      "step": 2940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980862980922948e-05,
+      "loss": 0.6087,
+      "step": 2941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808435704084825e-05,
+      "loss": 0.5448,
+      "step": 2942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808241501502458e-05,
+      "loss": 0.5569,
+      "step": 2943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980804720148431e-05,
+      "loss": 0.6229,
+      "step": 2944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.641,
+      "step": 2945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807658309148383e-05,
+      "loss": 0.5642,
+      "step": 2946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980746371683447e-05,
+      "loss": 0.7156,
+      "step": 2947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807269027092502e-05,
+      "loss": 0.5183,
+      "step": 2948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980707423992441e-05,
+      "loss": 0.6881,
+      "step": 2949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806879355332135e-05,
+      "loss": 0.5838,
+      "step": 2950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806684373317606e-05,
+      "loss": 0.6034,
+      "step": 2951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980648929388276e-05,
+      "loss": 0.6074,
+      "step": 2952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806294117029546e-05,
+      "loss": 0.5583,
+      "step": 2953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806098842759887e-05,
+      "loss": 0.5435,
+      "step": 2954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9805903471075735e-05,
+      "loss": 0.6189,
+      "step": 2955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9805708001979026e-05,
+      "loss": 0.545,
+      "step": 2956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9805512435471705e-05,
+      "loss": 0.6286,
+      "step": 2957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805316771555703e-05,
+      "loss": 0.6464,
+      "step": 2958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805121010232983e-05,
+      "loss": 0.5413,
+      "step": 2959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804925151505476e-05,
+      "loss": 0.5442,
+      "step": 2960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980472919537513e-05,
+      "loss": 0.5918,
+      "step": 2961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980453314184389e-05,
+      "loss": 0.5131,
+      "step": 2962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804336990913712e-05,
+      "loss": 0.6826,
+      "step": 2963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804140742586536e-05,
+      "loss": 0.6092,
+      "step": 2964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803944396864313e-05,
+      "loss": 0.5994,
+      "step": 2965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803747953749e-05,
+      "loss": 0.5957,
+      "step": 2966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980355141324254e-05,
+      "loss": 0.5722,
+      "step": 2967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980335477534689e-05,
+      "loss": 0.5779,
+      "step": 2968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803158040064003e-05,
+      "loss": 0.5698,
+      "step": 2969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980296120739583e-05,
+      "loss": 0.6018,
+      "step": 2970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802764277344334e-05,
+      "loss": 0.5822,
+      "step": 2971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802567249911463e-05,
+      "loss": 0.5804,
+      "step": 2972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802370125099177e-05,
+      "loss": 0.6229,
+      "step": 2973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802172902909436e-05,
+      "loss": 0.6336,
+      "step": 2974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98019755833442e-05,
+      "loss": 0.5844,
+      "step": 2975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801778166405427e-05,
+      "loss": 0.5487,
+      "step": 2976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801580652095076e-05,
+      "loss": 0.6934,
+      "step": 2977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801383040415113e-05,
+      "loss": 0.6462,
+      "step": 2978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98011853313675e-05,
+      "loss": 0.644,
+      "step": 2979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800987524954202e-05,
+      "loss": 0.6132,
+      "step": 2980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800789621177184e-05,
+      "loss": 0.5586,
+      "step": 2981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800591620038404e-05,
+      "loss": 0.5671,
+      "step": 2982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800393521539844e-05,
+      "loss": 0.5463,
+      "step": 2983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980019532568346e-05,
+      "loss": 0.5522,
+      "step": 2984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799997032471223e-05,
+      "loss": 0.6393,
+      "step": 2985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979979864190511e-05,
+      "loss": 0.5782,
+      "step": 2986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979960015398708e-05,
+      "loss": 0.5831,
+      "step": 2987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799401568719115e-05,
+      "loss": 0.5659,
+      "step": 2988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799202886103183e-05,
+      "loss": 0.5597,
+      "step": 2989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799004106141257e-05,
+      "loss": 0.5696,
+      "step": 2990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798805228835313e-05,
+      "loss": 0.5727,
+      "step": 2991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979860625418733e-05,
+      "loss": 0.5653,
+      "step": 2992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979840718219928e-05,
+      "loss": 0.5948,
+      "step": 2993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798208012873144e-05,
+      "loss": 0.6195,
+      "step": 2994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798008746210894e-05,
+      "loss": 0.6108,
+      "step": 2995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979780938221452e-05,
+      "loss": 0.5589,
+      "step": 2996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797609920885993e-05,
+      "loss": 0.6532,
+      "step": 2997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.97974103622273e-05,
+      "loss": 0.6448,
+      "step": 2998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979721070624042e-05,
+      "loss": 0.6249,
+      "step": 2999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979701095292734e-05,
+      "loss": 0.5401,
+      "step": 3000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979681110229004e-05,
+      "loss": 0.6464,
+      "step": 3001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979661115433051e-05,
+      "loss": 0.542,
+      "step": 3002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796411109050734e-05,
+      "loss": 0.4942,
+      "step": 3003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796210966452696e-05,
+      "loss": 0.6225,
+      "step": 3004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796010726538393e-05,
+      "loss": 0.6139,
+      "step": 3005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795810389309805e-05,
+      "loss": 0.6914,
+      "step": 3006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979560995476893e-05,
+      "loss": 0.7158,
+      "step": 3007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795409422917748e-05,
+      "loss": 0.6191,
+      "step": 3008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795208793758263e-05,
+      "loss": 0.6143,
+      "step": 3009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979500806729246e-05,
+      "loss": 0.5615,
+      "step": 3010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979480724352234e-05,
+      "loss": 0.5714,
+      "step": 3011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794606322449893e-05,
+      "loss": 0.6687,
+      "step": 3012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794405304077115e-05,
+      "loss": 0.6518,
+      "step": 3013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794204188406005e-05,
+      "loss": 0.6936,
+      "step": 3014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979400297543856e-05,
+      "loss": 0.7256,
+      "step": 3015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979380166517678e-05,
+      "loss": 0.5459,
+      "step": 3016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979360025762266e-05,
+      "loss": 0.5681,
+      "step": 3017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979339875277821e-05,
+      "loss": 0.5941,
+      "step": 3018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793197150645424e-05,
+      "loss": 0.5324,
+      "step": 3019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9792995451226304e-05,
+      "loss": 0.5449,
+      "step": 3020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979279365452286e-05,
+      "loss": 0.6177,
+      "step": 3021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9792591760537093e-05,
+      "loss": 0.5686,
+      "step": 3022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792389769271007e-05,
+      "loss": 0.5092,
+      "step": 3023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979218768072661e-05,
+      "loss": 0.5272,
+      "step": 3024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791985494905913e-05,
+      "loss": 0.5574,
+      "step": 3025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979178321181092e-05,
+      "loss": 0.5547,
+      "step": 3026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791580831443645e-05,
+      "loss": 0.5819,
+      "step": 3027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791378353806094e-05,
+      "loss": 0.5496,
+      "step": 3028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979117577890028e-05,
+      "loss": 0.5495,
+      "step": 3029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979097310672821e-05,
+      "loss": 0.5701,
+      "step": 3030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790770337291913e-05,
+      "loss": 0.5412,
+      "step": 3031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790567470593385e-05,
+      "loss": 0.6615,
+      "step": 3032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790364506634656e-05,
+      "loss": 0.6396,
+      "step": 3033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979016144541773e-05,
+      "loss": 0.6393,
+      "step": 3034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789958286944632e-05,
+      "loss": 0.6015,
+      "step": 3035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978975503121738e-05,
+      "loss": 0.6575,
+      "step": 3036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978955167823799e-05,
+      "loss": 0.5573,
+      "step": 3037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978934822800848e-05,
+      "loss": 0.6959,
+      "step": 3038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978914468053088e-05,
+      "loss": 0.5952,
+      "step": 3039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788941035807205e-05,
+      "loss": 0.5494,
+      "step": 3040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788737293839477e-05,
+      "loss": 0.5663,
+      "step": 3041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788533454629727e-05,
+      "loss": 0.5109,
+      "step": 3042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978832951817997e-05,
+      "loss": 0.556,
+      "step": 3043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978812548449224e-05,
+      "loss": 0.4983,
+      "step": 3044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978792135356856e-05,
+      "loss": 0.5419,
+      "step": 3045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978771712541096e-05,
+      "loss": 0.5551,
+      "step": 3046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978751280002147e-05,
+      "loss": 0.6351,
+      "step": 3047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787308377402118e-05,
+      "loss": 0.6128,
+      "step": 3048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978710385755493e-05,
+      "loss": 0.5932,
+      "step": 3049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786899240481947e-05,
+      "loss": 0.5869,
+      "step": 3050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786694526185193e-05,
+      "loss": 0.662,
+      "step": 3051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978648971466671e-05,
+      "loss": 0.5854,
+      "step": 3052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786284805928526e-05,
+      "loss": 0.6036,
+      "step": 3053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786079799972677e-05,
+      "loss": 0.5705,
+      "step": 3054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97858746968012e-05,
+      "loss": 0.5706,
+      "step": 3055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978566949641614e-05,
+      "loss": 0.5895,
+      "step": 3056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785464198819524e-05,
+      "loss": 0.6042,
+      "step": 3057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785258804013398e-05,
+      "loss": 0.6311,
+      "step": 3058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97850533119998e-05,
+      "loss": 0.5976,
+      "step": 3059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784847722780775e-05,
+      "loss": 0.5135,
+      "step": 3060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784642036358363e-05,
+      "loss": 0.6004,
+      "step": 3061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784436252734606e-05,
+      "loss": 0.7,
+      "step": 3062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784230371911552e-05,
+      "loss": 0.6616,
+      "step": 3063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978402439389124e-05,
+      "loss": 0.5856,
+      "step": 3064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978381831867572e-05,
+      "loss": 0.5785,
+      "step": 3065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783612146267036e-05,
+      "loss": 0.5185,
+      "step": 3066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783405876667247e-05,
+      "loss": 0.582,
+      "step": 3067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978319950987839e-05,
+      "loss": 0.5772,
+      "step": 3068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978299304590252e-05,
+      "loss": 0.5596,
+      "step": 3069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782786484741683e-05,
+      "loss": 0.5167,
+      "step": 3070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782579826397942e-05,
+      "loss": 0.5343,
+      "step": 3071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978237307087334e-05,
+      "loss": 0.6415,
+      "step": 3072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782166218169933e-05,
+      "loss": 0.6252,
+      "step": 3073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978195926828978e-05,
+      "loss": 0.5087,
+      "step": 3074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781752221234932e-05,
+      "loss": 0.6418,
+      "step": 3075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781545077007448e-05,
+      "loss": 0.5154,
+      "step": 3076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781337835609388e-05,
+      "loss": 0.595,
+      "step": 3077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781130497042804e-05,
+      "loss": 0.5236,
+      "step": 3078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780923061309765e-05,
+      "loss": 0.5228,
+      "step": 3079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780715528412325e-05,
+      "loss": 0.6304,
+      "step": 3080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780507898352543e-05,
+      "loss": 0.5613,
+      "step": 3081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978030017113249e-05,
+      "loss": 0.5425,
+      "step": 3082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780092346754227e-05,
+      "loss": 0.6446,
+      "step": 3083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779884425219813e-05,
+      "loss": 0.572,
+      "step": 3084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.977967640653132e-05,
+      "loss": 0.5638,
+      "step": 3085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779468290690813e-05,
+      "loss": 0.5401,
+      "step": 3086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779260077700354e-05,
+      "loss": 0.5764,
+      "step": 3087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977905176756202e-05,
+      "loss": 0.5292,
+      "step": 3088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778843360277877e-05,
+      "loss": 0.5752,
+      "step": 3089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977863485584999e-05,
+      "loss": 0.5957,
+      "step": 3090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977842625428044e-05,
+      "loss": 0.6404,
+      "step": 3091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977821755557129e-05,
+      "loss": 0.5373,
+      "step": 3092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977800875972462e-05,
+      "loss": 0.5506,
+      "step": 3093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777799866742502e-05,
+      "loss": 0.5787,
+      "step": 3094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777590876627012e-05,
+      "loss": 0.5151,
+      "step": 3095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977738178938022e-05,
+      "loss": 0.5291,
+      "step": 3096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977717260500421e-05,
+      "loss": 0.6006,
+      "step": 3097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977696332350106e-05,
+      "loss": 0.4991,
+      "step": 3098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776753944872844e-05,
+      "loss": 0.5808,
+      "step": 3099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776544469121646e-05,
+      "loss": 0.598,
+      "step": 3100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776334896249548e-05,
+      "loss": 0.5796,
+      "step": 3101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776125226258628e-05,
+      "loss": 0.6084,
+      "step": 3102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775915459150973e-05,
+      "loss": 0.5187,
+      "step": 3103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977570559492866e-05,
+      "loss": 0.5082,
+      "step": 3104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977549563359378e-05,
+      "loss": 0.4345,
+      "step": 3105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775285575148418e-05,
+      "loss": 0.621,
+      "step": 3106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977507541959466e-05,
+      "loss": 0.6882,
+      "step": 3107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774865166934595e-05,
+      "loss": 0.4762,
+      "step": 3108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977465481717031e-05,
+      "loss": 0.4844,
+      "step": 3109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774444370303894e-05,
+      "loss": 0.521,
+      "step": 3110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977423382633744e-05,
+      "loss": 0.5863,
+      "step": 3111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774023185273035e-05,
+      "loss": 0.6208,
+      "step": 3112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773812447112775e-05,
+      "loss": 0.6175,
+      "step": 3113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773601611858755e-05,
+      "loss": 0.5549,
+      "step": 3114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773390679513068e-05,
+      "loss": 0.5812,
+      "step": 3115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773179650077805e-05,
+      "loss": 0.4653,
+      "step": 3116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772968523555067e-05,
+      "loss": 0.5548,
+      "step": 3117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772757299946954e-05,
+      "loss": 0.6918,
+      "step": 3118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977254597925556e-05,
+      "loss": 0.5176,
+      "step": 3119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772334561482986e-05,
+      "loss": 0.5522,
+      "step": 3120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977212304663133e-05,
+      "loss": 0.5969,
+      "step": 3121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771911434702692e-05,
+      "loss": 0.582,
+      "step": 3122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977169972569918e-05,
+      "loss": 0.5247,
+      "step": 3123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771487919622895e-05,
+      "loss": 0.5138,
+      "step": 3124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977127601647594e-05,
+      "loss": 0.5958,
+      "step": 3125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771064016260416e-05,
+      "loss": 0.5769,
+      "step": 3126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977085191897844e-05,
+      "loss": 0.6147,
+      "step": 3127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770639724632102e-05,
+      "loss": 0.666,
+      "step": 3128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977042743322353e-05,
+      "loss": 0.5176,
+      "step": 3129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977021504475482e-05,
+      "loss": 0.5243,
+      "step": 3130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770002559228083e-05,
+      "loss": 0.5419,
+      "step": 3131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769789976645432e-05,
+      "loss": 0.6403,
+      "step": 3132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976957729700898e-05,
+      "loss": 0.5155,
+      "step": 3133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769364520320836e-05,
+      "loss": 0.6521,
+      "step": 3134
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769151646583122e-05,
+      "loss": 0.583,
+      "step": 3135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768938675797943e-05,
+      "loss": 0.423,
+      "step": 3136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976872560796742e-05,
+      "loss": 0.539,
+      "step": 3137
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768512443093663e-05,
+      "loss": 0.5626,
+      "step": 3138
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768299181178798e-05,
+      "loss": 0.5537,
+      "step": 3139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976808582222494e-05,
+      "loss": 0.5429,
+      "step": 3140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767872366234208e-05,
+      "loss": 0.533,
+      "step": 3141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767658813208725e-05,
+      "loss": 0.5532,
+      "step": 3142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976744516315061e-05,
+      "loss": 0.5296,
+      "step": 3143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767231416061988e-05,
+      "loss": 0.5956,
+      "step": 3144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767017571944976e-05,
+      "loss": 0.5438,
+      "step": 3145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976680363080171e-05,
+      "loss": 0.6184,
+      "step": 3146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97665895926343e-05,
+      "loss": 0.5922,
+      "step": 3147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766375457444886e-05,
+      "loss": 0.5601,
+      "step": 3148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766161225235585e-05,
+      "loss": 0.5753,
+      "step": 3149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9765946896008534e-05,
+      "loss": 0.5688,
+      "step": 3150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976573246976586e-05,
+      "loss": 0.6167,
+      "step": 3151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9765517946509685e-05,
+      "loss": 0.5718,
+      "step": 3152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976530332624215e-05,
+      "loss": 0.5214,
+      "step": 3153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976508860896538e-05,
+      "loss": 0.5198,
+      "step": 3154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764873794681514e-05,
+      "loss": 0.5816,
+      "step": 3155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764658883392683e-05,
+      "loss": 0.552,
+      "step": 3156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976444387510102e-05,
+      "loss": 0.5818,
+      "step": 3157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764228769808665e-05,
+      "loss": 0.6268,
+      "step": 3158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764013567517755e-05,
+      "loss": 0.5349,
+      "step": 3159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976379826823042e-05,
+      "loss": 0.6479,
+      "step": 3160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763582871948813e-05,
+      "loss": 0.6166,
+      "step": 3161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763367378675062e-05,
+      "loss": 0.5231,
+      "step": 3162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976315178841131e-05,
+      "loss": 0.5361,
+      "step": 3163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97629361011597e-05,
+      "loss": 0.6037,
+      "step": 3164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762720316922372e-05,
+      "loss": 0.5833,
+      "step": 3165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762504435701476e-05,
+      "loss": 0.5779,
+      "step": 3166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976228845749915e-05,
+      "loss": 0.644,
+      "step": 3167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762072382317545e-05,
+      "loss": 0.5945,
+      "step": 3168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761856210158802e-05,
+      "loss": 0.5604,
+      "step": 3169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761639941025074e-05,
+      "loss": 0.5826,
+      "step": 3170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761423574918503e-05,
+      "loss": 0.5505,
+      "step": 3171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761207111841247e-05,
+      "loss": 0.5536,
+      "step": 3172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760990551795448e-05,
+      "loss": 0.5688,
+      "step": 3173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976077389478326e-05,
+      "loss": 0.611,
+      "step": 3174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760557140806838e-05,
+      "loss": 0.5138,
+      "step": 3175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976034028986833e-05,
+      "loss": 0.549,
+      "step": 3176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760123341969898e-05,
+      "loss": 0.5553,
+      "step": 3177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975990629711369e-05,
+      "loss": 0.6121,
+      "step": 3178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759689155301864e-05,
+      "loss": 0.5973,
+      "step": 3179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759471916536578e-05,
+      "loss": 0.6832,
+      "step": 3180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759254580819992e-05,
+      "loss": 0.4993,
+      "step": 3181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975903714815426e-05,
+      "loss": 0.6371,
+      "step": 3182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975881961854155e-05,
+      "loss": 0.7232,
+      "step": 3183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758601991984012e-05,
+      "loss": 0.5172,
+      "step": 3184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758384268483816e-05,
+      "loss": 0.6419,
+      "step": 3185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758166448043123e-05,
+      "loss": 0.5692,
+      "step": 3186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757948530664096e-05,
+      "loss": 0.5138,
+      "step": 3187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757730516348904e-05,
+      "loss": 0.5807,
+      "step": 3188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757512405099705e-05,
+      "loss": 0.6,
+      "step": 3189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757294196918672e-05,
+      "loss": 0.6401,
+      "step": 3190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757075891807972e-05,
+      "loss": 0.5915,
+      "step": 3191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756857489769767e-05,
+      "loss": 0.6115,
+      "step": 3192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756638990806233e-05,
+      "loss": 0.5776,
+      "step": 3193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756420394919544e-05,
+      "loss": 0.5649,
+      "step": 3194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756201702111863e-05,
+      "loss": 0.479,
+      "step": 3195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755982912385367e-05,
+      "loss": 0.5617,
+      "step": 3196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975576402574223e-05,
+      "loss": 0.5431,
+      "step": 3197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755545042184627e-05,
+      "loss": 0.5587,
+      "step": 3198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755325961714726e-05,
+      "loss": 0.5432,
+      "step": 3199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755106784334714e-05,
+      "loss": 0.5764,
+      "step": 3200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975488751004676e-05,
+      "loss": 0.5929,
+      "step": 3201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975466813885305e-05,
+      "loss": 0.6508,
+      "step": 3202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754448670755758e-05,
+      "loss": 0.6552,
+      "step": 3203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754229105757065e-05,
+      "loss": 0.5698,
+      "step": 3204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975400944385915e-05,
+      "loss": 0.5975,
+      "step": 3205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97537896850642e-05,
+      "loss": 0.586,
+      "step": 3206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753569829374396e-05,
+      "loss": 0.5917,
+      "step": 3207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753349876791922e-05,
+      "loss": 0.686,
+      "step": 3208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975312982731896e-05,
+      "loss": 0.5837,
+      "step": 3209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752909680957698e-05,
+      "loss": 0.5512,
+      "step": 3210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752689437710325e-05,
+      "loss": 0.4942,
+      "step": 3211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975246909757903e-05,
+      "loss": 0.5874,
+      "step": 3212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752248660565998e-05,
+      "loss": 0.5702,
+      "step": 3213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752028126673418e-05,
+      "loss": 0.6562,
+      "step": 3214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9751807495903484e-05,
+      "loss": 0.4721,
+      "step": 3215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975158676825839e-05,
+      "loss": 0.6217,
+      "step": 3216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9751365943740325e-05,
+      "loss": 0.5821,
+      "step": 3217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975114502235148e-05,
+      "loss": 0.6019,
+      "step": 3218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750924004094056e-05,
+      "loss": 0.606,
+      "step": 3219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750702888970242e-05,
+      "loss": 0.6084,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975048167698224e-05,
+      "loss": 0.5831,
+      "step": 3221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750260368132245e-05,
+      "loss": 0.6922,
+      "step": 3222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750038962422456e-05,
+      "loss": 0.494,
+      "step": 3223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749817459855073e-05,
+      "loss": 0.5225,
+      "step": 3224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749595860432298e-05,
+      "loss": 0.5992,
+      "step": 3225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974937416415633e-05,
+      "loss": 0.5722,
+      "step": 3226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749152371029368e-05,
+      "loss": 0.5564,
+      "step": 3227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748930481053625e-05,
+      "loss": 0.5749,
+      "step": 3228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748708494231295e-05,
+      "loss": 0.5902,
+      "step": 3229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974848641056459e-05,
+      "loss": 0.58,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974826423005571e-05,
+      "loss": 0.4877,
+      "step": 3231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974804195270687e-05,
+      "loss": 0.6251,
+      "step": 3232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974781957852027e-05,
+      "loss": 0.5788,
+      "step": 3233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747597107498126e-05,
+      "loss": 0.6328,
+      "step": 3234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747374539642643e-05,
+      "loss": 0.5545,
+      "step": 3235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747151874956038e-05,
+      "loss": 0.5036,
+      "step": 3236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746929113440515e-05,
+      "loss": 0.5204,
+      "step": 3237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746706255098292e-05,
+      "loss": 0.6036,
+      "step": 3238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746483299931583e-05,
+      "loss": 0.6018,
+      "step": 3239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746260247942605e-05,
+      "loss": 0.653,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746037099133563e-05,
+      "loss": 0.5166,
+      "step": 3241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745813853506687e-05,
+      "loss": 0.6769,
+      "step": 3242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974559051106419e-05,
+      "loss": 0.6593,
+      "step": 3243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745367071808284e-05,
+      "loss": 0.5895,
+      "step": 3244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97451435357412e-05,
+      "loss": 0.5449,
+      "step": 3245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974491990286515e-05,
+      "loss": 0.6224,
+      "step": 3246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744696173182357e-05,
+      "loss": 0.828,
+      "step": 3247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744472346695048e-05,
+      "loss": 0.5722,
+      "step": 3248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744248423405446e-05,
+      "loss": 0.5434,
+      "step": 3249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744024403315773e-05,
+      "loss": 0.6257,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974380028642825e-05,
+      "loss": 0.5407,
+      "step": 3251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974357607274511e-05,
+      "loss": 0.5866,
+      "step": 3252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974335176226858e-05,
+      "loss": 0.6419,
+      "step": 3253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743127355000884e-05,
+      "loss": 0.5027,
+      "step": 3254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742902850944257e-05,
+      "loss": 0.4842,
+      "step": 3255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742678250100924e-05,
+      "loss": 0.5406,
+      "step": 3256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974245355247312e-05,
+      "loss": 0.5688,
+      "step": 3257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742228758063072e-05,
+      "loss": 0.6539,
+      "step": 3258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974200386687302e-05,
+      "loss": 0.565,
+      "step": 3259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741778878905196e-05,
+      "loss": 0.5989,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974155379416183e-05,
+      "loss": 0.5138,
+      "step": 3261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741328612645168e-05,
+      "loss": 0.5886,
+      "step": 3262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741103334357438e-05,
+      "loss": 0.6501,
+      "step": 3263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740877959300876e-05,
+      "loss": 0.5919,
+      "step": 3264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974065248747773e-05,
+      "loss": 0.479,
+      "step": 3265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740426918890235e-05,
+      "loss": 0.4704,
+      "step": 3266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974020125354063e-05,
+      "loss": 0.6306,
+      "step": 3267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973997549143116e-05,
+      "loss": 0.5214,
+      "step": 3268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973974963256407e-05,
+      "loss": 0.5675,
+      "step": 3269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97395236769416e-05,
+      "loss": 0.6397,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739297624565992e-05,
+      "loss": 0.6572,
+      "step": 3271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739071475439498e-05,
+      "loss": 0.6285,
+      "step": 3272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973884522956436e-05,
+      "loss": 0.5819,
+      "step": 3273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738618886942826e-05,
+      "loss": 0.5851,
+      "step": 3274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738392447577148e-05,
+      "loss": 0.5293,
+      "step": 3275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973816591146957e-05,
+      "loss": 0.7058,
+      "step": 3276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973793927862235e-05,
+      "loss": 0.6078,
+      "step": 3277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737712549037732e-05,
+      "loss": 0.5626,
+      "step": 3278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737485722717968e-05,
+      "loss": 0.6597,
+      "step": 3279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737258799665316e-05,
+      "loss": 0.5555,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973703177988203e-05,
+      "loss": 0.5367,
+      "step": 3281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973680466337036e-05,
+      "loss": 0.597,
+      "step": 3282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736577450132574e-05,
+      "loss": 0.5752,
+      "step": 3283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736350140170915e-05,
+      "loss": 0.573,
+      "step": 3284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736122733487648e-05,
+      "loss": 0.6015,
+      "step": 3285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735895230085033e-05,
+      "loss": 0.6229,
+      "step": 3286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735667629965326e-05,
+      "loss": 0.6529,
+      "step": 3287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735439933130794e-05,
+      "loss": 0.5343,
+      "step": 3288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735212139583694e-05,
+      "loss": 0.5845,
+      "step": 3289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973498424932629e-05,
+      "loss": 0.515,
+      "step": 3290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734756262360845e-05,
+      "loss": 0.5933,
+      "step": 3291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734528178689625e-05,
+      "loss": 0.6293,
+      "step": 3292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734299998314898e-05,
+      "loss": 0.5194,
+      "step": 3293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973407172123893e-05,
+      "loss": 0.5752,
+      "step": 3294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733843347463982e-05,
+      "loss": 0.5957,
+      "step": 3295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733614876992332e-05,
+      "loss": 0.5521,
+      "step": 3296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733386309826246e-05,
+      "loss": 0.6531,
+      "step": 3297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973315764596799e-05,
+      "loss": 0.5928,
+      "step": 3298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732928885419845e-05,
+      "loss": 0.5651,
+      "step": 3299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732700028184074e-05,
+      "loss": 0.5575,
+      "step": 3300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732471074262958e-05,
+      "loss": 0.5753,
+      "step": 3301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732242023658767e-05,
+      "loss": 0.6506,
+      "step": 3302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732012876373775e-05,
+      "loss": 0.5102,
+      "step": 3303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731783632410265e-05,
+      "loss": 0.5994,
+      "step": 3304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973155429177051e-05,
+      "loss": 0.5636,
+      "step": 3305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731324854456785e-05,
+      "loss": 0.5818,
+      "step": 3306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731095320471376e-05,
+      "loss": 0.6041,
+      "step": 3307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973086568981656e-05,
+      "loss": 0.5845,
+      "step": 3308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730635962494617e-05,
+      "loss": 0.683,
+      "step": 3309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730406138507834e-05,
+      "loss": 0.568,
+      "step": 3310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730176217858487e-05,
+      "loss": 0.5644,
+      "step": 3311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729946200548865e-05,
+      "loss": 0.4901,
+      "step": 3312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972971608658125e-05,
+      "loss": 0.5826,
+      "step": 3313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972948587595793e-05,
+      "loss": 0.5315,
+      "step": 3314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972925556868119e-05,
+      "loss": 0.6304,
+      "step": 3315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729025164753325e-05,
+      "loss": 0.6402,
+      "step": 3316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728794664176613e-05,
+      "loss": 0.5744,
+      "step": 3317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972856406695335e-05,
+      "loss": 0.5567,
+      "step": 3318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728333373085825e-05,
+      "loss": 0.6206,
+      "step": 3319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728102582576333e-05,
+      "loss": 0.6557,
+      "step": 3320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972787169542716e-05,
+      "loss": 0.5455,
+      "step": 3321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972764071164061e-05,
+      "loss": 0.6152,
+      "step": 3322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727409631218966e-05,
+      "loss": 0.5292,
+      "step": 3323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972717845416453e-05,
+      "loss": 0.5822,
+      "step": 3324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726947180479598e-05,
+      "loss": 0.577,
+      "step": 3325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726715810166468e-05,
+      "loss": 0.591,
+      "step": 3326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726484343227438e-05,
+      "loss": 0.6249,
+      "step": 3327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726252779664805e-05,
+      "loss": 0.558,
+      "step": 3328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726021119480873e-05,
+      "loss": 0.6201,
+      "step": 3329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972578936267794e-05,
+      "loss": 0.5884,
+      "step": 3330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725557509258307e-05,
+      "loss": 0.6252,
+      "step": 3331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725325559224285e-05,
+      "loss": 0.6406,
+      "step": 3332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725093512578172e-05,
+      "loss": 0.6193,
+      "step": 3333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724861369322274e-05,
+      "loss": 0.6665,
+      "step": 3334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97246291294589e-05,
+      "loss": 0.5323,
+      "step": 3335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972439679299035e-05,
+      "loss": 0.6741,
+      "step": 3336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972416435991894e-05,
+      "loss": 0.4818,
+      "step": 3337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723931830246978e-05,
+      "loss": 0.6306,
+      "step": 3338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723699203976768e-05,
+      "loss": 0.5255,
+      "step": 3339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723466481110623e-05,
+      "loss": 0.574,
+      "step": 3340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723233661650864e-05,
+      "loss": 0.5151,
+      "step": 3341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972300074559979e-05,
+      "loss": 0.5339,
+      "step": 3342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722767732959724e-05,
+      "loss": 0.6234,
+      "step": 3343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722534623732974e-05,
+      "loss": 0.5983,
+      "step": 3344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722301417921864e-05,
+      "loss": 0.5888,
+      "step": 3345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722068115528702e-05,
+      "loss": 0.6628,
+      "step": 3346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721834716555814e-05,
+      "loss": 0.5331,
+      "step": 3347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721601221005513e-05,
+      "loss": 0.5396,
+      "step": 3348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972136762888012e-05,
+      "loss": 0.6372,
+      "step": 3349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721133940181957e-05,
+      "loss": 0.556,
+      "step": 3350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720900154913344e-05,
+      "loss": 0.5324,
+      "step": 3351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97206662730766e-05,
+      "loss": 0.5634,
+      "step": 3352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720432294674058e-05,
+      "loss": 0.5547,
+      "step": 3353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972019821970803e-05,
+      "loss": 0.5763,
+      "step": 3354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719964048180853e-05,
+      "loss": 0.573,
+      "step": 3355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719729780094843e-05,
+      "loss": 0.5212,
+      "step": 3356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719495415452337e-05,
+      "loss": 0.514,
+      "step": 3357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719260954255656e-05,
+      "loss": 0.5537,
+      "step": 3358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719026396507134e-05,
+      "loss": 0.5515,
+      "step": 3359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718791742209097e-05,
+      "loss": 0.5537,
+      "step": 3360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718556991363876e-05,
+      "loss": 0.6177,
+      "step": 3361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718322143973806e-05,
+      "loss": 0.6372,
+      "step": 3362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971808720004122e-05,
+      "loss": 0.5864,
+      "step": 3363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971785215956845e-05,
+      "loss": 0.6143,
+      "step": 3364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717617022557828e-05,
+      "loss": 0.5281,
+      "step": 3365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97173817890117e-05,
+      "loss": 0.543,
+      "step": 3366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717146458932392e-05,
+      "loss": 0.5786,
+      "step": 3367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716911032322246e-05,
+      "loss": 0.5596,
+      "step": 3368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716675509183606e-05,
+      "loss": 0.5348,
+      "step": 3369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97164398895188e-05,
+      "loss": 0.5319,
+      "step": 3370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971620417333018e-05,
+      "loss": 0.5824,
+      "step": 3371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715968360620077e-05,
+      "loss": 0.522,
+      "step": 3372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715732451390846e-05,
+      "loss": 0.6909,
+      "step": 3373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971549644564482e-05,
+      "loss": 0.6086,
+      "step": 3374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971526034338435e-05,
+      "loss": 0.6456,
+      "step": 3375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715024144611776e-05,
+      "loss": 0.5247,
+      "step": 3376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714787849329448e-05,
+      "loss": 0.5055,
+      "step": 3377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714551457539716e-05,
+      "loss": 0.5878,
+      "step": 3378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714314969244918e-05,
+      "loss": 0.4807,
+      "step": 3379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714078384447416e-05,
+      "loss": 0.4924,
+      "step": 3380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971384170314955e-05,
+      "loss": 0.5057,
+      "step": 3381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971360492535368e-05,
+      "loss": 0.6296,
+      "step": 3382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713368051062157e-05,
+      "loss": 0.5304,
+      "step": 3383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713131080277325e-05,
+      "loss": 0.5298,
+      "step": 3384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971289401300155e-05,
+      "loss": 0.5095,
+      "step": 3385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712656849237172e-05,
+      "loss": 0.6332,
+      "step": 3386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712419588986562e-05,
+      "loss": 0.55,
+      "step": 3387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712182232252072e-05,
+      "loss": 0.5929,
+      "step": 3388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971194477903606e-05,
+      "loss": 0.6425,
+      "step": 3389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971170722934088e-05,
+      "loss": 0.5707,
+      "step": 3390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711469583168897e-05,
+      "loss": 0.6098,
+      "step": 3391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711231840522473e-05,
+      "loss": 0.6056,
+      "step": 3392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710994001403966e-05,
+      "loss": 0.576,
+      "step": 3393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710756065815738e-05,
+      "loss": 0.5819,
+      "step": 3394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710518033760158e-05,
+      "loss": 0.4415,
+      "step": 3395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710279905239584e-05,
+      "loss": 0.5996,
+      "step": 3396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710041680256386e-05,
+      "loss": 0.6417,
+      "step": 3397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970980335881293e-05,
+      "loss": 0.5197,
+      "step": 3398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709564940911584e-05,
+      "loss": 0.5942,
+      "step": 3399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709326426554715e-05,
+      "loss": 0.5603,
+      "step": 3400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970908781574469e-05,
+      "loss": 0.5722,
+      "step": 3401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708849108483888e-05,
+      "loss": 0.53,
+      "step": 3402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708610304774672e-05,
+      "loss": 0.5306,
+      "step": 3403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708371404619412e-05,
+      "loss": 0.6129,
+      "step": 3404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708132408020493e-05,
+      "loss": 0.6363,
+      "step": 3405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707893314980276e-05,
+      "loss": 0.5667,
+      "step": 3406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707654125501147e-05,
+      "loss": 0.5723,
+      "step": 3407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970741483958548e-05,
+      "loss": 0.6573,
+      "step": 3408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707175457235644e-05,
+      "loss": 0.6622,
+      "step": 3409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9706935978454024e-05,
+      "loss": 0.6348,
+      "step": 3410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9706696403243e-05,
+      "loss": 0.5916,
+      "step": 3411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706456731604947e-05,
+      "loss": 0.6013,
+      "step": 3412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706216963542253e-05,
+      "loss": 0.6108,
+      "step": 3413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705977099057292e-05,
+      "loss": 0.6019,
+      "step": 3414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970573713815245e-05,
+      "loss": 0.606,
+      "step": 3415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705497080830113e-05,
+      "loss": 0.552,
+      "step": 3416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705256927092664e-05,
+      "loss": 0.5002,
+      "step": 3417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705016676942486e-05,
+      "loss": 0.5474,
+      "step": 3418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970477633038197e-05,
+      "loss": 0.5453,
+      "step": 3419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97045358874135e-05,
+      "loss": 0.5462,
+      "step": 3420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704295348039472e-05,
+      "loss": 0.5692,
+      "step": 3421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704054712262268e-05,
+      "loss": 0.585,
+      "step": 3422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703813980084274e-05,
+      "loss": 0.5524,
+      "step": 3423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703573151507894e-05,
+      "loss": 0.6101,
+      "step": 3424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703332226535515e-05,
+      "loss": 0.5224,
+      "step": 3425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703091205169523e-05,
+      "loss": 0.5953,
+      "step": 3426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702850087412327e-05,
+      "loss": 0.5986,
+      "step": 3427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970260887326631e-05,
+      "loss": 0.5887,
+      "step": 3428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702367562733873e-05,
+      "loss": 0.561,
+      "step": 3429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702126155817413e-05,
+      "loss": 0.615,
+      "step": 3430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701884652519327e-05,
+      "loss": 0.6165,
+      "step": 3431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701643052842017e-05,
+      "loss": 0.5433,
+      "step": 3432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970140135678788e-05,
+      "loss": 0.5408,
+      "step": 3433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701159564359318e-05,
+      "loss": 0.5918,
+      "step": 3434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700917675558733e-05,
+      "loss": 0.5697,
+      "step": 3435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970067569038853e-05,
+      "loss": 0.5719,
+      "step": 3436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700433608851107e-05,
+      "loss": 0.6435,
+      "step": 3437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700191430948873e-05,
+      "loss": 0.5679,
+      "step": 3438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699949156684238e-05,
+      "loss": 0.4724,
+      "step": 3439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96997067860596e-05,
+      "loss": 0.6103,
+      "step": 3440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699464319077374e-05,
+      "loss": 0.5924,
+      "step": 3441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699221755739963e-05,
+      "loss": 0.5767,
+      "step": 3442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969897909604978e-05,
+      "loss": 0.6306,
+      "step": 3443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698736340009237e-05,
+      "loss": 0.5194,
+      "step": 3444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698493487620742e-05,
+      "loss": 0.5451,
+      "step": 3445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698250538886712e-05,
+      "loss": 0.5955,
+      "step": 3446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698007493809555e-05,
+      "loss": 0.5232,
+      "step": 3447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697764352391687e-05,
+      "loss": 0.5972,
+      "step": 3448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697521114635525e-05,
+      "loss": 0.527,
+      "step": 3449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697277780543486e-05,
+      "loss": 0.6717,
+      "step": 3450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697034350117983e-05,
+      "loss": 0.6119,
+      "step": 3451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969679082336144e-05,
+      "loss": 0.6105,
+      "step": 3452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696547200276274e-05,
+      "loss": 0.6029,
+      "step": 3453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96963034808649e-05,
+      "loss": 0.6345,
+      "step": 3454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969605966512975e-05,
+      "loss": 0.5519,
+      "step": 3455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695815753073236e-05,
+      "loss": 0.5259,
+      "step": 3456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695571744697787e-05,
+      "loss": 0.5839,
+      "step": 3457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969532764000582e-05,
+      "loss": 0.5757,
+      "step": 3458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695083438999772e-05,
+      "loss": 0.5612,
+      "step": 3459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694839141682056e-05,
+      "loss": 0.6268,
+      "step": 3460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969459474805511e-05,
+      "loss": 0.6077,
+      "step": 3461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969435025812135e-05,
+      "loss": 0.6055,
+      "step": 3462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694105671883216e-05,
+      "loss": 0.5366,
+      "step": 3463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969386098934313e-05,
+      "loss": 0.6635,
+      "step": 3464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693616210503528e-05,
+      "loss": 0.5176,
+      "step": 3465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969337133536684e-05,
+      "loss": 0.6132,
+      "step": 3466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693126363935496e-05,
+      "loss": 0.6727,
+      "step": 3467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692881296211932e-05,
+      "loss": 0.5592,
+      "step": 3468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969263613219858e-05,
+      "loss": 0.6849,
+      "step": 3469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692390871897878e-05,
+      "loss": 0.5973,
+      "step": 3470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692145515312267e-05,
+      "loss": 0.5427,
+      "step": 3471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691900062444174e-05,
+      "loss": 0.574,
+      "step": 3472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691654513296044e-05,
+      "loss": 0.6644,
+      "step": 3473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691408867870315e-05,
+      "loss": 0.5157,
+      "step": 3474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691163126169428e-05,
+      "loss": 0.5628,
+      "step": 3475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969091728819582e-05,
+      "loss": 0.5442,
+      "step": 3476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.969067135395194e-05,
+      "loss": 0.6574,
+      "step": 3477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9690425323440228e-05,
+      "loss": 0.54,
+      "step": 3478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9690179196663126e-05,
+      "loss": 0.6024,
+      "step": 3479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689932973623082e-05,
+      "loss": 0.5299,
+      "step": 3480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968968665432254e-05,
+      "loss": 0.4931,
+      "step": 3481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968944023876395e-05,
+      "loss": 0.6569,
+      "step": 3482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689193726949752e-05,
+      "loss": 0.5541,
+      "step": 3483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688947118882408e-05,
+      "loss": 0.5498,
+      "step": 3484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688700414564355e-05,
+      "loss": 0.6414,
+      "step": 3485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968845361399805e-05,
+      "loss": 0.5279,
+      "step": 3486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688206717185946e-05,
+      "loss": 0.625,
+      "step": 3487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687959724130492e-05,
+      "loss": 0.5043,
+      "step": 3488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687712634834146e-05,
+      "loss": 0.6219,
+      "step": 3489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687465449299355e-05,
+      "loss": 0.5515,
+      "step": 3490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687218167528585e-05,
+      "loss": 0.6181,
+      "step": 3491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968697078952428e-05,
+      "loss": 0.5073,
+      "step": 3492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686723315288908e-05,
+      "loss": 0.5745,
+      "step": 3493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686475744824928e-05,
+      "loss": 0.4661,
+      "step": 3494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968622807813479e-05,
+      "loss": 0.539,
+      "step": 3495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968598031522096e-05,
+      "loss": 0.4819,
+      "step": 3496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685732456085898e-05,
+      "loss": 0.6332,
+      "step": 3497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968548450073207e-05,
+      "loss": 0.6374,
+      "step": 3498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685236449161932e-05,
+      "loss": 0.5716,
+      "step": 3499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684988301377957e-05,
+      "loss": 0.5033,
+      "step": 3500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684740057382604e-05,
+      "loss": 0.575,
+      "step": 3501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968449171717834e-05,
+      "loss": 0.6911,
+      "step": 3502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684243280767635e-05,
+      "loss": 0.5767,
+      "step": 3503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968399474815295e-05,
+      "loss": 0.5783,
+      "step": 3504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683746119336763e-05,
+      "loss": 0.626,
+      "step": 3505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968349739432154e-05,
+      "loss": 0.5809,
+      "step": 3506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968324857310975e-05,
+      "loss": 0.5425,
+      "step": 3507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682999655703866e-05,
+      "loss": 0.5881,
+      "step": 3508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682750642106357e-05,
+      "loss": 0.6026,
+      "step": 3509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682501532319705e-05,
+      "loss": 0.5275,
+      "step": 3510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682252326346377e-05,
+      "loss": 0.5175,
+      "step": 3511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682003024188856e-05,
+      "loss": 0.5537,
+      "step": 3512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681753625849612e-05,
+      "loss": 0.5712,
+      "step": 3513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968150413133113e-05,
+      "loss": 0.5801,
+      "step": 3514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681254540635877e-05,
+      "loss": 0.6046,
+      "step": 3515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681004853766342e-05,
+      "loss": 0.5861,
+      "step": 3516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680755070724998e-05,
+      "loss": 0.5357,
+      "step": 3517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968050519151434e-05,
+      "loss": 0.5085,
+      "step": 3518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680255216136834e-05,
+      "loss": 0.6721,
+      "step": 3519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968000514459497e-05,
+      "loss": 0.6492,
+      "step": 3520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679754976891238e-05,
+      "loss": 0.5397,
+      "step": 3521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967950471302811e-05,
+      "loss": 0.527,
+      "step": 3522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679254353008086e-05,
+      "loss": 0.5863,
+      "step": 3523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679003896833647e-05,
+      "loss": 0.5754,
+      "step": 3524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967875334450728e-05,
+      "loss": 0.5946,
+      "step": 3525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678502696031473e-05,
+      "loss": 0.5592,
+      "step": 3526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967825195140872e-05,
+      "loss": 0.5663,
+      "step": 3527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967800111064151e-05,
+      "loss": 0.5205,
+      "step": 3528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677750173732336e-05,
+      "loss": 0.6043,
+      "step": 3529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967749914068369e-05,
+      "loss": 0.5475,
+      "step": 3530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677248011498062e-05,
+      "loss": 0.6005,
+      "step": 3531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967699678617795e-05,
+      "loss": 0.5888,
+      "step": 3532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676745464725854e-05,
+      "loss": 0.5778,
+      "step": 3533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676494047144262e-05,
+      "loss": 0.5537,
+      "step": 3534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967624253343568e-05,
+      "loss": 0.5659,
+      "step": 3535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96759909236026e-05,
+      "loss": 0.5665,
+      "step": 3536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675739217647523e-05,
+      "loss": 0.6234,
+      "step": 3537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675487415572957e-05,
+      "loss": 0.547,
+      "step": 3538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675235517381392e-05,
+      "loss": 0.5377,
+      "step": 3539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9674983523075334e-05,
+      "loss": 0.6262,
+      "step": 3540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967473143265729e-05,
+      "loss": 0.5741,
+      "step": 3541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674479246129762e-05,
+      "loss": 0.5932,
+      "step": 3542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674226963495254e-05,
+      "loss": 0.5479,
+      "step": 3543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673974584756274e-05,
+      "loss": 0.5315,
+      "step": 3544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673722109915333e-05,
+      "loss": 0.5841,
+      "step": 3545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967346953897493e-05,
+      "loss": 0.546,
+      "step": 3546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967321687193758e-05,
+      "loss": 0.5487,
+      "step": 3547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967296410880579e-05,
+      "loss": 0.5101,
+      "step": 3548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967271124958208e-05,
+      "loss": 0.6113,
+      "step": 3549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9672458294268946e-05,
+      "loss": 0.5668,
+      "step": 3550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967220524286891e-05,
+      "loss": 0.5928,
+      "step": 3551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967195209538449e-05,
+      "loss": 0.5585,
+      "step": 3552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671698851818193e-05,
+      "loss": 0.5083,
+      "step": 3553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967144551217254e-05,
+      "loss": 0.5382,
+      "step": 3554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671192076450047e-05,
+      "loss": 0.5777,
+      "step": 3555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670938544653226e-05,
+      "loss": 0.6521,
+      "step": 3556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96706849167846e-05,
+      "loss": 0.5973,
+      "step": 3557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670431192846694e-05,
+      "loss": 0.5936,
+      "step": 3558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670177372842017e-05,
+      "loss": 0.583,
+      "step": 3559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96699234567731e-05,
+      "loss": 0.5577,
+      "step": 3560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966966944464246e-05,
+      "loss": 0.6286,
+      "step": 3561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9669415336452623e-05,
+      "loss": 0.5999,
+      "step": 3562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966916113220611e-05,
+      "loss": 0.5687,
+      "step": 3563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668906831905454e-05,
+      "loss": 0.65,
+      "step": 3564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668652435553173e-05,
+      "loss": 0.5575,
+      "step": 3565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668397943151798e-05,
+      "loss": 0.5628,
+      "step": 3566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668143354703856e-05,
+      "loss": 0.5764,
+      "step": 3567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667888670211878e-05,
+      "loss": 0.6501,
+      "step": 3568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667633889678388e-05,
+      "loss": 0.5488,
+      "step": 3569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667379013105924e-05,
+      "loss": 0.5224,
+      "step": 3570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667124040497017e-05,
+      "loss": 0.6222,
+      "step": 3571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9666868971854195e-05,
+      "loss": 0.6285,
+      "step": 3572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966661380718e-05,
+      "loss": 0.5512,
+      "step": 3573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966635854647696e-05,
+      "loss": 0.577,
+      "step": 3574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966610318974761e-05,
+      "loss": 0.5482,
+      "step": 3575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665847736994493e-05,
+      "loss": 0.6751,
+      "step": 3576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966559218822014e-05,
+      "loss": 0.5341,
+      "step": 3577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96653365434271e-05,
+      "loss": 0.5106,
+      "step": 3578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665080802617898e-05,
+      "loss": 0.588,
+      "step": 3579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966482496579509e-05,
+      "loss": 0.5275,
+      "step": 3580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664569032961203e-05,
+      "loss": 0.5277,
+      "step": 3581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664313004118785e-05,
+      "loss": 0.599,
+      "step": 3582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664056879270384e-05,
+      "loss": 0.6086,
+      "step": 3583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966380065841854e-05,
+      "loss": 0.5285,
+      "step": 3584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96635443415658e-05,
+      "loss": 0.4933,
+      "step": 3585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966328792871471e-05,
+      "loss": 0.5615,
+      "step": 3586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663031419867814e-05,
+      "loss": 0.6374,
+      "step": 3587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662774815027665e-05,
+      "loss": 0.635,
+      "step": 3588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662518114196812e-05,
+      "loss": 0.5464,
+      "step": 3589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662261317377797e-05,
+      "loss": 0.6798,
+      "step": 3590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966200442457318e-05,
+      "loss": 0.6213,
+      "step": 3591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661747435785512e-05,
+      "loss": 0.5418,
+      "step": 3592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661490351017342e-05,
+      "loss": 0.5878,
+      "step": 3593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661233170271225e-05,
+      "loss": 0.4877,
+      "step": 3594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660975893549722e-05,
+      "loss": 0.6118,
+      "step": 3595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660718520855378e-05,
+      "loss": 0.6079,
+      "step": 3596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660461052190754e-05,
+      "loss": 0.5786,
+      "step": 3597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660203487558412e-05,
+      "loss": 0.5224,
+      "step": 3598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965994582696091e-05,
+      "loss": 0.6342,
+      "step": 3599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659688070400803e-05,
+      "loss": 0.5906,
+      "step": 3600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659430217880654e-05,
+      "loss": 0.5605,
+      "step": 3601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659172269403023e-05,
+      "loss": 0.62,
+      "step": 3602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658914224970474e-05,
+      "loss": 0.4917,
+      "step": 3603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658656084585576e-05,
+      "loss": 0.5806,
+      "step": 3604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658397848250882e-05,
+      "loss": 0.449,
+      "step": 3605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658139515968965e-05,
+      "loss": 0.6383,
+      "step": 3606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657881087742386e-05,
+      "loss": 0.5242,
+      "step": 3607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965762256357372e-05,
+      "loss": 0.5686,
+      "step": 3608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965736394346553e-05,
+      "loss": 0.5078,
+      "step": 3609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657105227420383e-05,
+      "loss": 0.6086,
+      "step": 3610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656846415440855e-05,
+      "loss": 0.5102,
+      "step": 3611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656587507529513e-05,
+      "loss": 0.5504,
+      "step": 3612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965632850368893e-05,
+      "loss": 0.5108,
+      "step": 3613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965606940392168e-05,
+      "loss": 0.4926,
+      "step": 3614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655810208230334e-05,
+      "loss": 0.5584,
+      "step": 3615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655550916617468e-05,
+      "loss": 0.6505,
+      "step": 3616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655291529085666e-05,
+      "loss": 0.6231,
+      "step": 3617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965503204563749e-05,
+      "loss": 0.606,
+      "step": 3618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654772466275527e-05,
+      "loss": 0.6006,
+      "step": 3619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654512791002354e-05,
+      "loss": 0.6089,
+      "step": 3620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965425301982055e-05,
+      "loss": 0.5124,
+      "step": 3621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96539931527327e-05,
+      "loss": 0.5638,
+      "step": 3622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965373318974138e-05,
+      "loss": 0.6529,
+      "step": 3623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965347313084917e-05,
+      "loss": 0.6401,
+      "step": 3624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653212976058664e-05,
+      "loss": 0.5158,
+      "step": 3625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652952725372435e-05,
+      "loss": 0.5207,
+      "step": 3626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652692378793074e-05,
+      "loss": 0.6304,
+      "step": 3627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652431936323166e-05,
+      "loss": 0.6886,
+      "step": 3628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96521713979653e-05,
+      "loss": 0.5172,
+      "step": 3629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651910763722064e-05,
+      "loss": 0.6204,
+      "step": 3630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651650033596044e-05,
+      "loss": 0.5929,
+      "step": 3631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651389207589835e-05,
+      "loss": 0.5557,
+      "step": 3632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965112828570602e-05,
+      "loss": 0.6239,
+      "step": 3633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650867267947206e-05,
+      "loss": 0.6027,
+      "step": 3634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650606154315967e-05,
+      "loss": 0.5511,
+      "step": 3635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650344944814913e-05,
+      "loss": 0.6079,
+      "step": 3636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650083639446626e-05,
+      "loss": 0.607,
+      "step": 3637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649822238213715e-05,
+      "loss": 0.6371,
+      "step": 3638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964956074111876e-05,
+      "loss": 0.4993,
+      "step": 3639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649299148164378e-05,
+      "loss": 0.6566,
+      "step": 3640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649037459353156e-05,
+      "loss": 0.6196,
+      "step": 3641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648775674687694e-05,
+      "loss": 0.5861,
+      "step": 3642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648513794170596e-05,
+      "loss": 0.5814,
+      "step": 3643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964825181780446e-05,
+      "loss": 0.5892,
+      "step": 3644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964798974559189e-05,
+      "loss": 0.5986,
+      "step": 3645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647727577535492e-05,
+      "loss": 0.5815,
+      "step": 3646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647465313637868e-05,
+      "loss": 0.4992,
+      "step": 3647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647202953901626e-05,
+      "loss": 0.5395,
+      "step": 3648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646940498329367e-05,
+      "loss": 0.5521,
+      "step": 3649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96466779469237e-05,
+      "loss": 0.6115,
+      "step": 3650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646415299687236e-05,
+      "loss": 0.6102,
+      "step": 3651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646152556622586e-05,
+      "loss": 0.6401,
+      "step": 3652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964588971773235e-05,
+      "loss": 0.5492,
+      "step": 3653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964562678301915e-05,
+      "loss": 0.6099,
+      "step": 3654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645363752485594e-05,
+      "loss": 0.5762,
+      "step": 3655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964510062613429e-05,
+      "loss": 0.562,
+      "step": 3656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964483740396786e-05,
+      "loss": 0.5495,
+      "step": 3657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644574085988915e-05,
+      "loss": 0.5283,
+      "step": 3658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644310672200073e-05,
+      "loss": 0.5809,
+      "step": 3659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644047162603948e-05,
+      "loss": 0.5618,
+      "step": 3660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964378355720316e-05,
+      "loss": 0.6562,
+      "step": 3661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643519856000327e-05,
+      "loss": 0.4924,
+      "step": 3662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643256058998062e-05,
+      "loss": 0.709,
+      "step": 3663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642992166198997e-05,
+      "loss": 0.5106,
+      "step": 3664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642728177605747e-05,
+      "loss": 0.5709,
+      "step": 3665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642464093220938e-05,
+      "loss": 0.5158,
+      "step": 3666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964219991304719e-05,
+      "loss": 0.6873,
+      "step": 3667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641935637087125e-05,
+      "loss": 0.5577,
+      "step": 3668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641671265343377e-05,
+      "loss": 0.529,
+      "step": 3669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641406797818562e-05,
+      "loss": 0.5666,
+      "step": 3670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641142234515317e-05,
+      "loss": 0.5709,
+      "step": 3671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640877575436265e-05,
+      "loss": 0.6158,
+      "step": 3672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.964061282058403e-05,
+      "loss": 0.4869,
+      "step": 3673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640347969961254e-05,
+      "loss": 0.5174,
+      "step": 3674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.964008302357056e-05,
+      "loss": 0.6652,
+      "step": 3675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639817981414583e-05,
+      "loss": 0.5822,
+      "step": 3676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963955284349595e-05,
+      "loss": 0.5635,
+      "step": 3677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639287609817306e-05,
+      "loss": 0.6065,
+      "step": 3678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639022280381278e-05,
+      "loss": 0.5767,
+      "step": 3679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.96387568551905e-05,
+      "loss": 0.4242,
+      "step": 3680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638491334247614e-05,
+      "loss": 0.5472,
+      "step": 3681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638225717555257e-05,
+      "loss": 0.6195,
+      "step": 3682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637960005116067e-05,
+      "loss": 0.5931,
+      "step": 3683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963769419693268e-05,
+      "loss": 0.6086,
+      "step": 3684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637428293007744e-05,
+      "loss": 0.6895,
+      "step": 3685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637162293343893e-05,
+      "loss": 0.6236,
+      "step": 3686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963689619794377e-05,
+      "loss": 0.6537,
+      "step": 3687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636630006810026e-05,
+      "loss": 0.5243,
+      "step": 3688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.96363637199453e-05,
+      "loss": 0.5546,
+      "step": 3689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636097337352232e-05,
+      "loss": 0.5233,
+      "step": 3690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963583085903348e-05,
+      "loss": 0.5747,
+      "step": 3691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635564284991683e-05,
+      "loss": 0.626,
+      "step": 3692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635297615229493e-05,
+      "loss": 0.5051,
+      "step": 3693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635030849749554e-05,
+      "loss": 0.6133,
+      "step": 3694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963476398855452e-05,
+      "loss": 0.4837,
+      "step": 3695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634497031647046e-05,
+      "loss": 0.5526,
+      "step": 3696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634229979029776e-05,
+      "loss": 0.4739,
+      "step": 3697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633962830705365e-05,
+      "loss": 0.552,
+      "step": 3698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963369558667647e-05,
+      "loss": 0.5771,
+      "step": 3699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633428246945745e-05,
+      "loss": 0.5434,
+      "step": 3700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633160811515847e-05,
+      "loss": 0.5852,
+      "step": 3701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632893280389427e-05,
+      "loss": 0.5883,
+      "step": 3702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632625653569146e-05,
+      "loss": 0.5167,
+      "step": 3703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632357931057665e-05,
+      "loss": 0.6298,
+      "step": 3704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963209011285764e-05,
+      "loss": 0.6401,
+      "step": 3705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631822198971733e-05,
+      "loss": 0.5455,
+      "step": 3706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963155418940261e-05,
+      "loss": 0.559,
+      "step": 3707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631286084152924e-05,
+      "loss": 0.4803,
+      "step": 3708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631017883225342e-05,
+      "loss": 0.6205,
+      "step": 3709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963074958662253e-05,
+      "loss": 0.5253,
+      "step": 3710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963048119434716e-05,
+      "loss": 0.5488,
+      "step": 3711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630212706401883e-05,
+      "loss": 0.5479,
+      "step": 3712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629944122789376e-05,
+      "loss": 0.4672,
+      "step": 3713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629675443512307e-05,
+      "loss": 0.5304,
+      "step": 3714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629406668573346e-05,
+      "loss": 0.592,
+      "step": 3715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629137797975156e-05,
+      "loss": 0.6696,
+      "step": 3716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628868831720415e-05,
+      "loss": 0.5185,
+      "step": 3717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962859976981179e-05,
+      "loss": 0.6425,
+      "step": 3718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962833061225196e-05,
+      "loss": 0.5273,
+      "step": 3719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962806135904359e-05,
+      "loss": 0.4563,
+      "step": 3720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627792010189366e-05,
+      "loss": 0.4866,
+      "step": 3721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627522565691955e-05,
+      "loss": 0.5328,
+      "step": 3722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627253025554034e-05,
+      "loss": 0.6654,
+      "step": 3723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626983389778288e-05,
+      "loss": 0.6705,
+      "step": 3724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626713658367385e-05,
+      "loss": 0.618,
+      "step": 3725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626443831324013e-05,
+      "loss": 0.596,
+      "step": 3726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626173908650848e-05,
+      "loss": 0.5835,
+      "step": 3727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625903890350578e-05,
+      "loss": 0.5948,
+      "step": 3728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625633776425874e-05,
+      "loss": 0.633,
+      "step": 3729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625363566879428e-05,
+      "loss": 0.671,
+      "step": 3730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625093261713923e-05,
+      "loss": 0.6258,
+      "step": 3731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624822860932045e-05,
+      "loss": 0.5886,
+      "step": 3732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624552364536472e-05,
+      "loss": 0.4867,
+      "step": 3733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624281772529904e-05,
+      "loss": 0.6056,
+      "step": 3734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962401108491502e-05,
+      "loss": 0.4919,
+      "step": 3735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962374030169451e-05,
+      "loss": 0.6031,
+      "step": 3736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962346942287107e-05,
+      "loss": 0.6523,
+      "step": 3737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623198448447384e-05,
+      "loss": 0.5757,
+      "step": 3738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622927378426147e-05,
+      "loss": 0.5494,
+      "step": 3739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962265621281005e-05,
+      "loss": 0.5842,
+      "step": 3740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962238495160179e-05,
+      "loss": 0.4821,
+      "step": 3741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962211359480406e-05,
+      "loss": 0.6195,
+      "step": 3742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962184214241955e-05,
+      "loss": 0.4721,
+      "step": 3743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621570594450968e-05,
+      "loss": 0.5667,
+      "step": 3744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621298950901005e-05,
+      "loss": 0.6212,
+      "step": 3745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962102721177236e-05,
+      "loss": 0.5778,
+      "step": 3746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620755377067732e-05,
+      "loss": 0.5269,
+      "step": 3747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620483446789822e-05,
+      "loss": 0.5323,
+      "step": 3748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962021142094133e-05,
+      "loss": 0.5603,
+      "step": 3749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961993929952496e-05,
+      "loss": 0.5695,
+      "step": 3750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961966708254342e-05,
+      "loss": 0.6137,
+      "step": 3751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.96193947699994e-05,
+      "loss": 0.6077,
+      "step": 3752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619122361895623e-05,
+      "loss": 0.6493,
+      "step": 3753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618849858234785e-05,
+      "loss": 0.4856,
+      "step": 3754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618577259019595e-05,
+      "loss": 0.4705,
+      "step": 3755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618304564252755e-05,
+      "loss": 0.577,
+      "step": 3756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618031773936984e-05,
+      "loss": 0.6753,
+      "step": 3757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617758888074986e-05,
+      "loss": 0.5594,
+      "step": 3758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617485906669476e-05,
+      "loss": 0.5764,
+      "step": 3759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617212829723164e-05,
+      "loss": 0.6202,
+      "step": 3760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616939657238758e-05,
+      "loss": 0.5059,
+      "step": 3761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961666638921898e-05,
+      "loss": 0.5156,
+      "step": 3762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616393025666537e-05,
+      "loss": 0.5805,
+      "step": 3763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961611956658415e-05,
+      "loss": 0.573,
+      "step": 3764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961584601197453e-05,
+      "loss": 0.5307,
+      "step": 3765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.96155723618404e-05,
+      "loss": 0.5769,
+      "step": 3766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9615298616184478e-05,
+      "loss": 0.6774,
+      "step": 3767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9615024775009482e-05,
+      "loss": 0.6622,
+      "step": 3768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614750838318134e-05,
+      "loss": 0.6706,
+      "step": 3769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614476806113146e-05,
+      "loss": 0.531,
+      "step": 3770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614202678397258e-05,
+      "loss": 0.6185,
+      "step": 3771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613928455173177e-05,
+      "loss": 0.5862,
+      "step": 3772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961365413644363e-05,
+      "loss": 0.5926,
+      "step": 3773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613379722211353e-05,
+      "loss": 0.5891,
+      "step": 3774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961310521247906e-05,
+      "loss": 0.4987,
+      "step": 3775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9612830607249485e-05,
+      "loss": 0.571,
+      "step": 3776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961255590652535e-05,
+      "loss": 0.5856,
+      "step": 3777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961228111030939e-05,
+      "loss": 0.6275,
+      "step": 3778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961200621860433e-05,
+      "loss": 0.6591,
+      "step": 3779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611731231412903e-05,
+      "loss": 0.6797,
+      "step": 3780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611456148737844e-05,
+      "loss": 0.5063,
+      "step": 3781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961118097058188e-05,
+      "loss": 0.5807,
+      "step": 3782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610905696947748e-05,
+      "loss": 0.6146,
+      "step": 3783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610630327838177e-05,
+      "loss": 0.5428,
+      "step": 3784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610354863255914e-05,
+      "loss": 0.5861,
+      "step": 3785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610079303203683e-05,
+      "loss": 0.5923,
+      "step": 3786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960980364768423e-05,
+      "loss": 0.4849,
+      "step": 3787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960952789670029e-05,
+      "loss": 0.6059,
+      "step": 3788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609252050254598e-05,
+      "loss": 0.6414,
+      "step": 3789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608976108349906e-05,
+      "loss": 0.564,
+      "step": 3790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608700070988944e-05,
+      "loss": 0.5243,
+      "step": 3791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608423938174458e-05,
+      "loss": 0.6464,
+      "step": 3792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608147709909194e-05,
+      "loss": 0.614,
+      "step": 3793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607871386195897e-05,
+      "loss": 0.5692,
+      "step": 3794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.96075949670373e-05,
+      "loss": 0.6072,
+      "step": 3795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607318452436164e-05,
+      "loss": 0.4739,
+      "step": 3796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607041842395228e-05,
+      "loss": 0.5534,
+      "step": 3797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960676513691724e-05,
+      "loss": 0.5348,
+      "step": 3798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960648833600495e-05,
+      "loss": 0.5515,
+      "step": 3799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960621143966111e-05,
+      "loss": 0.5717,
+      "step": 3800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9605934447888465e-05,
+      "loss": 0.6609,
+      "step": 3801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605657360689773e-05,
+      "loss": 0.7027,
+      "step": 3802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605380178067785e-05,
+      "loss": 0.6238,
+      "step": 3803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960510290002525e-05,
+      "loss": 0.5359,
+      "step": 3804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960482552656493e-05,
+      "loss": 0.574,
+      "step": 3805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604548057689574e-05,
+      "loss": 0.5001,
+      "step": 3806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960427049340194e-05,
+      "loss": 0.5535,
+      "step": 3807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603992833704788e-05,
+      "loss": 0.516,
+      "step": 3808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603715078600874e-05,
+      "loss": 0.5489,
+      "step": 3809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603437228092956e-05,
+      "loss": 0.559,
+      "step": 3810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603159282183795e-05,
+      "loss": 0.5821,
+      "step": 3811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602881240876156e-05,
+      "loss": 0.6068,
+      "step": 3812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.96026031041728e-05,
+      "loss": 0.6626,
+      "step": 3813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602324872076485e-05,
+      "loss": 0.6067,
+      "step": 3814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960204654458998e-05,
+      "loss": 0.56,
+      "step": 3815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601768121716044e-05,
+      "loss": 0.558,
+      "step": 3816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960148960345745e-05,
+      "loss": 0.6189,
+      "step": 3817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960121098981696e-05,
+      "loss": 0.6396,
+      "step": 3818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600932280797346e-05,
+      "loss": 0.6096,
+      "step": 3819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600653476401374e-05,
+      "loss": 0.5371,
+      "step": 3820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600374576631813e-05,
+      "loss": 0.5678,
+      "step": 3821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960009558149143e-05,
+      "loss": 0.5602,
+      "step": 3822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599816490983006e-05,
+      "loss": 0.6536,
+      "step": 3823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599537305109306e-05,
+      "loss": 0.5486,
+      "step": 3824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599258023873108e-05,
+      "loss": 0.5691,
+      "step": 3825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598978647277184e-05,
+      "loss": 0.561,
+      "step": 3826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959869917532431e-05,
+      "loss": 0.5435,
+      "step": 3827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598419608017262e-05,
+      "loss": 0.4682,
+      "step": 3828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598139945358817e-05,
+      "loss": 0.5486,
+      "step": 3829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597860187351753e-05,
+      "loss": 0.5289,
+      "step": 3830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597580333998846e-05,
+      "loss": 0.505,
+      "step": 3831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597300385302883e-05,
+      "loss": 0.5889,
+      "step": 3832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597020341266644e-05,
+      "loss": 0.5532,
+      "step": 3833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596740201892906e-05,
+      "loss": 0.6218,
+      "step": 3834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596459967184458e-05,
+      "loss": 0.5494,
+      "step": 3835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596179637144078e-05,
+      "loss": 0.5707,
+      "step": 3836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959589921177455e-05,
+      "loss": 0.4744,
+      "step": 3837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595618691078666e-05,
+      "loss": 0.5518,
+      "step": 3838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959533807505921e-05,
+      "loss": 0.6059,
+      "step": 3839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959505736371897e-05,
+      "loss": 0.6637,
+      "step": 3840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594776557060734e-05,
+      "loss": 0.5531,
+      "step": 3841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594495655087292e-05,
+      "loss": 0.5025,
+      "step": 3842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959421465780143e-05,
+      "loss": 0.5976,
+      "step": 3843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959393356520595e-05,
+      "loss": 0.5616,
+      "step": 3844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959365237730363e-05,
+      "loss": 0.5537,
+      "step": 3845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593371094097278e-05,
+      "loss": 0.4967,
+      "step": 3846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593089715589677e-05,
+      "loss": 0.7238,
+      "step": 3847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592808241783632e-05,
+      "loss": 0.5203,
+      "step": 3848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592526672681932e-05,
+      "loss": 0.5605,
+      "step": 3849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592245008287372e-05,
+      "loss": 0.6207,
+      "step": 3850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591963248602757e-05,
+      "loss": 0.5429,
+      "step": 3851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591681393630883e-05,
+      "loss": 0.5064,
+      "step": 3852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959139944337455e-05,
+      "loss": 0.5354,
+      "step": 3853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591117397836562e-05,
+      "loss": 0.5512,
+      "step": 3854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590835257019715e-05,
+      "loss": 0.6436,
+      "step": 3855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590553020926816e-05,
+      "loss": 0.5352,
+      "step": 3856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590270689560665e-05,
+      "loss": 0.5427,
+      "step": 3857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589988262924073e-05,
+      "loss": 0.5262,
+      "step": 3858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958970574101984e-05,
+      "loss": 0.5764,
+      "step": 3859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958942312385077e-05,
+      "loss": 0.5898,
+      "step": 3860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589140411419682e-05,
+      "loss": 0.5698,
+      "step": 3861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588857603729378e-05,
+      "loss": 0.5887,
+      "step": 3862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958857470078266e-05,
+      "loss": 0.386,
+      "step": 3863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958829170258235e-05,
+      "loss": 0.5984,
+      "step": 3864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588008609131253e-05,
+      "loss": 0.6077,
+      "step": 3865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9587725420432183e-05,
+      "loss": 0.4908,
+      "step": 3866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958744213648795e-05,
+      "loss": 0.5224,
+      "step": 3867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9587158757301376e-05,
+      "loss": 0.5587,
+      "step": 3868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958687528287527e-05,
+      "loss": 0.4577,
+      "step": 3869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586591713212453e-05,
+      "loss": 0.6416,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586308048315733e-05,
+      "loss": 0.4919,
+      "step": 3871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586024288187935e-05,
+      "loss": 0.5458,
+      "step": 3872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958574043283188e-05,
+      "loss": 0.6043,
+      "step": 3873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958545648225038e-05,
+      "loss": 0.5344,
+      "step": 3874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958517243644626e-05,
+      "loss": 0.5341,
+      "step": 3875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958488829542234e-05,
+      "loss": 0.5633,
+      "step": 3876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584604059181446e-05,
+      "loss": 0.5609,
+      "step": 3877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95843197277264e-05,
+      "loss": 0.51,
+      "step": 3878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584035301060026e-05,
+      "loss": 0.5973,
+      "step": 3879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958375077918515e-05,
+      "loss": 0.6968,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95834661621046e-05,
+      "loss": 0.5959,
+      "step": 3881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95831814498212e-05,
+      "loss": 0.608,
+      "step": 3882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582896642337784e-05,
+      "loss": 0.6937,
+      "step": 3883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582611739657174e-05,
+      "loss": 0.5235,
+      "step": 3884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582326741782204e-05,
+      "loss": 0.5474,
+      "step": 3885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582041648715706e-05,
+      "loss": 0.5856,
+      "step": 3886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958175646046051e-05,
+      "loss": 0.6237,
+      "step": 3887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958147117701945e-05,
+      "loss": 0.5659,
+      "step": 3888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958118579839536e-05,
+      "loss": 0.5426,
+      "step": 3889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958090032459108e-05,
+      "loss": 0.5903,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580614755609434e-05,
+      "loss": 0.5939,
+      "step": 3891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958032909145327e-05,
+      "loss": 0.6186,
+      "step": 3892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958004333212542e-05,
+      "loss": 0.5837,
+      "step": 3893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579757477628724e-05,
+      "loss": 0.6092,
+      "step": 3894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579471527966026e-05,
+      "loss": 0.6157,
+      "step": 3895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957918548314016e-05,
+      "loss": 0.5673,
+      "step": 3896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957889934315397e-05,
+      "loss": 0.609,
+      "step": 3897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95786131080103e-05,
+      "loss": 0.542,
+      "step": 3898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957832677771199e-05,
+      "loss": 0.5047,
+      "step": 3899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957804035226189e-05,
+      "loss": 0.5698,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577753831662846e-05,
+      "loss": 0.6378,
+      "step": 3901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577467215917694e-05,
+      "loss": 0.6019,
+      "step": 3902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577180505029288e-05,
+      "loss": 0.6039,
+      "step": 3903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957689369900048e-05,
+      "loss": 0.5938,
+      "step": 3904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576606797834114e-05,
+      "loss": 0.6713,
+      "step": 3905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576319801533037e-05,
+      "loss": 0.5593,
+      "step": 3906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576032710100108e-05,
+      "loss": 0.5389,
+      "step": 3907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575745523538176e-05,
+      "loss": 0.5623,
+      "step": 3908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575458241850093e-05,
+      "loss": 0.5616,
+      "step": 3909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957517086503871e-05,
+      "loss": 0.6419,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9574883393106888e-05,
+      "loss": 0.5981,
+      "step": 3911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9574595826057478e-05,
+      "loss": 0.5333,
+      "step": 3912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9574308163893342e-05,
+      "loss": 0.5376,
+      "step": 3913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957402040661733e-05,
+      "loss": 0.5745,
+      "step": 3914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957373255423231e-05,
+      "loss": 0.6796,
+      "step": 3915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573444606741127e-05,
+      "loss": 0.64,
+      "step": 3916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573156564146657e-05,
+      "loss": 0.6108,
+      "step": 3917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572868426451754e-05,
+      "loss": 0.5819,
+      "step": 3918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572580193659282e-05,
+      "loss": 0.659,
+      "step": 3919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572291865772105e-05,
+      "loss": 0.6192,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572003442793083e-05,
+      "loss": 0.6129,
+      "step": 3921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957171492472509e-05,
+      "loss": 0.5798,
+      "step": 3922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957142631157098e-05,
+      "loss": 0.7045,
+      "step": 3923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571137603333627e-05,
+      "loss": 0.5645,
+      "step": 3924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570848800015904e-05,
+      "loss": 0.5474,
+      "step": 3925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570559901620673e-05,
+      "loss": 0.6062,
+      "step": 3926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570270908150802e-05,
+      "loss": 0.4856,
+      "step": 3927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569981819609167e-05,
+      "loss": 0.6185,
+      "step": 3928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569692635998637e-05,
+      "loss": 0.4982,
+      "step": 3929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 0.5431,
+      "step": 3930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569113983582394e-05,
+      "loss": 0.5866,
+      "step": 3931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568824514782424e-05,
+      "loss": 0.5884,
+      "step": 3932
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956853495092506e-05,
+      "loss": 0.5744,
+      "step": 3933
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568245292013172e-05,
+      "loss": 0.6191,
+      "step": 3934
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567955538049643e-05,
+      "loss": 0.5494,
+      "step": 3935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956766568903735e-05,
+      "loss": 0.5224,
+      "step": 3936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567375744979172e-05,
+      "loss": 0.5145,
+      "step": 3937
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956708570587799e-05,
+      "loss": 0.5546,
+      "step": 3938
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566795571736682e-05,
+      "loss": 0.587,
+      "step": 3939
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566505342558132e-05,
+      "loss": 0.5705,
+      "step": 3940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566215018345226e-05,
+      "loss": 0.4984,
+      "step": 3941
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565924599100848e-05,
+      "loss": 0.5398,
+      "step": 3942
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565634084827877e-05,
+      "loss": 0.5945,
+      "step": 3943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565343475529207e-05,
+      "loss": 0.6025,
+      "step": 3944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956505277120772e-05,
+      "loss": 0.5145,
+      "step": 3945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95647619718663e-05,
+      "loss": 0.4594,
+      "step": 3946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564471077507848e-05,
+      "loss": 0.5876,
+      "step": 3947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564180088135242e-05,
+      "loss": 0.5411,
+      "step": 3948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563889003751378e-05,
+      "loss": 0.6121,
+      "step": 3949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563597824359144e-05,
+      "loss": 0.5627,
+      "step": 3950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956330654996144e-05,
+      "loss": 0.6146,
+      "step": 3951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563015180561153e-05,
+      "loss": 0.5416,
+      "step": 3952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562723716161178e-05,
+      "loss": 0.5113,
+      "step": 3953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956243215676441e-05,
+      "loss": 0.6666,
+      "step": 3954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562140502373753e-05,
+      "loss": 0.5913,
+      "step": 3955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561848752992098e-05,
+      "loss": 0.5465,
+      "step": 3956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561556908622338e-05,
+      "loss": 0.5836,
+      "step": 3957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561264969267384e-05,
+      "loss": 0.5356,
+      "step": 3958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560972934930123e-05,
+      "loss": 0.542,
+      "step": 3959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560680805613472e-05,
+      "loss": 0.6052,
+      "step": 3960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560388581320318e-05,
+      "loss": 0.5977,
+      "step": 3961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560096262053574e-05,
+      "loss": 0.6572,
+      "step": 3962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559803847816138e-05,
+      "loss": 0.6295,
+      "step": 3963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559511338610916e-05,
+      "loss": 0.6524,
+      "step": 3964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559218734440818e-05,
+      "loss": 0.5888,
+      "step": 3965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558926035308746e-05,
+      "loss": 0.6559,
+      "step": 3966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955863324121761e-05,
+      "loss": 0.6521,
+      "step": 3967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558340352170314e-05,
+      "loss": 0.5373,
+      "step": 3968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558047368169776e-05,
+      "loss": 0.5068,
+      "step": 3969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95577542892189e-05,
+      "loss": 0.5332,
+      "step": 3970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557461115320602e-05,
+      "loss": 0.5943,
+      "step": 3971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557167846477788e-05,
+      "loss": 0.5458,
+      "step": 3972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556874482693378e-05,
+      "loss": 0.525,
+      "step": 3973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556581023970283e-05,
+      "loss": 0.484,
+      "step": 3974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556287470311418e-05,
+      "loss": 0.5879,
+      "step": 3975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95559938217197e-05,
+      "loss": 0.5975,
+      "step": 3976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555700078198048e-05,
+      "loss": 0.5268,
+      "step": 3977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555406239749375e-05,
+      "loss": 0.5789,
+      "step": 3978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555112306376606e-05,
+      "loss": 0.5087,
+      "step": 3979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554818278082655e-05,
+      "loss": 0.5723,
+      "step": 3980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554524154870447e-05,
+      "loss": 0.5732,
+      "step": 3981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554229936742907e-05,
+      "loss": 0.5456,
+      "step": 3982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553935623702948e-05,
+      "loss": 0.5808,
+      "step": 3983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553641215753502e-05,
+      "loss": 0.5568,
+      "step": 3984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955334671289749e-05,
+      "loss": 0.5795,
+      "step": 3985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553052115137842e-05,
+      "loss": 0.6339,
+      "step": 3986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955275742247748e-05,
+      "loss": 0.5036,
+      "step": 3987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552462634919328e-05,
+      "loss": 0.6472,
+      "step": 3988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552167752466325e-05,
+      "loss": 0.6394,
+      "step": 3989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551872775121393e-05,
+      "loss": 0.5804,
+      "step": 3990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955157770288746e-05,
+      "loss": 0.5846,
+      "step": 3991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551282535767466e-05,
+      "loss": 0.5462,
+      "step": 3992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550987273764336e-05,
+      "loss": 0.5673,
+      "step": 3993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550691916881006e-05,
+      "loss": 0.4462,
+      "step": 3994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955039646512041e-05,
+      "loss": 0.5191,
+      "step": 3995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550100918485483e-05,
+      "loss": 0.6183,
+      "step": 3996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549805276979158e-05,
+      "loss": 0.6003,
+      "step": 3997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549509540604377e-05,
+      "loss": 0.4643,
+      "step": 3998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954921370936407e-05,
+      "loss": 0.6635,
+      "step": 3999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954891778326119e-05,
+      "loss": 0.5653,
+      "step": 4000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548621762298663e-05,
+      "loss": 0.5385,
+      "step": 4001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548325646479437e-05,
+      "loss": 0.6399,
+      "step": 4002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548029435806447e-05,
+      "loss": 0.6264,
+      "step": 4003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547733130282645e-05,
+      "loss": 0.4676,
+      "step": 4004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547436729910968e-05,
+      "loss": 0.5841,
+      "step": 4005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547140234694358e-05,
+      "loss": 0.6518,
+      "step": 4006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9546843644635768e-05,
+      "loss": 0.5168,
+      "step": 4007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954654695973814e-05,
+      "loss": 0.5576,
+      "step": 4008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954625018000442e-05,
+      "loss": 0.5286,
+      "step": 4009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545953305437564e-05,
+      "loss": 0.5158,
+      "step": 4010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954565633604051e-05,
+      "loss": 0.5402,
+      "step": 4011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545359271816212e-05,
+      "loss": 0.5651,
+      "step": 4012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545062112767627e-05,
+      "loss": 0.5017,
+      "step": 4013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95447648588977e-05,
+      "loss": 0.5426,
+      "step": 4014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954446751020939e-05,
+      "loss": 0.6045,
+      "step": 4015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954417006670564e-05,
+      "loss": 0.4905,
+      "step": 4016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954387252838942e-05,
+      "loss": 0.6121,
+      "step": 4017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543574895263673e-05,
+      "loss": 0.6263,
+      "step": 4018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543277167331362e-05,
+      "loss": 0.6145,
+      "step": 4019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542979344595444e-05,
+      "loss": 0.6134,
+      "step": 4020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542681427058877e-05,
+      "loss": 0.5927,
+      "step": 4021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954238341472462e-05,
+      "loss": 0.5329,
+      "step": 4022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954208530759563e-05,
+      "loss": 0.4951,
+      "step": 4023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541787105674876e-05,
+      "loss": 0.6691,
+      "step": 4024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541488808965318e-05,
+      "loss": 0.5487,
+      "step": 4025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541190417469917e-05,
+      "loss": 0.6006,
+      "step": 4026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540891931191637e-05,
+      "loss": 0.5534,
+      "step": 4027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540593350133444e-05,
+      "loss": 0.4534,
+      "step": 4028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540294674298305e-05,
+      "loss": 0.596,
+      "step": 4029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539995903689188e-05,
+      "loss": 0.5514,
+      "step": 4030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539697038309054e-05,
+      "loss": 0.5834,
+      "step": 4031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539398078160886e-05,
+      "loss": 0.643,
+      "step": 4032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539099023247637e-05,
+      "loss": 0.4797,
+      "step": 4033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9538799873572293e-05,
+      "loss": 0.5375,
+      "step": 4034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9538500629137814e-05,
+      "loss": 0.5726,
+      "step": 4035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9538201289947182e-05,
+      "loss": 0.5988,
+      "step": 4036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537901856003362e-05,
+      "loss": 0.5837,
+      "step": 4037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537602327309336e-05,
+      "loss": 0.5515,
+      "step": 4038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953730270386808e-05,
+      "loss": 0.5235,
+      "step": 4039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953700298568256e-05,
+      "loss": 0.6237,
+      "step": 4040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536703172755765e-05,
+      "loss": 0.5668,
+      "step": 4041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536403265090667e-05,
+      "loss": 0.572,
+      "step": 4042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536103262690246e-05,
+      "loss": 0.6117,
+      "step": 4043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535803165557484e-05,
+      "loss": 0.4835,
+      "step": 4044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535502973695362e-05,
+      "loss": 0.6007,
+      "step": 4045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535202687106863e-05,
+      "loss": 0.6328,
+      "step": 4046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953490230579497e-05,
+      "loss": 0.5594,
+      "step": 4047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953460182976266e-05,
+      "loss": 0.4877,
+      "step": 4048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9534301259012927e-05,
+      "loss": 0.5616,
+      "step": 4049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9534000593548756e-05,
+      "loss": 0.5993,
+      "step": 4050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953369983337313e-05,
+      "loss": 0.4774,
+      "step": 4051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533398978489035e-05,
+      "loss": 0.6167,
+      "step": 4052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533098028899465e-05,
+      "loss": 0.5304,
+      "step": 4053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532796984607408e-05,
+      "loss": 0.5469,
+      "step": 4054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532495845615854e-05,
+      "loss": 0.5999,
+      "step": 4055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532194611927795e-05,
+      "loss": 0.5334,
+      "step": 4056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531893283546223e-05,
+      "loss": 0.4798,
+      "step": 4057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953159186047413e-05,
+      "loss": 0.5801,
+      "step": 4058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531290342714514e-05,
+      "loss": 0.5386,
+      "step": 4059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9530988730270367e-05,
+      "loss": 0.5387,
+      "step": 4060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9530687023144688e-05,
+      "loss": 0.5925,
+      "step": 4061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.953038522134047e-05,
+      "loss": 0.6112,
+      "step": 4062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9530083324860718e-05,
+      "loss": 0.4356,
+      "step": 4063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529781333708425e-05,
+      "loss": 0.5507,
+      "step": 4064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952947924788659e-05,
+      "loss": 0.5786,
+      "step": 4065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529177067398218e-05,
+      "loss": 0.5547,
+      "step": 4066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952887479224631e-05,
+      "loss": 0.6466,
+      "step": 4067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528572422433872e-05,
+      "loss": 0.7622,
+      "step": 4068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528269957963898e-05,
+      "loss": 0.4867,
+      "step": 4069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527967398839402e-05,
+      "loss": 0.5266,
+      "step": 4070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527664745063388e-05,
+      "loss": 0.598,
+      "step": 4071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952736199663886e-05,
+      "loss": 0.6038,
+      "step": 4072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527059153568828e-05,
+      "loss": 0.4377,
+      "step": 4073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526756215856298e-05,
+      "loss": 0.5493,
+      "step": 4074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952645318350428e-05,
+      "loss": 0.5127,
+      "step": 4075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526150056515786e-05,
+      "loss": 0.5994,
+      "step": 4076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525846834893824e-05,
+      "loss": 0.5591,
+      "step": 4077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525543518641413e-05,
+      "loss": 0.5454,
+      "step": 4078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525240107761558e-05,
+      "loss": 0.5894,
+      "step": 4079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524936602257277e-05,
+      "loss": 0.6048,
+      "step": 4080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524633002131582e-05,
+      "loss": 0.6122,
+      "step": 4081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524329307387496e-05,
+      "loss": 0.633,
+      "step": 4082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952402551802803e-05,
+      "loss": 0.5492,
+      "step": 4083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523721634056204e-05,
+      "loss": 0.5267,
+      "step": 4084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952341765547504e-05,
+      "loss": 0.5708,
+      "step": 4085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523113582287546e-05,
+      "loss": 0.6013,
+      "step": 4086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952280941449676e-05,
+      "loss": 0.5305,
+      "step": 4087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952250515210569e-05,
+      "loss": 0.5048,
+      "step": 4088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952220079511736e-05,
+      "loss": 0.6353,
+      "step": 4089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521896343534797e-05,
+      "loss": 0.7093,
+      "step": 4090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952159179736103e-05,
+      "loss": 0.4544,
+      "step": 4091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521287156599075e-05,
+      "loss": 0.6346,
+      "step": 4092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520982421251964e-05,
+      "loss": 0.5435,
+      "step": 4093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520677591322724e-05,
+      "loss": 0.59,
+      "step": 4094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952037266681438e-05,
+      "loss": 0.4517,
+      "step": 4095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520067647729967e-05,
+      "loss": 0.6861,
+      "step": 4096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519762534072507e-05,
+      "loss": 0.6378,
+      "step": 4097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519457325845034e-05,
+      "loss": 0.4662,
+      "step": 4098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951915202305059e-05,
+      "loss": 0.5562,
+      "step": 4099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951884662569219e-05,
+      "loss": 0.5888,
+      "step": 4100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951854113377288e-05,
+      "loss": 0.57,
+      "step": 4101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9518235547295693e-05,
+      "loss": 0.6212,
+      "step": 4102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517929866263662e-05,
+      "loss": 0.5611,
+      "step": 4103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517624090679824e-05,
+      "loss": 0.5995,
+      "step": 4104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951731822054722e-05,
+      "loss": 0.6164,
+      "step": 4105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517012255868886e-05,
+      "loss": 0.6321,
+      "step": 4106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516706196647862e-05,
+      "loss": 0.5999,
+      "step": 4107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516400042887187e-05,
+      "loss": 0.5789,
+      "step": 4108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516093794589904e-05,
+      "loss": 0.675,
+      "step": 4109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515787451759057e-05,
+      "loss": 0.5749,
+      "step": 4110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515481014397684e-05,
+      "loss": 0.5468,
+      "step": 4111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515174482508833e-05,
+      "loss": 0.5327,
+      "step": 4112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951486785609555e-05,
+      "loss": 0.5303,
+      "step": 4113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514561135160876e-05,
+      "loss": 0.5608,
+      "step": 4114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514254319707868e-05,
+      "loss": 0.616,
+      "step": 4115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513947409739562e-05,
+      "loss": 0.5658,
+      "step": 4116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513640405259012e-05,
+      "loss": 0.6217,
+      "step": 4117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513333306269267e-05,
+      "loss": 0.5882,
+      "step": 4118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513026112773383e-05,
+      "loss": 0.5441,
+      "step": 4119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512718824774405e-05,
+      "loss": 0.654,
+      "step": 4120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951241144227539e-05,
+      "loss": 0.5534,
+      "step": 4121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512103965279388e-05,
+      "loss": 0.5671,
+      "step": 4122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511796393789453e-05,
+      "loss": 0.6028,
+      "step": 4123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511488727808643e-05,
+      "loss": 0.5705,
+      "step": 4124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951118096734002e-05,
+      "loss": 0.5607,
+      "step": 4125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951087311238663e-05,
+      "loss": 0.6072,
+      "step": 4126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.5653,
+      "step": 4127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.95102571190378e-05,
+      "loss": 0.601,
+      "step": 4128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950994898064848e-05,
+      "loss": 0.6104,
+      "step": 4129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509640747786632e-05,
+      "loss": 0.579,
+      "step": 4130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509332420455328e-05,
+      "loss": 0.5519,
+      "step": 4131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509023998657627e-05,
+      "loss": 0.562,
+      "step": 4132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950871548239659e-05,
+      "loss": 0.5896,
+      "step": 4133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950840687167528e-05,
+      "loss": 0.655,
+      "step": 4134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950809816649677e-05,
+      "loss": 0.5837,
+      "step": 4135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950778936686412e-05,
+      "loss": 0.5662,
+      "step": 4136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9507480472780406e-05,
+      "loss": 0.5978,
+      "step": 4137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9507171484248684e-05,
+      "loss": 0.6529,
+      "step": 4138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506862401272037e-05,
+      "loss": 0.5652,
+      "step": 4139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506553223853527e-05,
+      "loss": 0.5824,
+      "step": 4140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506243951996227e-05,
+      "loss": 0.5381,
+      "step": 4141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950593458570321e-05,
+      "loss": 0.4799,
+      "step": 4142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950562512497755e-05,
+      "loss": 0.5384,
+      "step": 4143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950531556982232e-05,
+      "loss": 0.6256,
+      "step": 4144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505005920240596e-05,
+      "loss": 0.5638,
+      "step": 4145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504696176235453e-05,
+      "loss": 0.4881,
+      "step": 4146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504386337809967e-05,
+      "loss": 0.5727,
+      "step": 4147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504076404967217e-05,
+      "loss": 0.66,
+      "step": 4148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950376637771029e-05,
+      "loss": 0.5813,
+      "step": 4149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950345625604225e-05,
+      "loss": 0.5219,
+      "step": 4150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503146039966188e-05,
+      "loss": 0.5992,
+      "step": 4151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502835729485186e-05,
+      "loss": 0.5236,
+      "step": 4152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502525324602324e-05,
+      "loss": 0.5885,
+      "step": 4153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502214825320683e-05,
+      "loss": 0.58,
+      "step": 4154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501904231643353e-05,
+      "loss": 0.5435,
+      "step": 4155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501593543573416e-05,
+      "loss": 0.5878,
+      "step": 4156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501282761113964e-05,
+      "loss": 0.5541,
+      "step": 4157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500971884268074e-05,
+      "loss": 0.5869,
+      "step": 4158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500660913038842e-05,
+      "loss": 0.5378,
+      "step": 4159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500349847429353e-05,
+      "loss": 0.5449,
+      "step": 4160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.95000386874427e-05,
+      "loss": 0.5377,
+      "step": 4161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499727433081976e-05,
+      "loss": 0.6689,
+      "step": 4162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949941608435027e-05,
+      "loss": 0.5501,
+      "step": 4163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499104641250676e-05,
+      "loss": 0.5834,
+      "step": 4164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498793103786288e-05,
+      "loss": 0.6119,
+      "step": 4165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498481471960197e-05,
+      "loss": 0.6461,
+      "step": 4166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498169745775502e-05,
+      "loss": 0.5112,
+      "step": 4167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497857925235302e-05,
+      "loss": 0.5741,
+      "step": 4168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497546010342693e-05,
+      "loss": 0.6095,
+      "step": 4169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949723400110077e-05,
+      "loss": 0.5687,
+      "step": 4170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496921897512635e-05,
+      "loss": 0.4739,
+      "step": 4171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949660969958139e-05,
+      "loss": 0.614,
+      "step": 4172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496297407310135e-05,
+      "loss": 0.5027,
+      "step": 4173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495985020701977e-05,
+      "loss": 0.5799,
+      "step": 4174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949567253976001e-05,
+      "loss": 0.5704,
+      "step": 4175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495359964487344e-05,
+      "loss": 0.6057,
+      "step": 4176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949504729488708e-05,
+      "loss": 0.5622,
+      "step": 4177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494734530962333e-05,
+      "loss": 0.5005,
+      "step": 4178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.94944216727162e-05,
+      "loss": 0.5533,
+      "step": 4179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494108720151795e-05,
+      "loss": 0.6519,
+      "step": 4180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493795673272228e-05,
+      "loss": 0.5491,
+      "step": 4181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.94934825320806e-05,
+      "loss": 0.535,
+      "step": 4182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493169296580035e-05,
+      "loss": 0.4637,
+      "step": 4183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9492855966773632e-05,
+      "loss": 0.5593,
+      "step": 4184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949254254266451e-05,
+      "loss": 0.5554,
+      "step": 4185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9492229024255783e-05,
+      "loss": 0.5099,
+      "step": 4186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491915411550564e-05,
+      "loss": 0.5351,
+      "step": 4187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491601704551967e-05,
+      "loss": 0.6734,
+      "step": 4188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491287903263113e-05,
+      "loss": 0.5361,
+      "step": 4189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9490974007687114e-05,
+      "loss": 0.528,
+      "step": 4190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9490660017827096e-05,
+      "loss": 0.5634,
+      "step": 4191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490345933686165e-05,
+      "loss": 0.5293,
+      "step": 4192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490031755267454e-05,
+      "loss": 0.5661,
+      "step": 4193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489717482574074e-05,
+      "loss": 0.6183,
+      "step": 4194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489403115609157e-05,
+      "loss": 0.5711,
+      "step": 4195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489088654375822e-05,
+      "loss": 0.458,
+      "step": 4196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488774098877186e-05,
+      "loss": 0.4397,
+      "step": 4197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488459449116384e-05,
+      "loss": 0.5833,
+      "step": 4198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488144705096535e-05,
+      "loss": 0.5313,
+      "step": 4199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948782986682077e-05,
+      "loss": 0.5155,
+      "step": 4200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487514934292216e-05,
+      "loss": 0.5129,
+      "step": 4201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487199907514e-05,
+      "loss": 0.6063,
+      "step": 4202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948688478648925e-05,
+      "loss": 0.5089,
+      "step": 4203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486569571221096e-05,
+      "loss": 0.6043,
+      "step": 4204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486254261712677e-05,
+      "loss": 0.5354,
+      "step": 4205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485938857967116e-05,
+      "loss": 0.6392,
+      "step": 4206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948562335998755e-05,
+      "loss": 0.6155,
+      "step": 4207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485307767777115e-05,
+      "loss": 0.5871,
+      "step": 4208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484992081338946e-05,
+      "loss": 0.5715,
+      "step": 4209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484676300676172e-05,
+      "loss": 0.5069,
+      "step": 4210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484360425791943e-05,
+      "loss": 0.5355,
+      "step": 4211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484044456689383e-05,
+      "loss": 0.5558,
+      "step": 4212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483728393371643e-05,
+      "loss": 0.5997,
+      "step": 4213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483412235841853e-05,
+      "loss": 0.5865,
+      "step": 4214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948309598410316e-05,
+      "loss": 0.55,
+      "step": 4215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94827796381587e-05,
+      "loss": 0.5314,
+      "step": 4216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482463198011625e-05,
+      "loss": 0.5572,
+      "step": 4217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482146663665066e-05,
+      "loss": 0.4975,
+      "step": 4218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481830035122176e-05,
+      "loss": 0.6334,
+      "step": 4219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481513312386102e-05,
+      "loss": 0.5052,
+      "step": 4220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481196495459984e-05,
+      "loss": 0.4935,
+      "step": 4221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948087958434697e-05,
+      "loss": 0.6011,
+      "step": 4222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9480562579050216e-05,
+      "loss": 0.5588,
+      "step": 4223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948024547957286e-05,
+      "loss": 0.5939,
+      "step": 4224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479928285918065e-05,
+      "loss": 0.4362,
+      "step": 4225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479610998088967e-05,
+      "loss": 0.4918,
+      "step": 4226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947929361608873e-05,
+      "loss": 0.5926,
+      "step": 4227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94789761399205e-05,
+      "loss": 0.5583,
+      "step": 4228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478658569587433e-05,
+      "loss": 0.5992,
+      "step": 4229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478340905092687e-05,
+      "loss": 0.4955,
+      "step": 4230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478023146439415e-05,
+      "loss": 0.5,
+      "step": 4231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947770529363077e-05,
+      "loss": 0.6047,
+      "step": 4232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477387346669916e-05,
+      "loss": 0.6216,
+      "step": 4233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477069305560006e-05,
+      "loss": 0.6152,
+      "step": 4234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476751170304206e-05,
+      "loss": 0.4689,
+      "step": 4235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476432940905668e-05,
+      "loss": 0.5704,
+      "step": 4236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947611461736756e-05,
+      "loss": 0.5313,
+      "step": 4237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947579619969304e-05,
+      "loss": 0.5621,
+      "step": 4238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475477687885272e-05,
+      "loss": 0.5701,
+      "step": 4239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475159081947423e-05,
+      "loss": 0.4595,
+      "step": 4240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474840381882658e-05,
+      "loss": 0.6615,
+      "step": 4241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947452158769414e-05,
+      "loss": 0.5273,
+      "step": 4242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474202699385038e-05,
+      "loss": 0.543,
+      "step": 4243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473883716958516e-05,
+      "loss": 0.5166,
+      "step": 4244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947356464041775e-05,
+      "loss": 0.5874,
+      "step": 4245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473245469765904e-05,
+      "loss": 0.5585,
+      "step": 4246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947292620500615e-05,
+      "loss": 0.6279,
+      "step": 4247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947260684614166e-05,
+      "loss": 0.5712,
+      "step": 4248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472287393175605e-05,
+      "loss": 0.5861,
+      "step": 4249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947196784611116e-05,
+      "loss": 0.5458,
+      "step": 4250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94716482049515e-05,
+      "loss": 0.6226,
+      "step": 4251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94713284696998e-05,
+      "loss": 0.5728,
+      "step": 4252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947100864035923e-05,
+      "loss": 0.5082,
+      "step": 4253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470688716932983e-05,
+      "loss": 0.5714,
+      "step": 4254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947036869942422e-05,
+      "loss": 0.5187,
+      "step": 4255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470048587836132e-05,
+      "loss": 0.6038,
+      "step": 4256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469728382171892e-05,
+      "loss": 0.6743,
+      "step": 4257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469408082434682e-05,
+      "loss": 0.5451,
+      "step": 4258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469087688627687e-05,
+      "loss": 0.6348,
+      "step": 4259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468767200754085e-05,
+      "loss": 0.4785,
+      "step": 4260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468446618817064e-05,
+      "loss": 0.462,
+      "step": 4261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946812594281981e-05,
+      "loss": 0.507,
+      "step": 4262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467805172765506e-05,
+      "loss": 0.5516,
+      "step": 4263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467484308657335e-05,
+      "loss": 0.5203,
+      "step": 4264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946716335049849e-05,
+      "loss": 0.6052,
+      "step": 4265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466842298292156e-05,
+      "loss": 0.5374,
+      "step": 4266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466521152041528e-05,
+      "loss": 0.4881,
+      "step": 4267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946619991174979e-05,
+      "loss": 0.533,
+      "step": 4268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465878577420133e-05,
+      "loss": 0.5117,
+      "step": 4269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946555714905575e-05,
+      "loss": 0.5316,
+      "step": 4270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946523562665984e-05,
+      "loss": 0.5517,
+      "step": 4271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9464914010235592e-05,
+      "loss": 0.5837,
+      "step": 4272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94645922997862e-05,
+      "loss": 0.5572,
+      "step": 4273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946427049531486e-05,
+      "loss": 0.6397,
+      "step": 4274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463948596824774e-05,
+      "loss": 0.5236,
+      "step": 4275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463626604319135e-05,
+      "loss": 0.5408,
+      "step": 4276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946330451780114e-05,
+      "loss": 0.5585,
+      "step": 4277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462982337273995e-05,
+      "loss": 0.5846,
+      "step": 4278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462660062740896e-05,
+      "loss": 0.4462,
+      "step": 4279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462337694205042e-05,
+      "loss": 0.5354,
+      "step": 4280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462015231669646e-05,
+      "loss": 0.5202,
+      "step": 4281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94616926751379e-05,
+      "loss": 0.5759,
+      "step": 4282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946137002461301e-05,
+      "loss": 0.5885,
+      "step": 4283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461047280098184e-05,
+      "loss": 0.5408,
+      "step": 4284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946072444159663e-05,
+      "loss": 0.5704,
+      "step": 4285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460401509111552e-05,
+      "loss": 0.6353,
+      "step": 4286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460078482646163e-05,
+      "loss": 0.5786,
+      "step": 4287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945975536220366e-05,
+      "loss": 0.5284,
+      "step": 4288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945943214778727e-05,
+      "loss": 0.5256,
+      "step": 4289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945910883940019e-05,
+      "loss": 0.5327,
+      "step": 4290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458785437045636e-05,
+      "loss": 0.5864,
+      "step": 4291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458461940726826e-05,
+      "loss": 0.5956,
+      "step": 4292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458138350446963e-05,
+      "loss": 0.6185,
+      "step": 4293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457814666209272e-05,
+      "loss": 0.5909,
+      "step": 4294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457490888016962e-05,
+      "loss": 0.6797,
+      "step": 4295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945716701587325e-05,
+      "loss": 0.5212,
+      "step": 4296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456843049781354e-05,
+      "loss": 0.5214,
+      "step": 4297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456518989744496e-05,
+      "loss": 0.5801,
+      "step": 4298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945619483576589e-05,
+      "loss": 0.5152,
+      "step": 4299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945587058784876e-05,
+      "loss": 0.586,
+      "step": 4300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455546245996328e-05,
+      "loss": 0.5384,
+      "step": 4301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945522181021181e-05,
+      "loss": 0.5363,
+      "step": 4302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945489728049843e-05,
+      "loss": 0.503,
+      "step": 4303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945457265685942e-05,
+      "loss": 0.5545,
+      "step": 4304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9454247939297996e-05,
+      "loss": 0.506,
+      "step": 4305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945392312781739e-05,
+      "loss": 0.5991,
+      "step": 4306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9453598222420824e-05,
+      "loss": 0.5428,
+      "step": 4307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9453273223111526e-05,
+      "loss": 0.5327,
+      "step": 4308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452948129892725e-05,
+      "loss": 0.5611,
+      "step": 4309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452622942767654e-05,
+      "loss": 0.585,
+      "step": 4310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452297661739542e-05,
+      "loss": 0.5767,
+      "step": 4311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945197228681162e-05,
+      "loss": 0.495,
+      "step": 4312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451646817987115e-05,
+      "loss": 0.5528,
+      "step": 4313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451321255269268e-05,
+      "loss": 0.57,
+      "step": 4314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945099559866131e-05,
+      "loss": 0.4972,
+      "step": 4315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450669848166476e-05,
+      "loss": 0.5857,
+      "step": 4316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450344003788e-05,
+      "loss": 0.5215,
+      "step": 4317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450018065529123e-05,
+      "loss": 0.5936,
+      "step": 4318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.944969203339308e-05,
+      "loss": 0.5825,
+      "step": 4319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.944936590738311e-05,
+      "loss": 0.4942,
+      "step": 4320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449039687502457e-05,
+      "loss": 0.544,
+      "step": 4321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9448713373754353e-05,
+      "loss": 0.4893,
+      "step": 4322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944838696614205e-05,
+      "loss": 0.5316,
+      "step": 4323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944806046466878e-05,
+      "loss": 0.5299,
+      "step": 4324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94477338693378e-05,
+      "loss": 0.4954,
+      "step": 4325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944740718015234e-05,
+      "loss": 0.4772,
+      "step": 4326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447080397115654e-05,
+      "loss": 0.5922,
+      "step": 4327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944675352023098e-05,
+      "loss": 0.4997,
+      "step": 4328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446426549501578e-05,
+      "loss": 0.6009,
+      "step": 4329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446099484930686e-05,
+      "loss": 0.5239,
+      "step": 4330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445772326521558e-05,
+      "loss": 0.5364,
+      "step": 4331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445445074277442e-05,
+      "loss": 0.5138,
+      "step": 4332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944511772820159e-05,
+      "loss": 0.5796,
+      "step": 4333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444790288297252e-05,
+      "loss": 0.5168,
+      "step": 4334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444462754567682e-05,
+      "loss": 0.5733,
+      "step": 4335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444135127016136e-05,
+      "loss": 0.5886,
+      "step": 4336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443807405645865e-05,
+      "loss": 0.4565,
+      "step": 4337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443479590460125e-05,
+      "loss": 0.5443,
+      "step": 4338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443151681462177e-05,
+      "loss": 0.5563,
+      "step": 4339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442823678655275e-05,
+      "loss": 0.5174,
+      "step": 4340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442495582042673e-05,
+      "loss": 0.6045,
+      "step": 4341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442167391627636e-05,
+      "loss": 0.5548,
+      "step": 4342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441839107413428e-05,
+      "loss": 0.5927,
+      "step": 4343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441510729403297e-05,
+      "loss": 0.6162,
+      "step": 4344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944118225760052e-05,
+      "loss": 0.5914,
+      "step": 4345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944085369200835e-05,
+      "loss": 0.5722,
+      "step": 4346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440525032630057e-05,
+      "loss": 0.5417,
+      "step": 4347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440196279468905e-05,
+      "loss": 0.4878,
+      "step": 4348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439867432528155e-05,
+      "loss": 0.5227,
+      "step": 4349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439538491811076e-05,
+      "loss": 0.556,
+      "step": 4350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943920945732094e-05,
+      "loss": 0.4922,
+      "step": 4351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943888032906101e-05,
+      "loss": 0.5844,
+      "step": 4352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943855110703456e-05,
+      "loss": 0.6202,
+      "step": 4353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9438221791244854e-05,
+      "loss": 0.5651,
+      "step": 4354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437892381695174e-05,
+      "loss": 0.4772,
+      "step": 4355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437562878388783e-05,
+      "loss": 0.6005,
+      "step": 4356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943723328132896e-05,
+      "loss": 0.5373,
+      "step": 4357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436903590518975e-05,
+      "loss": 0.535,
+      "step": 4358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436573805962108e-05,
+      "loss": 0.5327,
+      "step": 4359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436243927661627e-05,
+      "loss": 0.5625,
+      "step": 4360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435913955620816e-05,
+      "loss": 0.5622,
+      "step": 4361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435583889842956e-05,
+      "loss": 0.5898,
+      "step": 4362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435253730331314e-05,
+      "loss": 0.6561,
+      "step": 4363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434923477089182e-05,
+      "loss": 0.6295,
+      "step": 4364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434593130119835e-05,
+      "loss": 0.5212,
+      "step": 4365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434262689426554e-05,
+      "loss": 0.5226,
+      "step": 4366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943393215501262e-05,
+      "loss": 0.5314,
+      "step": 4367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433601526881326e-05,
+      "loss": 0.6243,
+      "step": 4368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433270805035946e-05,
+      "loss": 0.5488,
+      "step": 4369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432939989479767e-05,
+      "loss": 0.5632,
+      "step": 4370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432609080216082e-05,
+      "loss": 0.4997,
+      "step": 4371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432278077248175e-05,
+      "loss": 0.4874,
+      "step": 4372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943194698057933e-05,
+      "loss": 0.5797,
+      "step": 4373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431615790212842e-05,
+      "loss": 0.5478,
+      "step": 4374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431284506151996e-05,
+      "loss": 0.5126,
+      "step": 4375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430953128400087e-05,
+      "loss": 0.5168,
+      "step": 4376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430621656960405e-05,
+      "loss": 0.6127,
+      "step": 4377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430290091836242e-05,
+      "loss": 0.5253,
+      "step": 4378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429958433030897e-05,
+      "loss": 0.5143,
+      "step": 4379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429626680547658e-05,
+      "loss": 0.6021,
+      "step": 4380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429294834389823e-05,
+      "loss": 0.558,
+      "step": 4381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428962894560687e-05,
+      "loss": 0.6023,
+      "step": 4382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428630861063554e-05,
+      "loss": 0.5769,
+      "step": 4383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428298733901715e-05,
+      "loss": 0.5286,
+      "step": 4384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9427966513078476e-05,
+      "loss": 0.566,
+      "step": 4385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.942763419859713e-05,
+      "loss": 0.585,
+      "step": 4386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942730179046098e-05,
+      "loss": 0.4885,
+      "step": 4387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426969288673335e-05,
+      "loss": 0.6295,
+      "step": 4388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426636693237493e-05,
+      "loss": 0.5195,
+      "step": 4389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426304004156758e-05,
+      "loss": 0.5544,
+      "step": 4390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425971221434434e-05,
+      "loss": 0.537,
+      "step": 4391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425638345073826e-05,
+      "loss": 0.5668,
+      "step": 4392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425305375078243e-05,
+      "loss": 0.5788,
+      "step": 4393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424972311450997e-05,
+      "loss": 0.4631,
+      "step": 4394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424639154195386e-05,
+      "loss": 0.6425,
+      "step": 4395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942430590331473e-05,
+      "loss": 0.5518,
+      "step": 4396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423972558812334e-05,
+      "loss": 0.5969,
+      "step": 4397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423639120691514e-05,
+      "loss": 0.5879,
+      "step": 4398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423305588955576e-05,
+      "loss": 0.5917,
+      "step": 4399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9422971963607836e-05,
+      "loss": 0.6205,
+      "step": 4400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942263824465161e-05,
+      "loss": 0.5947,
+      "step": 4401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9422304432090212e-05,
+      "loss": 0.5286,
+      "step": 4402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421970525926958e-05,
+      "loss": 0.6084,
+      "step": 4403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942163652616517e-05,
+      "loss": 0.6187,
+      "step": 4404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421302432808153e-05,
+      "loss": 0.5208,
+      "step": 4405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942096824585924e-05,
+      "loss": 0.5987,
+      "step": 4406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942063396532174e-05,
+      "loss": 0.6457,
+      "step": 4407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420299591198987e-05,
+      "loss": 0.6213,
+      "step": 4408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419965123494288e-05,
+      "loss": 0.7177,
+      "step": 4409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941963056221097e-05,
+      "loss": 0.5589,
+      "step": 4410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419295907352364e-05,
+      "loss": 0.5578,
+      "step": 4411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9418961158921787e-05,
+      "loss": 0.5563,
+      "step": 4412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9418626316922566e-05,
+      "loss": 0.5146,
+      "step": 4413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9418291381358026e-05,
+      "loss": 0.5521,
+      "step": 4414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94179563522315e-05,
+      "loss": 0.4965,
+      "step": 4415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417621229546314e-05,
+      "loss": 0.5575,
+      "step": 4416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941728601330579e-05,
+      "loss": 0.533,
+      "step": 4417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416950703513264e-05,
+      "loss": 0.562,
+      "step": 4418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416615300172072e-05,
+      "loss": 0.6247,
+      "step": 4419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416279803285537e-05,
+      "loss": 0.5325,
+      "step": 4420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9415944212856993e-05,
+      "loss": 0.6132,
+      "step": 4421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9415608528889777e-05,
+      "loss": 0.6529,
+      "step": 4422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9415272751387224e-05,
+      "loss": 0.5699,
+      "step": 4423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941493688035267e-05,
+      "loss": 0.5842,
+      "step": 4424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414600915789446e-05,
+      "loss": 0.5204,
+      "step": 4425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414264857700897e-05,
+      "loss": 0.5061,
+      "step": 4426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413928706090355e-05,
+      "loss": 0.561,
+      "step": 4427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413592460961162e-05,
+      "loss": 0.5696,
+      "step": 4428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413256122316662e-05,
+      "loss": 0.5681,
+      "step": 4429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412919690160193e-05,
+      "loss": 0.5812,
+      "step": 4430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412583164495094e-05,
+      "loss": 0.619,
+      "step": 4431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941224654532471e-05,
+      "loss": 0.5118,
+      "step": 4432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941190983265239e-05,
+      "loss": 0.5822,
+      "step": 4433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941157302648147e-05,
+      "loss": 0.6629,
+      "step": 4434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411236126815306e-05,
+      "loss": 0.5588,
+      "step": 4435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410899133657237e-05,
+      "loss": 0.5159,
+      "step": 4436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410562047010613e-05,
+      "loss": 0.6676,
+      "step": 4437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410224866878785e-05,
+      "loss": 0.586,
+      "step": 4438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409887593265098e-05,
+      "loss": 0.5713,
+      "step": 4439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409550226172902e-05,
+      "loss": 0.5524,
+      "step": 4440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940921276560556e-05,
+      "loss": 0.5914,
+      "step": 4441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408875211566406e-05,
+      "loss": 0.4797,
+      "step": 4442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940853756405881e-05,
+      "loss": 0.4973,
+      "step": 4443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408199823086115e-05,
+      "loss": 0.5543,
+      "step": 4444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407861988651683e-05,
+      "loss": 0.5716,
+      "step": 4445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407524060758865e-05,
+      "loss": 0.4836,
+      "step": 4446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407186039411025e-05,
+      "loss": 0.6614,
+      "step": 4447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9406847924611513e-05,
+      "loss": 0.5482,
+      "step": 4448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940650971636369e-05,
+      "loss": 0.5926,
+      "step": 4449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940617141467092e-05,
+      "loss": 0.6119,
+      "step": 4450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940583301953656e-05,
+      "loss": 0.6045,
+      "step": 4451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9405494530963975e-05,
+      "loss": 0.5536,
+      "step": 4452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940515594895652e-05,
+      "loss": 0.5393,
+      "step": 4453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404817273517565e-05,
+      "loss": 0.6945,
+      "step": 4454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404478504650473e-05,
+      "loss": 0.5888,
+      "step": 4455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940413964235861e-05,
+      "loss": 0.5621,
+      "step": 4456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403800686645343e-05,
+      "loss": 0.5173,
+      "step": 4457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403461637514037e-05,
+      "loss": 0.5678,
+      "step": 4458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403122494968057e-05,
+      "loss": 0.4439,
+      "step": 4459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940278325901078e-05,
+      "loss": 0.5003,
+      "step": 4460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402443929645568e-05,
+      "loss": 0.5484,
+      "step": 4461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.94021045068758e-05,
+      "loss": 0.4556,
+      "step": 4462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9401764990704842e-05,
+      "loss": 0.5303,
+      "step": 4463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940142538113607e-05,
+      "loss": 0.5692,
+      "step": 4464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9401085678172852e-05,
+      "loss": 0.5546,
+      "step": 4465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400745881818573e-05,
+      "loss": 0.5556,
+      "step": 4466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400405992076598e-05,
+      "loss": 0.4884,
+      "step": 4467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940006600895031e-05,
+      "loss": 0.5192,
+      "step": 4468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399725932443083e-05,
+      "loss": 0.5393,
+      "step": 4469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399385762558295e-05,
+      "loss": 0.5342,
+      "step": 4470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939904549929933e-05,
+      "loss": 0.5758,
+      "step": 4471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398705142669562e-05,
+      "loss": 0.5135,
+      "step": 4472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939836469267238e-05,
+      "loss": 0.545,
+      "step": 4473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398024149311155e-05,
+      "loss": 0.5121,
+      "step": 4474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397683512589282e-05,
+      "loss": 0.5328,
+      "step": 4475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397342782510134e-05,
+      "loss": 0.5459,
+      "step": 4476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397001959077107e-05,
+      "loss": 0.5008,
+      "step": 4477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396661042293575e-05,
+      "loss": 0.5188,
+      "step": 4478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396320032162936e-05,
+      "loss": 0.5653,
+      "step": 4479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939597892868857e-05,
+      "loss": 0.5519,
+      "step": 4480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9395637731873864e-05,
+      "loss": 0.5991,
+      "step": 4481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9395296441722216e-05,
+      "loss": 0.5938,
+      "step": 4482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939495505823701e-05,
+      "loss": 0.5339,
+      "step": 4483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394613581421638e-05,
+      "loss": 0.6748,
+      "step": 4484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394272011279492e-05,
+      "loss": 0.4976,
+      "step": 4485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939393034781397e-05,
+      "loss": 0.5957,
+      "step": 4486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393588591028458e-05,
+      "loss": 0.5776,
+      "step": 4487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393246740926358e-05,
+      "loss": 0.5436,
+      "step": 4488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392904797511064e-05,
+      "loss": 0.507,
+      "step": 4489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939256276078597e-05,
+      "loss": 0.5192,
+      "step": 4490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392220630754477e-05,
+      "loss": 0.5759,
+      "step": 4491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391878407419984e-05,
+      "loss": 0.5403,
+      "step": 4492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391536090785887e-05,
+      "loss": 0.5373,
+      "step": 4493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391193680855592e-05,
+      "loss": 0.6386,
+      "step": 4494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390851177632496e-05,
+      "loss": 0.516,
+      "step": 4495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390508581120003e-05,
+      "loss": 0.5483,
+      "step": 4496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939016589132152e-05,
+      "loss": 0.5524,
+      "step": 4497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938982310824044e-05,
+      "loss": 0.5503,
+      "step": 4498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389480231880185e-05,
+      "loss": 0.5723,
+      "step": 4499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938913726224415e-05,
+      "loss": 0.5808,
+      "step": 4500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938879419933574e-05,
+      "loss": 0.6432,
+      "step": 4501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938845104315837e-05,
+      "loss": 0.5063,
+      "step": 4502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388107793715446e-05,
+      "loss": 0.5361,
+      "step": 4503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938776445101038e-05,
+      "loss": 0.6374,
+      "step": 4504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9387421015046576e-05,
+      "loss": 0.5817,
+      "step": 4505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9387077485827453e-05,
+      "loss": 0.5301,
+      "step": 4506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938673386335642e-05,
+      "loss": 0.5347,
+      "step": 4507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386390147636895e-05,
+      "loss": 0.545,
+      "step": 4508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386046338672286e-05,
+      "loss": 0.6459,
+      "step": 4509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385702436466013e-05,
+      "loss": 0.674,
+      "step": 4510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938535844102149e-05,
+      "loss": 0.5882,
+      "step": 4511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385014352342138e-05,
+      "loss": 0.5235,
+      "step": 4512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384670170431373e-05,
+      "loss": 0.4406,
+      "step": 4513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384325895292614e-05,
+      "loss": 0.6175,
+      "step": 4514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9383981526929277e-05,
+      "loss": 0.4392,
+      "step": 4515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9383637065344787e-05,
+      "loss": 0.5888,
+      "step": 4516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9383292510542566e-05,
+      "loss": 0.585,
+      "step": 4517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938294786252604e-05,
+      "loss": 0.5816,
+      "step": 4518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382603121298623e-05,
+      "loss": 0.596,
+      "step": 4519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938225828686375e-05,
+      "loss": 0.5444,
+      "step": 4520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381913359224844e-05,
+      "loss": 0.5257,
+      "step": 4521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938156833838533e-05,
+      "loss": 0.6617,
+      "step": 4522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381223224348632e-05,
+      "loss": 0.551,
+      "step": 4523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380878017118183e-05,
+      "loss": 0.4943,
+      "step": 4524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380532716697412e-05,
+      "loss": 0.5312,
+      "step": 4525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380187323089746e-05,
+      "loss": 0.5682,
+      "step": 4526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937984183629862e-05,
+      "loss": 0.5975,
+      "step": 4527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379496256327466e-05,
+      "loss": 0.6347,
+      "step": 4528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379150583179713e-05,
+      "loss": 0.6855,
+      "step": 4529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.93788048168588e-05,
+      "loss": 0.6406,
+      "step": 4530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937845895736816e-05,
+      "loss": 0.6361,
+      "step": 4531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9378113004711227e-05,
+      "loss": 0.5401,
+      "step": 4532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937776695889144e-05,
+      "loss": 0.5097,
+      "step": 4533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377420819912236e-05,
+      "loss": 0.5736,
+      "step": 4534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937707458777705e-05,
+      "loss": 0.6249,
+      "step": 4535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937672826248933e-05,
+      "loss": 0.4875,
+      "step": 4536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376381844052507e-05,
+      "loss": 0.5892,
+      "step": 4537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376035332470028e-05,
+      "loss": 0.637,
+      "step": 4538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375688727745333e-05,
+      "loss": 0.5446,
+      "step": 4539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375342029881868e-05,
+      "loss": 0.4991,
+      "step": 4540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937499523888307e-05,
+      "loss": 0.6203,
+      "step": 4541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374648354752398e-05,
+      "loss": 0.4557,
+      "step": 4542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374301377493282e-05,
+      "loss": 0.5105,
+      "step": 4543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373954307109178e-05,
+      "loss": 0.5965,
+      "step": 4544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373607143603533e-05,
+      "loss": 0.5759,
+      "step": 4545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937325988697979e-05,
+      "loss": 0.4128,
+      "step": 4546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9372912537241406e-05,
+      "loss": 0.5697,
+      "step": 4547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937256509439183e-05,
+      "loss": 0.5438,
+      "step": 4548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937221755843451e-05,
+      "loss": 0.5195,
+      "step": 4549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371869929372903e-05,
+      "loss": 0.4838,
+      "step": 4550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371522207210462e-05,
+      "loss": 0.5518,
+      "step": 4551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371174391950636e-05,
+      "loss": 0.6047,
+      "step": 4552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370826483596885e-05,
+      "loss": 0.5949,
+      "step": 4553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370478482152665e-05,
+      "loss": 0.6096,
+      "step": 4554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937013038762143e-05,
+      "loss": 0.5705,
+      "step": 4555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936978220000664e-05,
+      "loss": 0.6169,
+      "step": 4556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369433919311754e-05,
+      "loss": 0.6222,
+      "step": 4557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369085545540233e-05,
+      "loss": 0.6842,
+      "step": 4558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368737078695535e-05,
+      "loss": 0.6134,
+      "step": 4559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368388518781124e-05,
+      "loss": 0.6321,
+      "step": 4560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368039865800465e-05,
+      "loss": 0.5277,
+      "step": 4561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367691119757014e-05,
+      "loss": 0.5325,
+      "step": 4562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936734228065424e-05,
+      "loss": 0.4929,
+      "step": 4563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936699334849561e-05,
+      "loss": 0.5819,
+      "step": 4564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366644323284584e-05,
+      "loss": 0.5841,
+      "step": 4565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366295205024637e-05,
+      "loss": 0.5214,
+      "step": 4566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365945993719237e-05,
+      "loss": 0.5111,
+      "step": 4567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365596689371847e-05,
+      "loss": 0.5177,
+      "step": 4568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936524729198594e-05,
+      "loss": 0.5463,
+      "step": 4569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364897801564984e-05,
+      "loss": 0.5366,
+      "step": 4570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364548218112456e-05,
+      "loss": 0.6156,
+      "step": 4571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364198541631828e-05,
+      "loss": 0.5893,
+      "step": 4572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936384877212657e-05,
+      "loss": 0.6048,
+      "step": 4573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363498909600164e-05,
+      "loss": 0.528,
+      "step": 4574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363148954056077e-05,
+      "loss": 0.5582,
+      "step": 4575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936279890549779e-05,
+      "loss": 0.5588,
+      "step": 4576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362448763928777e-05,
+      "loss": 0.5572,
+      "step": 4577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362098529352522e-05,
+      "loss": 0.5473,
+      "step": 4578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.93617482017725e-05,
+      "loss": 0.475,
+      "step": 4579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936139778119219e-05,
+      "loss": 0.6021,
+      "step": 4580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9361047267615077e-05,
+      "loss": 0.5442,
+      "step": 4581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.936069666104464e-05,
+      "loss": 0.691,
+      "step": 4582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360345961484366e-05,
+      "loss": 0.5627,
+      "step": 4583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9359995168937734e-05,
+      "loss": 0.5677,
+      "step": 4584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9359644283408233e-05,
+      "loss": 0.6486,
+      "step": 4585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9359293304899342e-05,
+      "loss": 0.5638,
+      "step": 4586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358942233414555e-05,
+      "loss": 0.5063,
+      "step": 4587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935859106895736e-05,
+      "loss": 0.5337,
+      "step": 4588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935823981153124e-05,
+      "loss": 0.6855,
+      "step": 4589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357888461139685e-05,
+      "loss": 0.6457,
+      "step": 4590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357537017786188e-05,
+      "loss": 0.5,
+      "step": 4591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935718548147424e-05,
+      "loss": 0.5576,
+      "step": 4592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356833852207334e-05,
+      "loss": 0.579,
+      "step": 4593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356482129988957e-05,
+      "loss": 0.5147,
+      "step": 4594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935613031482261e-05,
+      "loss": 0.5719,
+      "step": 4595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355778406711786e-05,
+      "loss": 0.548,
+      "step": 4596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935542640565998e-05,
+      "loss": 0.5538,
+      "step": 4597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355074311670692e-05,
+      "loss": 0.5408,
+      "step": 4598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354722124747417e-05,
+      "loss": 0.5714,
+      "step": 4599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935436984489365e-05,
+      "loss": 0.6033,
+      "step": 4600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354017472112896e-05,
+      "loss": 0.6082,
+      "step": 4601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353665006408656e-05,
+      "loss": 0.6046,
+      "step": 4602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353312447784428e-05,
+      "loss": 0.5974,
+      "step": 4603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352959796243714e-05,
+      "loss": 0.5906,
+      "step": 4604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935260705179002e-05,
+      "loss": 0.4789,
+      "step": 4605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935225421442685e-05,
+      "loss": 0.5528,
+      "step": 4606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351901284157707e-05,
+      "loss": 0.6041,
+      "step": 4607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351548260986095e-05,
+      "loss": 0.5631,
+      "step": 4608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351195144915528e-05,
+      "loss": 0.5847,
+      "step": 4609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935084193594951e-05,
+      "loss": 0.5947,
+      "step": 4610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350488634091545e-05,
+      "loss": 0.5832,
+      "step": 4611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350135239345157e-05,
+      "loss": 0.6054,
+      "step": 4612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934978175171384e-05,
+      "loss": 0.5871,
+      "step": 4613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934942817120111e-05,
+      "loss": 0.5812,
+      "step": 4614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934907449781049e-05,
+      "loss": 0.5475,
+      "step": 4615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934872073154548e-05,
+      "loss": 0.5973,
+      "step": 4616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348366872409607e-05,
+      "loss": 0.5474,
+      "step": 4617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348012920406375e-05,
+      "loss": 0.5324,
+      "step": 4618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9347658875539302e-05,
+      "loss": 0.5437,
+      "step": 4619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934730473781191e-05,
+      "loss": 0.5046,
+      "step": 4620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346950507227713e-05,
+      "loss": 0.5264,
+      "step": 4621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346596183790233e-05,
+      "loss": 0.5337,
+      "step": 4622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346241767502988e-05,
+      "loss": 0.5481,
+      "step": 4623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93458872583695e-05,
+      "loss": 0.5749,
+      "step": 4624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934553265639329e-05,
+      "loss": 0.522,
+      "step": 4625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934517796157788e-05,
+      "loss": 0.5832,
+      "step": 4626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344823173926792e-05,
+      "loss": 0.5025,
+      "step": 4627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344468293443553e-05,
+      "loss": 0.5818,
+      "step": 4628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344113320131693e-05,
+      "loss": 0.6032,
+      "step": 4629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934375825399473e-05,
+      "loss": 0.5286,
+      "step": 4630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343403095036192e-05,
+      "loss": 0.6905,
+      "step": 4631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934304784325961e-05,
+      "loss": 0.5328,
+      "step": 4632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9342692498668518e-05,
+      "loss": 0.4417,
+      "step": 4633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934233706126644e-05,
+      "loss": 0.471,
+      "step": 4634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9341981531056905e-05,
+      "loss": 0.5587,
+      "step": 4635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934162590804345e-05,
+      "loss": 0.5083,
+      "step": 4636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9341270192229603e-05,
+      "loss": 0.4816,
+      "step": 4637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93409143836189e-05,
+      "loss": 0.5501,
+      "step": 4638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934055848221488e-05,
+      "loss": 0.609,
+      "step": 4639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934020248802107e-05,
+      "loss": 0.4841,
+      "step": 4640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933984640104101e-05,
+      "loss": 0.4865,
+      "step": 4641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9339490221278243e-05,
+      "loss": 0.5217,
+      "step": 4642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93391339487363e-05,
+      "loss": 0.542,
+      "step": 4643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338777583418726e-05,
+      "loss": 0.6469,
+      "step": 4644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338421125329053e-05,
+      "loss": 0.4756,
+      "step": 4645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933806457447083e-05,
+      "loss": 0.5262,
+      "step": 4646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337707930847595e-05,
+      "loss": 0.5952,
+      "step": 4647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933735119446289e-05,
+      "loss": 0.5137,
+      "step": 4648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9336994365320267e-05,
+      "loss": 0.572,
+      "step": 4649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9336637443423257e-05,
+      "loss": 0.4406,
+      "step": 4650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933628042877542e-05,
+      "loss": 0.4719,
+      "step": 4651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933592332138029e-05,
+      "loss": 0.5653,
+      "step": 4652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933556612124142e-05,
+      "loss": 0.4807,
+      "step": 4653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9335208828362365e-05,
+      "loss": 0.5332,
+      "step": 4654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334851442746665e-05,
+      "loss": 0.5412,
+      "step": 4655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334493964397873e-05,
+      "loss": 0.6132,
+      "step": 4656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334136393319543e-05,
+      "loss": 0.5706,
+      "step": 4657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933377872951522e-05,
+      "loss": 0.6691,
+      "step": 4658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333420972988466e-05,
+      "loss": 0.615,
+      "step": 4659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933306312374283e-05,
+      "loss": 0.5932,
+      "step": 4660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933270518178187e-05,
+      "loss": 0.5155,
+      "step": 4661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332347147109138e-05,
+      "loss": 0.5327,
+      "step": 4662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933198901972819e-05,
+      "loss": 0.5561,
+      "step": 4663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9331630799642587e-05,
+      "loss": 0.5169,
+      "step": 4664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933127248685589e-05,
+      "loss": 0.5087,
+      "step": 4665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330914081371653e-05,
+      "loss": 0.4732,
+      "step": 4666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330555583193437e-05,
+      "loss": 0.5695,
+      "step": 4667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933019699232481e-05,
+      "loss": 0.5269,
+      "step": 4668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329838308769323e-05,
+      "loss": 0.5851,
+      "step": 4669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329479532530547e-05,
+      "loss": 0.5171,
+      "step": 4670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329120663612044e-05,
+      "loss": 0.4803,
+      "step": 4671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932876170201738e-05,
+      "loss": 0.4668,
+      "step": 4672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932840264775012e-05,
+      "loss": 0.56,
+      "step": 4673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328043500813837e-05,
+      "loss": 0.5654,
+      "step": 4674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9327684261212084e-05,
+      "loss": 0.5618,
+      "step": 4675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9327324928948446e-05,
+      "loss": 0.5473,
+      "step": 4676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932696550402648e-05,
+      "loss": 0.6043,
+      "step": 4677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9326605986449765e-05,
+      "loss": 0.4802,
+      "step": 4678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932624637622187e-05,
+      "loss": 0.4821,
+      "step": 4679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325886673346367e-05,
+      "loss": 0.5195,
+      "step": 4680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325526877826828e-05,
+      "loss": 0.5388,
+      "step": 4681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932516698966683e-05,
+      "loss": 0.5505,
+      "step": 4682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324807008869947e-05,
+      "loss": 0.5579,
+      "step": 4683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932444693543975e-05,
+      "loss": 0.5655,
+      "step": 4684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324086769379825e-05,
+      "loss": 0.5312,
+      "step": 4685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9323726510693745e-05,
+      "loss": 0.5965,
+      "step": 4686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9323366159385088e-05,
+      "loss": 0.4815,
+      "step": 4687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9323005715457436e-05,
+      "loss": 0.5489,
+      "step": 4688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932264517891437e-05,
+      "loss": 0.527,
+      "step": 4689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322284549759466e-05,
+      "loss": 0.5129,
+      "step": 4690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321923827996317e-05,
+      "loss": 0.5571,
+      "step": 4691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321563013628496e-05,
+      "loss": 0.5266,
+      "step": 4692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321202106659594e-05,
+      "loss": 0.5105,
+      "step": 4693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320841107093193e-05,
+      "loss": 0.5661,
+      "step": 4694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932048001493288e-05,
+      "loss": 0.4458,
+      "step": 4695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932011883018224e-05,
+      "loss": 0.4975,
+      "step": 4696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931975755284487e-05,
+      "loss": 0.5542,
+      "step": 4697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319396182924347e-05,
+      "loss": 0.4922,
+      "step": 4698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931903472042427e-05,
+      "loss": 0.6838,
+      "step": 4699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318673165348222e-05,
+      "loss": 0.5952,
+      "step": 4700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93183115176998e-05,
+      "loss": 0.5365,
+      "step": 4701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317949777482595e-05,
+      "loss": 0.5355,
+      "step": 4702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317587944700204e-05,
+      "loss": 0.6133,
+      "step": 4703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317226019356214e-05,
+      "loss": 0.5704,
+      "step": 4704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316864001454226e-05,
+      "loss": 0.5437,
+      "step": 4705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316501890997836e-05,
+      "loss": 0.5396,
+      "step": 4706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931613968799064e-05,
+      "loss": 0.5574,
+      "step": 4707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315777392436235e-05,
+      "loss": 0.5677,
+      "step": 4708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315415004338224e-05,
+      "loss": 0.5668,
+      "step": 4709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315052523700203e-05,
+      "loss": 0.6036,
+      "step": 4710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9314689950525777e-05,
+      "loss": 0.4769,
+      "step": 4711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931432728481854e-05,
+      "loss": 0.5234,
+      "step": 4712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313964526582104e-05,
+      "loss": 0.6192,
+      "step": 4713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313601675820067e-05,
+      "loss": 0.5248,
+      "step": 4714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931323873253604e-05,
+      "loss": 0.6002,
+      "step": 4715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312875696733616e-05,
+      "loss": 0.6018,
+      "step": 4716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312512568416416e-05,
+      "loss": 0.4944,
+      "step": 4717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312149347588035e-05,
+      "loss": 0.5476,
+      "step": 4718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931178603425209e-05,
+      "loss": 0.5688,
+      "step": 4719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311422628412188e-05,
+      "loss": 0.4915,
+      "step": 4720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311059130071938e-05,
+      "loss": 0.5619,
+      "step": 4721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310695539234948e-05,
+      "loss": 0.5465,
+      "step": 4722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310331855904833e-05,
+      "loss": 0.5856,
+      "step": 4723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930996808008521e-05,
+      "loss": 0.5042,
+      "step": 4724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309604211779684e-05,
+      "loss": 0.5066,
+      "step": 4725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930924025099188e-05,
+      "loss": 0.4901,
+      "step": 4726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93088761977254e-05,
+      "loss": 0.5264,
+      "step": 4727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9308512051983875e-05,
+      "loss": 0.5734,
+      "step": 4728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9308147813770913e-05,
+      "loss": 0.6027,
+      "step": 4729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9307783483090136e-05,
+      "loss": 0.4822,
+      "step": 4730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930741905994516e-05,
+      "loss": 0.486,
+      "step": 4731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9307054544339612e-05,
+      "loss": 0.5516,
+      "step": 4732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306689936277107e-05,
+      "loss": 0.6274,
+      "step": 4733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930632523576127e-05,
+      "loss": 0.4702,
+      "step": 4734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930596044279572e-05,
+      "loss": 0.5637,
+      "step": 4735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305595557384088e-05,
+      "loss": 0.5561,
+      "step": 4736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305230579529993e-05,
+      "loss": 0.4644,
+      "step": 4737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930486550923706e-05,
+      "loss": 0.561,
+      "step": 4738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304500346508922e-05,
+      "loss": 0.5092,
+      "step": 4739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304135091349203e-05,
+      "loss": 0.5901,
+      "step": 4740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303769743761526e-05,
+      "loss": 0.5802,
+      "step": 4741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303404303749528e-05,
+      "loss": 0.5087,
+      "step": 4742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303038771316837e-05,
+      "loss": 0.5285,
+      "step": 4743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9302673146467084e-05,
+      "loss": 0.5575,
+      "step": 4744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9302307429203902e-05,
+      "loss": 0.5324,
+      "step": 4745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930194161953092e-05,
+      "loss": 0.5662,
+      "step": 4746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301575717451777e-05,
+      "loss": 0.5291,
+      "step": 4747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301209722970107e-05,
+      "loss": 0.5802,
+      "step": 4748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930084363608954e-05,
+      "loss": 0.5397,
+      "step": 4749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300477456813725e-05,
+      "loss": 0.5863,
+      "step": 4750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300111185146286e-05,
+      "loss": 0.6045,
+      "step": 4751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929974482109087e-05,
+      "loss": 0.5489,
+      "step": 4752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9299378364651113e-05,
+      "loss": 0.5881,
+      "step": 4753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929901181583066e-05,
+      "loss": 0.5756,
+      "step": 4754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298645174633147e-05,
+      "loss": 0.4902,
+      "step": 4755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929827844106222e-05,
+      "loss": 0.5145,
+      "step": 4756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9297911615121517e-05,
+      "loss": 0.6645,
+      "step": 4757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9297544696814686e-05,
+      "loss": 0.5409,
+      "step": 4758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929717768614537e-05,
+      "loss": 0.6177,
+      "step": 4759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9296810583117222e-05,
+      "loss": 0.5999,
+      "step": 4760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929644338773388e-05,
+      "loss": 0.5443,
+      "step": 4761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9296076099998996e-05,
+      "loss": 0.5112,
+      "step": 4762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295708719916217e-05,
+      "loss": 0.5742,
+      "step": 4763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929534124748919e-05,
+      "loss": 0.6115,
+      "step": 4764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9294973682721573e-05,
+      "loss": 0.5219,
+      "step": 4765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929460602561701e-05,
+      "loss": 0.5378,
+      "step": 4766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9294238276179158e-05,
+      "loss": 0.5527,
+      "step": 4767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929387043441167e-05,
+      "loss": 0.5933,
+      "step": 4768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293502500318194e-05,
+      "loss": 0.5082,
+      "step": 4769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293134473902394e-05,
+      "loss": 0.4936,
+      "step": 4770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929276635516792e-05,
+      "loss": 0.5865,
+      "step": 4771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9292398144118434e-05,
+      "loss": 0.5772,
+      "step": 4772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929202984075759e-05,
+      "loss": 0.5524,
+      "step": 4773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291661445089044e-05,
+      "loss": 0.5113,
+      "step": 4774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929129295711646e-05,
+      "loss": 0.6107,
+      "step": 4775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9290924376843502e-05,
+      "loss": 0.5784,
+      "step": 4776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9290555704273823e-05,
+      "loss": 0.4822,
+      "step": 4777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.929018693941109e-05,
+      "loss": 0.6045,
+      "step": 4778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289818082258966e-05,
+      "loss": 0.5323,
+      "step": 4779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289449132821115e-05,
+      "loss": 0.5488,
+      "step": 4780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289080091101206e-05,
+      "loss": 0.5206,
+      "step": 4781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9288710957102898e-05,
+      "loss": 0.501,
+      "step": 4782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928834173082986e-05,
+      "loss": 0.5396,
+      "step": 4783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287972412285764e-05,
+      "loss": 0.4396,
+      "step": 4784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287603001474274e-05,
+      "loss": 0.5921,
+      "step": 4785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287233498399066e-05,
+      "loss": 0.5754,
+      "step": 4786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286863903063803e-05,
+      "loss": 0.611,
+      "step": 4787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286494215472164e-05,
+      "loss": 0.5234,
+      "step": 4788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286124435627815e-05,
+      "loss": 0.601,
+      "step": 4789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928575456353443e-05,
+      "loss": 0.5013,
+      "step": 4790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928538459919569e-05,
+      "loss": 0.6809,
+      "step": 4791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928501454261526e-05,
+      "loss": 0.5692,
+      "step": 4792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284644393796827e-05,
+      "loss": 0.5812,
+      "step": 4793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284274152744062e-05,
+      "loss": 0.566,
+      "step": 4794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283903819460644e-05,
+      "loss": 0.5195,
+      "step": 4795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283533393950253e-05,
+      "loss": 0.6036,
+      "step": 4796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283162876216564e-05,
+      "loss": 0.5774,
+      "step": 4797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9282792266263265e-05,
+      "loss": 0.5,
+      "step": 4798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928242156409403e-05,
+      "loss": 0.5528,
+      "step": 4799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928205076971255e-05,
+      "loss": 0.5461,
+      "step": 4800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92816798831225e-05,
+      "loss": 0.4502,
+      "step": 4801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9281308904327574e-05,
+      "loss": 0.5845,
+      "step": 4802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928093783333145e-05,
+      "loss": 0.5212,
+      "step": 4803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280566670137814e-05,
+      "loss": 0.5191,
+      "step": 4804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928019541475036e-05,
+      "loss": 0.5716,
+      "step": 4805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279824067172766e-05,
+      "loss": 0.5176,
+      "step": 4806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927945262740873e-05,
+      "loss": 0.642,
+      "step": 4807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279081095461936e-05,
+      "loss": 0.5499,
+      "step": 4808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278709471336078e-05,
+      "loss": 0.5136,
+      "step": 4809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278337755034848e-05,
+      "loss": 0.6129,
+      "step": 4810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9277965946561936e-05,
+      "loss": 0.5507,
+      "step": 4811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927759404592104e-05,
+      "loss": 0.6845,
+      "step": 4812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927722205311585e-05,
+      "loss": 0.5248,
+      "step": 4813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276849968150062e-05,
+      "loss": 0.4645,
+      "step": 4814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276477791027374e-05,
+      "loss": 0.6596,
+      "step": 4815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276105521751486e-05,
+      "loss": 0.5499,
+      "step": 4816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275733160326087e-05,
+      "loss": 0.6022,
+      "step": 4817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275360706754885e-05,
+      "loss": 0.4654,
+      "step": 4818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274988161041578e-05,
+      "loss": 0.5367,
+      "step": 4819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274615523189864e-05,
+      "loss": 0.5258,
+      "step": 4820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274242793203445e-05,
+      "loss": 0.5834,
+      "step": 4821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927386997108603e-05,
+      "loss": 0.5032,
+      "step": 4822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273497056841314e-05,
+      "loss": 0.6062,
+      "step": 4823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273124050473005e-05,
+      "loss": 0.6841,
+      "step": 4824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272750951984808e-05,
+      "loss": 0.5572,
+      "step": 4825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272377761380433e-05,
+      "loss": 0.5891,
+      "step": 4826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272004478663582e-05,
+      "loss": 0.5108,
+      "step": 4827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271631103837967e-05,
+      "loss": 0.4906,
+      "step": 4828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271257636907297e-05,
+      "loss": 0.6316,
+      "step": 4829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9270884077875283e-05,
+      "loss": 0.588,
+      "step": 4830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9270510426745633e-05,
+      "loss": 0.5374,
+      "step": 4831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9270136683522055e-05,
+      "loss": 0.5325,
+      "step": 4832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269762848208272e-05,
+      "loss": 0.5739,
+      "step": 4833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269388920807988e-05,
+      "loss": 0.5339,
+      "step": 4834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269014901324926e-05,
+      "loss": 0.6134,
+      "step": 4835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268640789762796e-05,
+      "loss": 0.5783,
+      "step": 4836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268266586125316e-05,
+      "loss": 0.5264,
+      "step": 4837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267892290416204e-05,
+      "loss": 0.5396,
+      "step": 4838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267517902639178e-05,
+      "loss": 0.6275,
+      "step": 4839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267143422797954e-05,
+      "loss": 0.4913,
+      "step": 4840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9266768850896257e-05,
+      "loss": 0.6265,
+      "step": 4841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9266394186937806e-05,
+      "loss": 0.5249,
+      "step": 4842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926601943092632e-05,
+      "loss": 0.5505,
+      "step": 4843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265644582865527e-05,
+      "loss": 0.5502,
+      "step": 4844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926526964275915e-05,
+      "loss": 0.5719,
+      "step": 4845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264894610610913e-05,
+      "loss": 0.5455,
+      "step": 4846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264519486424537e-05,
+      "loss": 0.4897,
+      "step": 4847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264144270203752e-05,
+      "loss": 0.5544,
+      "step": 4848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263768961952288e-05,
+      "loss": 0.5805,
+      "step": 4849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926339356167387e-05,
+      "loss": 0.5073,
+      "step": 4850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926301806937223e-05,
+      "loss": 0.591,
+      "step": 4851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262642485051095e-05,
+      "loss": 0.5744,
+      "step": 4852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262266808714197e-05,
+      "loss": 0.5055,
+      "step": 4853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261891040365272e-05,
+      "loss": 0.5165,
+      "step": 4854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261515180008047e-05,
+      "loss": 0.5099,
+      "step": 4855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926113922764626e-05,
+      "loss": 0.5555,
+      "step": 4856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926076318328364e-05,
+      "loss": 0.5829,
+      "step": 4857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926038704692393e-05,
+      "loss": 0.5895,
+      "step": 4858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926001081857086e-05,
+      "loss": 0.5915,
+      "step": 4859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9259634498228172e-05,
+      "loss": 0.5705,
+      "step": 4860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9259258085899602e-05,
+      "loss": 0.5446,
+      "step": 4861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258881581588892e-05,
+      "loss": 0.4121,
+      "step": 4862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258504985299783e-05,
+      "loss": 0.5225,
+      "step": 4863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925812829703601e-05,
+      "loss": 0.5171,
+      "step": 4864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9257751516801323e-05,
+      "loss": 0.6024,
+      "step": 4865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9257374644599457e-05,
+      "loss": 0.5394,
+      "step": 4866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925699768043416e-05,
+      "loss": 0.5355,
+      "step": 4867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925662062430918e-05,
+      "loss": 0.503,
+      "step": 4868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9256243476228257e-05,
+      "loss": 0.53,
+      "step": 4869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255866236195137e-05,
+      "loss": 0.5627,
+      "step": 4870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255488904213573e-05,
+      "loss": 0.6065,
+      "step": 4871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925511148028731e-05,
+      "loss": 0.5064,
+      "step": 4872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254733964420097e-05,
+      "loss": 0.5741,
+      "step": 4873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925435635661569e-05,
+      "loss": 0.5284,
+      "step": 4874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925397865687783e-05,
+      "loss": 0.5149,
+      "step": 4875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925360086521028e-05,
+      "loss": 0.5901,
+      "step": 4876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9253222981616783e-05,
+      "loss": 0.5495,
+      "step": 4877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.92528450061011e-05,
+      "loss": 0.4589,
+      "step": 4878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925246693866698e-05,
+      "loss": 0.5746,
+      "step": 4879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252088779318184e-05,
+      "loss": 0.5826,
+      "step": 4880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925171052805847e-05,
+      "loss": 0.6029,
+      "step": 4881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9251332184891592e-05,
+      "loss": 0.5913,
+      "step": 4882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9250953749821308e-05,
+      "loss": 0.3972,
+      "step": 4883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9250575222851373e-05,
+      "loss": 0.4814,
+      "step": 4884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9250196603985557e-05,
+      "loss": 0.5373,
+      "step": 4885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249817893227617e-05,
+      "loss": 0.6213,
+      "step": 4886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249439090581313e-05,
+      "loss": 0.5623,
+      "step": 4887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249060196050413e-05,
+      "loss": 0.5513,
+      "step": 4888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248681209638675e-05,
+      "loss": 0.5543,
+      "step": 4889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248302131349872e-05,
+      "loss": 0.5152,
+      "step": 4890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924792296118776e-05,
+      "loss": 0.496,
+      "step": 4891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247543699156112e-05,
+      "loss": 0.5216,
+      "step": 4892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247164345258692e-05,
+      "loss": 0.609,
+      "step": 4893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246784899499272e-05,
+      "loss": 0.5921,
+      "step": 4894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246405361881623e-05,
+      "loss": 0.5504,
+      "step": 4895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246025732409505e-05,
+      "loss": 0.582,
+      "step": 4896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245646011086702e-05,
+      "loss": 0.5595,
+      "step": 4897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924526619791698e-05,
+      "loss": 0.5674,
+      "step": 4898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244886292904113e-05,
+      "loss": 0.4299,
+      "step": 4899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244506296051874e-05,
+      "loss": 0.4802,
+      "step": 4900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924412620736404e-05,
+      "loss": 0.4874,
+      "step": 4901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243746026844384e-05,
+      "loss": 0.5123,
+      "step": 4902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243365754496682e-05,
+      "loss": 0.6083,
+      "step": 4903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924298539032472e-05,
+      "loss": 0.545,
+      "step": 4904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242604934332268e-05,
+      "loss": 0.5938,
+      "step": 4905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242224386523105e-05,
+      "loss": 0.514,
+      "step": 4906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.924184374690102e-05,
+      "loss": 0.5229,
+      "step": 4907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241463015469787e-05,
+      "loss": 0.5712,
+      "step": 4908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241082192233187e-05,
+      "loss": 0.5081,
+      "step": 4909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240701277195008e-05,
+      "loss": 0.7136,
+      "step": 4910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.924032027035903e-05,
+      "loss": 0.4911,
+      "step": 4911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9239939171729044e-05,
+      "loss": 0.5031,
+      "step": 4912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923955798130883e-05,
+      "loss": 0.5947,
+      "step": 4913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9239176699102175e-05,
+      "loss": 0.4716,
+      "step": 4914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.5499,
+      "step": 4915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92384138593447e-05,
+      "loss": 0.5911,
+      "step": 4916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923803230180146e-05,
+      "loss": 0.5625,
+      "step": 4917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9237650652486935e-05,
+      "loss": 0.6428,
+      "step": 4918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9237268911404916e-05,
+      "loss": 0.5828,
+      "step": 4919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236887078559202e-05,
+      "loss": 0.6426,
+      "step": 4920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923650515395358e-05,
+      "loss": 0.5262,
+      "step": 4921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236123137591846e-05,
+      "loss": 0.663,
+      "step": 4922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235741029477792e-05,
+      "loss": 0.6299,
+      "step": 4923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923535882961522e-05,
+      "loss": 0.5732,
+      "step": 4924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923497653800792e-05,
+      "loss": 0.5486,
+      "step": 4925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9234594154659696e-05,
+      "loss": 0.6394,
+      "step": 4926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923421167957434e-05,
+      "loss": 0.5701,
+      "step": 4927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923382911275566e-05,
+      "loss": 0.4888,
+      "step": 4928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233446454207446e-05,
+      "loss": 0.6019,
+      "step": 4929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923306370393351e-05,
+      "loss": 0.6425,
+      "step": 4930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923268086193765e-05,
+      "loss": 0.4964,
+      "step": 4931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9232297928223665e-05,
+      "loss": 0.5838,
+      "step": 4932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231914902795362e-05,
+      "loss": 0.6343,
+      "step": 4933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923153178565655e-05,
+      "loss": 0.5373,
+      "step": 4934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923114857681103e-05,
+      "loss": 0.5821,
+      "step": 4935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923076527626261e-05,
+      "loss": 0.6256,
+      "step": 4936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9230381884015096e-05,
+      "loss": 0.4919,
+      "step": 4937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9229998400072304e-05,
+      "loss": 0.5042,
+      "step": 4938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9229614824438033e-05,
+      "loss": 0.474,
+      "step": 4939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92292311571161e-05,
+      "loss": 0.5584,
+      "step": 4940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922884739811032e-05,
+      "loss": 0.5307,
+      "step": 4941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9228463547424495e-05,
+      "loss": 0.5713,
+      "step": 4942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9228079605062443e-05,
+      "loss": 0.6271,
+      "step": 4943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922769557102798e-05,
+      "loss": 0.5705,
+      "step": 4944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9227311445324924e-05,
+      "loss": 0.4785,
+      "step": 4945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226927227957082e-05,
+      "loss": 0.6005,
+      "step": 4946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922654291892828e-05,
+      "loss": 0.5384,
+      "step": 4947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226158518242324e-05,
+      "loss": 0.5245,
+      "step": 4948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225774025903044e-05,
+      "loss": 0.4759,
+      "step": 4949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225389441914253e-05,
+      "loss": 0.4736,
+      "step": 4950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922500476627978e-05,
+      "loss": 0.4875,
+      "step": 4951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224619999003436e-05,
+      "loss": 0.5295,
+      "step": 4952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224235140089048e-05,
+      "loss": 0.4797,
+      "step": 4953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223850189540438e-05,
+      "loss": 0.6368,
+      "step": 4954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223465147361433e-05,
+      "loss": 0.5594,
+      "step": 4955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223080013555855e-05,
+      "loss": 0.5131,
+      "step": 4956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922269478812753e-05,
+      "loss": 0.5306,
+      "step": 4957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922230947108029e-05,
+      "loss": 0.4843,
+      "step": 4958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9221924062417955e-05,
+      "loss": 0.5036,
+      "step": 4959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922153856214436e-05,
+      "loss": 0.5153,
+      "step": 4960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922115297026333e-05,
+      "loss": 0.5601,
+      "step": 4961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220767286778695e-05,
+      "loss": 0.5379,
+      "step": 4962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220381511694292e-05,
+      "loss": 0.4904,
+      "step": 4963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9219995645013948e-05,
+      "loss": 0.5404,
+      "step": 4964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92196096867415e-05,
+      "loss": 0.5695,
+      "step": 4965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921922363688078e-05,
+      "loss": 0.5907,
+      "step": 4966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218837495435623e-05,
+      "loss": 0.5308,
+      "step": 4967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218451262409867e-05,
+      "loss": 0.5709,
+      "step": 4968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218064937807347e-05,
+      "loss": 0.5277,
+      "step": 4969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92176785216319e-05,
+      "loss": 0.5341,
+      "step": 4970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9217292013887367e-05,
+      "loss": 0.57,
+      "step": 4971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921690541457759e-05,
+      "loss": 0.5657,
+      "step": 4972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216518723706398e-05,
+      "loss": 0.5888,
+      "step": 4973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216131941277643e-05,
+      "loss": 0.5975,
+      "step": 4974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215745067295168e-05,
+      "loss": 0.5579,
+      "step": 4975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921535810176281e-05,
+      "loss": 0.5064,
+      "step": 4976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214971044684416e-05,
+      "loss": 0.4383,
+      "step": 4977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214583896063833e-05,
+      "loss": 0.4897,
+      "step": 4978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214196655904905e-05,
+      "loss": 0.5177,
+      "step": 4979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9213809324211476e-05,
+      "loss": 0.4671,
+      "step": 4980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.92134219009874e-05,
+      "loss": 0.5756,
+      "step": 4981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921303438623652e-05,
+      "loss": 0.4979,
+      "step": 4982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921264677996269e-05,
+      "loss": 0.5066,
+      "step": 4983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921225908216976e-05,
+      "loss": 0.5592,
+      "step": 4984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211871292861576e-05,
+      "loss": 0.5286,
+      "step": 4985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211483412041994e-05,
+      "loss": 0.5886,
+      "step": 4986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921109543971487e-05,
+      "loss": 0.5037,
+      "step": 4987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210707375884055e-05,
+      "loss": 0.5767,
+      "step": 4988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210319220553403e-05,
+      "loss": 0.5475,
+      "step": 4989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9209930973726775e-05,
+      "loss": 0.5845,
+      "step": 4990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920954263540802e-05,
+      "loss": 0.4876,
+      "step": 4991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9209154205601005e-05,
+      "loss": 0.5817,
+      "step": 4992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208765684309582e-05,
+      "loss": 0.5212,
+      "step": 4993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208377071537612e-05,
+      "loss": 0.5623,
+      "step": 4994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207988367288956e-05,
+      "loss": 0.5063,
+      "step": 4995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207599571567478e-05,
+      "loss": 0.5244,
+      "step": 4996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207210684377036e-05,
+      "loss": 0.5966,
+      "step": 4997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9206821705721493e-05,
+      "loss": 0.6081,
+      "step": 4998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920643263560472e-05,
+      "loss": 0.4685,
+      "step": 4999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920604347403057e-05,
+      "loss": 0.5939,
+      "step": 5000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205654221002926e-05,
+      "loss": 0.6105,
+      "step": 5001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205264876525637e-05,
+      "loss": 0.6367,
+      "step": 5002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204875440602583e-05,
+      "loss": 0.5675,
+      "step": 5003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204485913237626e-05,
+      "loss": 0.463,
+      "step": 5004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920409629443464e-05,
+      "loss": 0.4876,
+      "step": 5005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203706584197492e-05,
+      "loss": 0.5455,
+      "step": 5006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920331678253006e-05,
+      "loss": 0.5291,
+      "step": 5007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202926889436206e-05,
+      "loss": 0.5184,
+      "step": 5008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920253690491981e-05,
+      "loss": 0.4905,
+      "step": 5009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202146828984745e-05,
+      "loss": 0.5568,
+      "step": 5010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201756661634883e-05,
+      "loss": 0.5481,
+      "step": 5011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201366402874107e-05,
+      "loss": 0.5356,
+      "step": 5012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200976052706292e-05,
+      "loss": 0.6115,
+      "step": 5013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200585611135304e-05,
+      "loss": 0.4974,
+      "step": 5014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920019507816504e-05,
+      "loss": 0.5303,
+      "step": 5015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9199804453799365e-05,
+      "loss": 0.5446,
+      "step": 5016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919941373804217e-05,
+      "loss": 0.5468,
+      "step": 5017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9199022930897328e-05,
+      "loss": 0.5257,
+      "step": 5018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198632032368724e-05,
+      "loss": 0.631,
+      "step": 5019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198241042460245e-05,
+      "loss": 0.5981,
+      "step": 5020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919784996117577e-05,
+      "loss": 0.5702,
+      "step": 5021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919745878851919e-05,
+      "loss": 0.5713,
+      "step": 5022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919706752449438e-05,
+      "loss": 0.4821,
+      "step": 5023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919667616910524e-05,
+      "loss": 0.5267,
+      "step": 5024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196284722355648e-05,
+      "loss": 0.5209,
+      "step": 5025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9195893184249498e-05,
+      "loss": 0.473,
+      "step": 5026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919550155479068e-05,
+      "loss": 0.4859,
+      "step": 5027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919510983398308e-05,
+      "loss": 0.4473,
+      "step": 5028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919471802183059e-05,
+      "loss": 0.6218,
+      "step": 5029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194326118337104e-05,
+      "loss": 0.5232,
+      "step": 5030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919393412350652e-05,
+      "loss": 0.551,
+      "step": 5031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193542037342725e-05,
+      "loss": 0.546,
+      "step": 5032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193149859849618e-05,
+      "loss": 0.5476,
+      "step": 5033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9192757591031092e-05,
+      "loss": 0.5384,
+      "step": 5034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9192365230891044e-05,
+      "loss": 0.5756,
+      "step": 5035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9191972779433374e-05,
+      "loss": 0.5537,
+      "step": 5036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.919158023666198e-05,
+      "loss": 0.5353,
+      "step": 5037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191187602580758e-05,
+      "loss": 0.5848,
+      "step": 5038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9190794877193615e-05,
+      "loss": 0.5608,
+      "step": 5039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.919040206050445e-05,
+      "loss": 0.5687,
+      "step": 5040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9190009152517158e-05,
+      "loss": 0.5359,
+      "step": 5041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9189616153235654e-05,
+      "loss": 0.599,
+      "step": 5042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9189223062663835e-05,
+      "loss": 0.5692,
+      "step": 5043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188829880805605e-05,
+      "loss": 0.5892,
+      "step": 5044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188436607664878e-05,
+      "loss": 0.5087,
+      "step": 5045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188043243245553e-05,
+      "loss": 0.4557,
+      "step": 5046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9187649787551536e-05,
+      "loss": 0.5609,
+      "step": 5047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918725624058674e-05,
+      "loss": 0.6145,
+      "step": 5048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186862602355077e-05,
+      "loss": 0.5386,
+      "step": 5049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186468872860455e-05,
+      "loss": 0.5383,
+      "step": 5050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918607505210678e-05,
+      "loss": 0.5781,
+      "step": 5051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9185681140097975e-05,
+      "loss": 0.4225,
+      "step": 5052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918528713683794e-05,
+      "loss": 0.5718,
+      "step": 5053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9184893042330604e-05,
+      "loss": 0.4968,
+      "step": 5054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918449885657987e-05,
+      "loss": 0.645,
+      "step": 5055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918410457958966e-05,
+      "loss": 0.455,
+      "step": 5056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183710211363883e-05,
+      "loss": 0.7101,
+      "step": 5057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918331575190647e-05,
+      "loss": 0.5319,
+      "step": 5058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918292120122133e-05,
+      "loss": 0.5125,
+      "step": 5059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918252655931238e-05,
+      "loss": 0.5569,
+      "step": 5060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918213182618355e-05,
+      "loss": 0.5099,
+      "step": 5061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181737001838754e-05,
+      "loss": 0.6267,
+      "step": 5062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181342086281917e-05,
+      "loss": 0.5212,
+      "step": 5063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918094707951696e-05,
+      "loss": 0.5455,
+      "step": 5064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9180551981547813e-05,
+      "loss": 0.554,
+      "step": 5065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9180156792378396e-05,
+      "loss": 0.453,
+      "step": 5066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917976151201263e-05,
+      "loss": 0.486,
+      "step": 5067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917936614045445e-05,
+      "loss": 0.5886,
+      "step": 5068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9178970677707783e-05,
+      "loss": 0.6206,
+      "step": 5069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9178575123776553e-05,
+      "loss": 0.5537,
+      "step": 5070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9178179478664693e-05,
+      "loss": 0.5698,
+      "step": 5071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9177783742376133e-05,
+      "loss": 0.5515,
+      "step": 5072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91773879149148e-05,
+      "loss": 0.6052,
+      "step": 5073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917699199628463e-05,
+      "loss": 0.5484,
+      "step": 5074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176595986489558e-05,
+      "loss": 0.6094,
+      "step": 5075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176199885533514e-05,
+      "loss": 0.4945,
+      "step": 5076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9175803693420438e-05,
+      "loss": 0.5626,
+      "step": 5077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917540741015426e-05,
+      "loss": 0.5833,
+      "step": 5078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917501103573892e-05,
+      "loss": 0.5469,
+      "step": 5079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174614570178354e-05,
+      "loss": 0.5573,
+      "step": 5080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174218013476504e-05,
+      "loss": 0.5515,
+      "step": 5081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91738213656373e-05,
+      "loss": 0.5358,
+      "step": 5082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173424626664692e-05,
+      "loss": 0.5112,
+      "step": 5083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917302779656262e-05,
+      "loss": 0.5296,
+      "step": 5084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9172630875335024e-05,
+      "loss": 0.5598,
+      "step": 5085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9172233862985847e-05,
+      "loss": 0.4815,
+      "step": 5086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917183675951903e-05,
+      "loss": 0.5275,
+      "step": 5087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171439564938525e-05,
+      "loss": 0.5463,
+      "step": 5088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171042279248275e-05,
+      "loss": 0.562,
+      "step": 5089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170644902452225e-05,
+      "loss": 0.5804,
+      "step": 5090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170247434554323e-05,
+      "loss": 0.6402,
+      "step": 5091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169849875558518e-05,
+      "loss": 0.5982,
+      "step": 5092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169452225468755e-05,
+      "loss": 0.587,
+      "step": 5093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916905448428899e-05,
+      "loss": 0.503,
+      "step": 5094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9168656652023176e-05,
+      "loss": 0.5333,
+      "step": 5095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9168258728675257e-05,
+      "loss": 0.5519,
+      "step": 5096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167860714249197e-05,
+      "loss": 0.5963,
+      "step": 5097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167462608748937e-05,
+      "loss": 0.541,
+      "step": 5098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167064412178443e-05,
+      "loss": 0.5381,
+      "step": 5099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9166666124541664e-05,
+      "loss": 0.5578,
+      "step": 5100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916626774584256e-05,
+      "loss": 0.4749,
+      "step": 5101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165869276085084e-05,
+      "loss": 0.6423,
+      "step": 5102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165470715273197e-05,
+      "loss": 0.5556,
+      "step": 5103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165072063410862e-05,
+      "loss": 0.594,
+      "step": 5104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916467332050204e-05,
+      "loss": 0.6146,
+      "step": 5105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916427448655068e-05,
+      "loss": 0.5035,
+      "step": 5106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916387556156076e-05,
+      "loss": 0.5798,
+      "step": 5107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916347654553623e-05,
+      "loss": 0.5675,
+      "step": 5108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9163077438481062e-05,
+      "loss": 0.4808,
+      "step": 5109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9162678240399218e-05,
+      "loss": 0.5337,
+      "step": 5110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916227895129466e-05,
+      "loss": 0.5394,
+      "step": 5111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916187957117136e-05,
+      "loss": 0.6113,
+      "step": 5112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9161480100033285e-05,
+      "loss": 0.6413,
+      "step": 5113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91610805378844e-05,
+      "loss": 0.5657,
+      "step": 5114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916068088472868e-05,
+      "loss": 0.5798,
+      "step": 5115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9160281140570086e-05,
+      "loss": 0.5561,
+      "step": 5116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91598813054126e-05,
+      "loss": 0.4911,
+      "step": 5117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159481379260184e-05,
+      "loss": 0.5255,
+      "step": 5118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915908136211682e-05,
+      "loss": 0.5216,
+      "step": 5119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158681253986476e-05,
+      "loss": 0.6746,
+      "step": 5120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158281054873126e-05,
+      "loss": 0.5641,
+      "step": 5121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157880764780746e-05,
+      "loss": 0.6061,
+      "step": 5122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157480383713317e-05,
+      "loss": 0.593,
+      "step": 5123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157079911674815e-05,
+      "loss": 0.5108,
+      "step": 5124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9156679348669215e-05,
+      "loss": 0.591,
+      "step": 5125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9156278694700497e-05,
+      "loss": 0.4966,
+      "step": 5126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155877949772644e-05,
+      "loss": 0.5403,
+      "step": 5127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155477113889635e-05,
+      "loss": 0.5033,
+      "step": 5128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915507618705545e-05,
+      "loss": 0.4874,
+      "step": 5129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154675169274076e-05,
+      "loss": 0.4914,
+      "step": 5130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154274060549492e-05,
+      "loss": 0.5372,
+      "step": 5131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915387286088569e-05,
+      "loss": 0.604,
+      "step": 5132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9153471570286646e-05,
+      "loss": 0.6038,
+      "step": 5133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915307018875635e-05,
+      "loss": 0.5364,
+      "step": 5134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91526687162988e-05,
+      "loss": 0.5316,
+      "step": 5135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9152267152917968e-05,
+      "loss": 0.5184,
+      "step": 5136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915186549861785e-05,
+      "loss": 0.504,
+      "step": 5137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915146375340244e-05,
+      "loss": 0.5175,
+      "step": 5138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151061917275723e-05,
+      "loss": 0.4085,
+      "step": 5139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9150659990241693e-05,
+      "loss": 0.6276,
+      "step": 5140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915025797230434e-05,
+      "loss": 0.5726,
+      "step": 5141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149855863467665e-05,
+      "loss": 0.584,
+      "step": 5142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149453663735653e-05,
+      "loss": 0.498,
+      "step": 5143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914905137311231e-05,
+      "loss": 0.5478,
+      "step": 5144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9148648991601623e-05,
+      "loss": 0.5697,
+      "step": 5145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9148246519207595e-05,
+      "loss": 0.5048,
+      "step": 5146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9147843955934222e-05,
+      "loss": 0.5398,
+      "step": 5147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91474413017855e-05,
+      "loss": 0.5191,
+      "step": 5148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9147038556765438e-05,
+      "loss": 0.4859,
+      "step": 5149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914663572087803e-05,
+      "loss": 0.524,
+      "step": 5150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146232794127277e-05,
+      "loss": 0.5136,
+      "step": 5151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145829776517185e-05,
+      "loss": 0.4858,
+      "step": 5152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145426668051757e-05,
+      "loss": 0.5751,
+      "step": 5153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145023468734996e-05,
+      "loss": 0.5768,
+      "step": 5154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914462017857091e-05,
+      "loss": 0.5585,
+      "step": 5155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9144216797563505e-05,
+      "loss": 0.5224,
+      "step": 5156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914381332571678e-05,
+      "loss": 0.6188,
+      "step": 5157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914340976303476e-05,
+      "loss": 0.5108,
+      "step": 5158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143006109521437e-05,
+      "loss": 0.4813,
+      "step": 5159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9142602365180827e-05,
+      "loss": 0.5839,
+      "step": 5160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9142198530016944e-05,
+      "loss": 0.6181,
+      "step": 5161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141794604033798e-05,
+      "loss": 0.5549,
+      "step": 5162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91413905872354e-05,
+      "loss": 0.5036,
+      "step": 5163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9140986479625766e-05,
+      "loss": 0.5307,
+      "step": 5164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914058228120891e-05,
+      "loss": 0.5808,
+      "step": 5165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9140177991988847e-05,
+      "loss": 0.568,
+      "step": 5166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139773611969594e-05,
+      "loss": 0.4837,
+      "step": 5167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913936914115516e-05,
+      "loss": 0.5568,
+      "step": 5168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138964579549574e-05,
+      "loss": 0.5118,
+      "step": 5169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913855992715685e-05,
+      "loss": 0.5894,
+      "step": 5170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913815518398101e-05,
+      "loss": 0.57,
+      "step": 5171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9137750350026074e-05,
+      "loss": 0.5414,
+      "step": 5172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913734542529606e-05,
+      "loss": 0.5128,
+      "step": 5173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136940409794996e-05,
+      "loss": 0.5118,
+      "step": 5174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136535303526906e-05,
+      "loss": 0.5752,
+      "step": 5175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136130106495806e-05,
+      "loss": 0.5134,
+      "step": 5176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913572481870573e-05,
+      "loss": 0.481,
+      "step": 5177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9135319440160696e-05,
+      "loss": 0.503,
+      "step": 5178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913491397086474e-05,
+      "loss": 0.5536,
+      "step": 5179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913450841082189e-05,
+      "loss": 0.5528,
+      "step": 5180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134102760036164e-05,
+      "loss": 0.474,
+      "step": 5181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9133697018511598e-05,
+      "loss": 0.5333,
+      "step": 5182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913329118625223e-05,
+      "loss": 0.5497,
+      "step": 5183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913288526326208e-05,
+      "loss": 0.5836,
+      "step": 5184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132479249545186e-05,
+      "loss": 0.538,
+      "step": 5185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913207314510558e-05,
+      "loss": 0.5235,
+      "step": 5186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9131666949947297e-05,
+      "loss": 0.5379,
+      "step": 5187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913126066407437e-05,
+      "loss": 0.5294,
+      "step": 5188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913085428749084e-05,
+      "loss": 0.5667,
+      "step": 5189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913044782020074e-05,
+      "loss": 0.5148,
+      "step": 5190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913004126220811e-05,
+      "loss": 0.5709,
+      "step": 5191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912963461351699e-05,
+      "loss": 0.5404,
+      "step": 5192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9129227874131412e-05,
+      "loss": 0.5162,
+      "step": 5193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128821044055423e-05,
+      "loss": 0.6183,
+      "step": 5194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128414123293065e-05,
+      "loss": 0.546,
+      "step": 5195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912800711184838e-05,
+      "loss": 0.6812,
+      "step": 5196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9127600009725408e-05,
+      "loss": 0.5825,
+      "step": 5197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9127192816928194e-05,
+      "loss": 0.5701,
+      "step": 5198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912678553346079e-05,
+      "loss": 0.7129,
+      "step": 5199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912637815932723e-05,
+      "loss": 0.598,
+      "step": 5200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912597069453157e-05,
+      "loss": 0.5584,
+      "step": 5201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9125563139077857e-05,
+      "loss": 0.5355,
+      "step": 5202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9125155492970134e-05,
+      "loss": 0.4783,
+      "step": 5203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9124747756212457e-05,
+      "loss": 0.5673,
+      "step": 5204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912433992880887e-05,
+      "loss": 0.5699,
+      "step": 5205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123932010763437e-05,
+      "loss": 0.5785,
+      "step": 5206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912352400208019e-05,
+      "loss": 0.5504,
+      "step": 5207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123115902763203e-05,
+      "loss": 0.4813,
+      "step": 5208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9122707712816516e-05,
+      "loss": 0.4922,
+      "step": 5209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9122299432244188e-05,
+      "loss": 0.5598,
+      "step": 5210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912189106105028e-05,
+      "loss": 0.4948,
+      "step": 5211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912148259923884e-05,
+      "loss": 0.5125,
+      "step": 5212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9121074046813933e-05,
+      "loss": 0.5223,
+      "step": 5213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120665403779612e-05,
+      "loss": 0.6404,
+      "step": 5214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120256670139942e-05,
+      "loss": 0.5553,
+      "step": 5215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119847845898978e-05,
+      "loss": 0.5737,
+      "step": 5216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119438931060787e-05,
+      "loss": 0.5177,
+      "step": 5217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119029925629423e-05,
+      "loss": 0.5075,
+      "step": 5218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911862082960896e-05,
+      "loss": 0.5314,
+      "step": 5219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9118211643003452e-05,
+      "loss": 0.5505,
+      "step": 5220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911780236581697e-05,
+      "loss": 0.5006,
+      "step": 5221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117392998053574e-05,
+      "loss": 0.5671,
+      "step": 5222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911698353971734e-05,
+      "loss": 0.5334,
+      "step": 5223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116573990812327e-05,
+      "loss": 0.6258,
+      "step": 5224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911616435134261e-05,
+      "loss": 0.5839,
+      "step": 5225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911575462131225e-05,
+      "loss": 0.5759,
+      "step": 5226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115344800725324e-05,
+      "loss": 0.5465,
+      "step": 5227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114934889585902e-05,
+      "loss": 0.5689,
+      "step": 5228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114524887898052e-05,
+      "loss": 0.5001,
+      "step": 5229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911411479566586e-05,
+      "loss": 0.6139,
+      "step": 5230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9113704612893383e-05,
+      "loss": 0.5085,
+      "step": 5231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113294339584706e-05,
+      "loss": 0.5435,
+      "step": 5232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112883975743898e-05,
+      "loss": 0.6512,
+      "step": 5233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911247352137504e-05,
+      "loss": 0.5789,
+      "step": 5234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112062976482213e-05,
+      "loss": 0.4907,
+      "step": 5235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9111652341069486e-05,
+      "loss": 0.5877,
+      "step": 5236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9111241615140947e-05,
+      "loss": 0.4542,
+      "step": 5237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110830798700673e-05,
+      "loss": 0.5322,
+      "step": 5238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110419891752744e-05,
+      "loss": 0.5063,
+      "step": 5239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110008894301245e-05,
+      "loss": 0.467,
+      "step": 5240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9109597806350252e-05,
+      "loss": 0.55,
+      "step": 5241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910918662790386e-05,
+      "loss": 0.5466,
+      "step": 5242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910877535896614e-05,
+      "loss": 0.4567,
+      "step": 5243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910836399954119e-05,
+      "loss": 0.5516,
+      "step": 5244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910795254963309e-05,
+      "loss": 0.5884,
+      "step": 5245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9107541009245925e-05,
+      "loss": 0.5588,
+      "step": 5246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910712937838379e-05,
+      "loss": 0.5583,
+      "step": 5247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106717657050773e-05,
+      "loss": 0.5151,
+      "step": 5248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910630584525096e-05,
+      "loss": 0.6401,
+      "step": 5249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910589394298844e-05,
+      "loss": 0.5917,
+      "step": 5250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105481950267317e-05,
+      "loss": 0.5373,
+      "step": 5251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105069867091673e-05,
+      "loss": 0.587,
+      "step": 5252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104657693465602e-05,
+      "loss": 0.5719,
+      "step": 5253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91042454293932e-05,
+      "loss": 0.4994,
+      "step": 5254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103833074878565e-05,
+      "loss": 0.5105,
+      "step": 5255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103420629925794e-05,
+      "loss": 0.5037,
+      "step": 5256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910300809453898e-05,
+      "loss": 0.5051,
+      "step": 5257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102595468722224e-05,
+      "loss": 0.5029,
+      "step": 5258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102182752479624e-05,
+      "loss": 0.5917,
+      "step": 5259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101769945815277e-05,
+      "loss": 0.6111,
+      "step": 5260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910135704873329e-05,
+      "loss": 0.5341,
+      "step": 5261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9100944061237764e-05,
+      "loss": 0.5293,
+      "step": 5262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9100530983332793e-05,
+      "loss": 0.5475,
+      "step": 5263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910011781502249e-05,
+      "loss": 0.5341,
+      "step": 5264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099704556310962e-05,
+      "loss": 0.5047,
+      "step": 5265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099291207202305e-05,
+      "loss": 0.5591,
+      "step": 5266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098877767700626e-05,
+      "loss": 0.4597,
+      "step": 5267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098464237810038e-05,
+      "loss": 0.608,
+      "step": 5268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909805061753464e-05,
+      "loss": 0.4988,
+      "step": 5269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097636906878556e-05,
+      "loss": 0.5986,
+      "step": 5270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097223105845885e-05,
+      "loss": 0.5264,
+      "step": 5271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9096809214440737e-05,
+      "loss": 0.4345,
+      "step": 5272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909639523266723e-05,
+      "loss": 0.52,
+      "step": 5273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9095981160529467e-05,
+      "loss": 0.5647,
+      "step": 5274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909556699803157e-05,
+      "loss": 0.5377,
+      "step": 5275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909515274517765e-05,
+      "loss": 0.4863,
+      "step": 5276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094738401971824e-05,
+      "loss": 0.5548,
+      "step": 5277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094323968418207e-05,
+      "loss": 0.5711,
+      "step": 5278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9093909444520914e-05,
+      "loss": 0.6493,
+      "step": 5279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909349483028407e-05,
+      "loss": 0.5422,
+      "step": 5280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9093080125711782e-05,
+      "loss": 0.5318,
+      "step": 5281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909266533080818e-05,
+      "loss": 0.5279,
+      "step": 5282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909225044557738e-05,
+      "loss": 0.5741,
+      "step": 5283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091835470023507e-05,
+      "loss": 0.5544,
+      "step": 5284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091420404150674e-05,
+      "loss": 0.5288,
+      "step": 5285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091005247963018e-05,
+      "loss": 0.6104,
+      "step": 5286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909059000146465e-05,
+      "loss": 0.481,
+      "step": 5287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9090174664659708e-05,
+      "loss": 0.5438,
+      "step": 5288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908975923755231e-05,
+      "loss": 0.5557,
+      "step": 5289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9089343720146583e-05,
+      "loss": 0.5813,
+      "step": 5290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088928112446656e-05,
+      "loss": 0.5295,
+      "step": 5291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088512414456658e-05,
+      "loss": 0.6596,
+      "step": 5292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088096626180718e-05,
+      "loss": 0.5113,
+      "step": 5293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087680747622967e-05,
+      "loss": 0.5572,
+      "step": 5294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087264778787534e-05,
+      "loss": 0.5059,
+      "step": 5295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9086848719678558e-05,
+      "loss": 0.6135,
+      "step": 5296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9086432570300163e-05,
+      "loss": 0.6011,
+      "step": 5297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908601633065649e-05,
+      "loss": 0.478,
+      "step": 5298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085600000751673e-05,
+      "loss": 0.5334,
+      "step": 5299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085183580589845e-05,
+      "loss": 0.4786,
+      "step": 5300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084767070175142e-05,
+      "loss": 0.5581,
+      "step": 5301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084350469511704e-05,
+      "loss": 0.5738,
+      "step": 5302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9083933778603672e-05,
+      "loss": 0.5002,
+      "step": 5303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9083516997455182e-05,
+      "loss": 0.5682,
+      "step": 5304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9083100126070375e-05,
+      "loss": 0.6053,
+      "step": 5305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908268316445339e-05,
+      "loss": 0.5254,
+      "step": 5306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082266112608377e-05,
+      "loss": 0.5367,
+      "step": 5307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081848970539467e-05,
+      "loss": 0.618,
+      "step": 5308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081431738250815e-05,
+      "loss": 0.5375,
+      "step": 5309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908101441574656e-05,
+      "loss": 0.6322,
+      "step": 5310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080597003030847e-05,
+      "loss": 0.499,
+      "step": 5311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908017950010783e-05,
+      "loss": 0.5094,
+      "step": 5312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9079761906981646e-05,
+      "loss": 0.5876,
+      "step": 5313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907934422365645e-05,
+      "loss": 0.5401,
+      "step": 5314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907892645013639e-05,
+      "loss": 0.6443,
+      "step": 5315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078508586425615e-05,
+      "loss": 0.5293,
+      "step": 5316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907809063252828e-05,
+      "loss": 0.5022,
+      "step": 5317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907767258844853e-05,
+      "loss": 0.5817,
+      "step": 5318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9077254454190524e-05,
+      "loss": 0.6128,
+      "step": 5319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076836229758414e-05,
+      "loss": 0.5328,
+      "step": 5320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076417915156354e-05,
+      "loss": 0.5512,
+      "step": 5321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.90759995103885e-05,
+      "loss": 0.4919,
+      "step": 5322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9075581015459007e-05,
+      "loss": 0.5271,
+      "step": 5323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9075162430372032e-05,
+      "loss": 0.4898,
+      "step": 5324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074743755131743e-05,
+      "loss": 0.5252,
+      "step": 5325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907432498974228e-05,
+      "loss": 0.5363,
+      "step": 5326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073906134207824e-05,
+      "loss": 0.5119,
+      "step": 5327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907348718853252e-05,
+      "loss": 0.5139,
+      "step": 5328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907306815272054e-05,
+      "loss": 0.5757,
+      "step": 5329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907264902677604e-05,
+      "loss": 0.6359,
+      "step": 5330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907222981070319e-05,
+      "loss": 0.4939,
+      "step": 5331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9071810504506145e-05,
+      "loss": 0.4693,
+      "step": 5332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9071391108189083e-05,
+      "loss": 0.4557,
+      "step": 5333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907097162175616e-05,
+      "loss": 0.4875,
+      "step": 5334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070552045211544e-05,
+      "loss": 0.5854,
+      "step": 5335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907013237855941e-05,
+      "loss": 0.5017,
+      "step": 5336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9069712621803923e-05,
+      "loss": 0.5348,
+      "step": 5337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9069292774949253e-05,
+      "loss": 0.548,
+      "step": 5338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9068872837999568e-05,
+      "loss": 0.5482,
+      "step": 5339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906845281095904e-05,
+      "loss": 0.6457,
+      "step": 5340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9068032693831852e-05,
+      "loss": 0.5563,
+      "step": 5341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9067612486622163e-05,
+      "loss": 0.4691,
+      "step": 5342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9067192189334155e-05,
+      "loss": 0.6488,
+      "step": 5343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066771801972003e-05,
+      "loss": 0.5023,
+      "step": 5344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906635132453988e-05,
+      "loss": 0.5822,
+      "step": 5345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065930757041967e-05,
+      "loss": 0.5001,
+      "step": 5346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906551009948244e-05,
+      "loss": 0.5236,
+      "step": 5347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906508935186548e-05,
+      "loss": 0.5349,
+      "step": 5348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906466851419526e-05,
+      "loss": 0.5856,
+      "step": 5349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9064247586475968e-05,
+      "loss": 0.6061,
+      "step": 5350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906382656871178e-05,
+      "loss": 0.5653,
+      "step": 5351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906340546090689e-05,
+      "loss": 0.6345,
+      "step": 5352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062984263065465e-05,
+      "loss": 0.4796,
+      "step": 5353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062562975191697e-05,
+      "loss": 0.5872,
+      "step": 5354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062141597289773e-05,
+      "loss": 0.469,
+      "step": 5355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9061720129363878e-05,
+      "loss": 0.58,
+      "step": 5356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9061298571418195e-05,
+      "loss": 0.605,
+      "step": 5357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060876923456915e-05,
+      "loss": 0.4777,
+      "step": 5358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906045518548423e-05,
+      "loss": 0.5445,
+      "step": 5359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906003335750432e-05,
+      "loss": 0.4894,
+      "step": 5360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9059611439521384e-05,
+      "loss": 0.6326,
+      "step": 5361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9059189431539612e-05,
+      "loss": 0.5677,
+      "step": 5362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058767333563192e-05,
+      "loss": 0.5983,
+      "step": 5363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905834514559632e-05,
+      "loss": 0.5211,
+      "step": 5364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9057922867643196e-05,
+      "loss": 0.5938,
+      "step": 5365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9057500499708004e-05,
+      "loss": 0.504,
+      "step": 5366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9057078041794944e-05,
+      "loss": 0.5663,
+      "step": 5367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905665549390822e-05,
+      "loss": 0.5359,
+      "step": 5368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056232856052015e-05,
+      "loss": 0.5094,
+      "step": 5369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905581012823054e-05,
+      "loss": 0.4584,
+      "step": 5370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9055387310447992e-05,
+      "loss": 0.6186,
+      "step": 5371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905496440270857e-05,
+      "loss": 0.6464,
+      "step": 5372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905454140501647e-05,
+      "loss": 0.5275,
+      "step": 5373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9054118317375897e-05,
+      "loss": 0.5759,
+      "step": 5374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053695139791058e-05,
+      "loss": 0.5388,
+      "step": 5375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905327187226616e-05,
+      "loss": 0.4285,
+      "step": 5376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052848514805398e-05,
+      "loss": 0.4535,
+      "step": 5377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052425067412983e-05,
+      "loss": 0.5718,
+      "step": 5378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905200153009312e-05,
+      "loss": 0.4388,
+      "step": 5379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051577902850016e-05,
+      "loss": 0.5477,
+      "step": 5380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905115418568788e-05,
+      "loss": 0.5082,
+      "step": 5381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9050730378610923e-05,
+      "loss": 0.5533,
+      "step": 5382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9050306481623357e-05,
+      "loss": 0.5294,
+      "step": 5383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049882494729386e-05,
+      "loss": 0.5308,
+      "step": 5384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049458417933225e-05,
+      "loss": 0.5196,
+      "step": 5385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049034251239093e-05,
+      "loss": 0.5264,
+      "step": 5386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048609994651194e-05,
+      "loss": 0.6022,
+      "step": 5387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048185648173746e-05,
+      "loss": 0.491,
+      "step": 5388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047761211810967e-05,
+      "loss": 0.5496,
+      "step": 5389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047336685567073e-05,
+      "loss": 0.6534,
+      "step": 5390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046912069446275e-05,
+      "loss": 0.4197,
+      "step": 5391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90464873634528e-05,
+      "loss": 0.6046,
+      "step": 5392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046062567590866e-05,
+      "loss": 0.5799,
+      "step": 5393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9045637681864684e-05,
+      "loss": 0.5002,
+      "step": 5394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9045212706278485e-05,
+      "loss": 0.4945,
+      "step": 5395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9044787640836483e-05,
+      "loss": 0.5502,
+      "step": 5396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904436248554291e-05,
+      "loss": 0.5199,
+      "step": 5397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904393724040198e-05,
+      "loss": 0.5104,
+      "step": 5398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904351190541792e-05,
+      "loss": 0.519,
+      "step": 5399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904308648059496e-05,
+      "loss": 0.5969,
+      "step": 5400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042660965937322e-05,
+      "loss": 0.5224,
+      "step": 5401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042235361449234e-05,
+      "loss": 0.4611,
+      "step": 5402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9041809667134922e-05,
+      "loss": 0.6015,
+      "step": 5403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9041383882998617e-05,
+      "loss": 0.4902,
+      "step": 5404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904095800904455e-05,
+      "loss": 0.6214,
+      "step": 5405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9040532045276953e-05,
+      "loss": 0.5582,
+      "step": 5406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904010599170005e-05,
+      "loss": 0.4879,
+      "step": 5407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903967984831808e-05,
+      "loss": 0.6029,
+      "step": 5408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9039253615135276e-05,
+      "loss": 0.532,
+      "step": 5409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9038827292155872e-05,
+      "loss": 0.6005,
+      "step": 5410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9038400879384097e-05,
+      "loss": 0.5258,
+      "step": 5411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9037974376824194e-05,
+      "loss": 0.4802,
+      "step": 5412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9037547784480398e-05,
+      "loss": 0.5455,
+      "step": 5413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9037121102356947e-05,
+      "loss": 0.586,
+      "step": 5414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9036694330458077e-05,
+      "loss": 0.4932,
+      "step": 5415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9036267468788032e-05,
+      "loss": 0.6195,
+      "step": 5416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9035840517351052e-05,
+      "loss": 0.5526,
+      "step": 5417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9035413476151375e-05,
+      "loss": 0.5388,
+      "step": 5418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9034986345193243e-05,
+      "loss": 0.5243,
+      "step": 5419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9034559124480904e-05,
+      "loss": 0.5837,
+      "step": 5420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90341318140186e-05,
+      "loss": 0.5357,
+      "step": 5421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9033704413810572e-05,
+      "loss": 0.5287,
+      "step": 5422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903327692386107e-05,
+      "loss": 0.5022,
+      "step": 5423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903284934417434e-05,
+      "loss": 0.5781,
+      "step": 5424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903242167475463e-05,
+      "loss": 0.5188,
+      "step": 5425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9031993915606188e-05,
+      "loss": 0.5285,
+      "step": 5426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.903156606673326e-05,
+      "loss": 0.5705,
+      "step": 5427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031138128140102e-05,
+      "loss": 0.5063,
+      "step": 5428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9030710099830966e-05,
+      "loss": 0.5226,
+      "step": 5429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9030281981810093e-05,
+      "loss": 0.5216,
+      "step": 5430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029853774081747e-05,
+      "loss": 0.4874,
+      "step": 5431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029425476650182e-05,
+      "loss": 0.5891,
+      "step": 5432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9028997089519645e-05,
+      "loss": 0.551,
+      "step": 5433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.90285686126944e-05,
+      "loss": 0.5862,
+      "step": 5434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9028140046178696e-05,
+      "loss": 0.6112,
+      "step": 5435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9027711389976792e-05,
+      "loss": 0.5208,
+      "step": 5436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902728264409295e-05,
+      "loss": 0.5887,
+      "step": 5437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026853808531428e-05,
+      "loss": 0.4662,
+      "step": 5438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026424883296485e-05,
+      "loss": 0.5585,
+      "step": 5439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9025995868392387e-05,
+      "loss": 0.5128,
+      "step": 5440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9025566763823385e-05,
+      "loss": 0.5257,
+      "step": 5441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902513756959375e-05,
+      "loss": 0.4751,
+      "step": 5442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9024708285707743e-05,
+      "loss": 0.487,
+      "step": 5443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902427891216963e-05,
+      "loss": 0.579,
+      "step": 5444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9023849448983674e-05,
+      "loss": 0.5558,
+      "step": 5445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9023419896154146e-05,
+      "loss": 0.4835,
+      "step": 5446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9022990253685307e-05,
+      "loss": 0.635,
+      "step": 5447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902256052158143e-05,
+      "loss": 0.522,
+      "step": 5448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902213069984678e-05,
+      "loss": 0.4876,
+      "step": 5449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902170078848563e-05,
+      "loss": 0.4889,
+      "step": 5450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021270787502254e-05,
+      "loss": 0.5578,
+      "step": 5451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9020840696900913e-05,
+      "loss": 0.4904,
+      "step": 5452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902041051668589e-05,
+      "loss": 0.6541,
+      "step": 5453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901998024686145e-05,
+      "loss": 0.523,
+      "step": 5454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901954988743188e-05,
+      "loss": 0.5657,
+      "step": 5455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019119438401442e-05,
+      "loss": 0.5203,
+      "step": 5456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018688899774414e-05,
+      "loss": 0.5289,
+      "step": 5457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018258271555082e-05,
+      "loss": 0.4919,
+      "step": 5458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9017827553747718e-05,
+      "loss": 0.5418,
+      "step": 5459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9017396746356594e-05,
+      "loss": 0.5568,
+      "step": 5460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016965849386003e-05,
+      "loss": 0.5303,
+      "step": 5461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016534862840218e-05,
+      "loss": 0.4773,
+      "step": 5462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901610378672352e-05,
+      "loss": 0.5765,
+      "step": 5463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9015672621040195e-05,
+      "loss": 0.5299,
+      "step": 5464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901524136579452e-05,
+      "loss": 0.4755,
+      "step": 5465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9014810020990785e-05,
+      "loss": 0.6832,
+      "step": 5466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9014378586633276e-05,
+      "loss": 0.5376,
+      "step": 5467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013947062726276e-05,
+      "loss": 0.5412,
+      "step": 5468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013515449274066e-05,
+      "loss": 0.448,
+      "step": 5469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013083746280946e-05,
+      "loss": 0.4704,
+      "step": 5470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012651953751192e-05,
+      "loss": 0.5264,
+      "step": 5471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012220071689107e-05,
+      "loss": 0.5879,
+      "step": 5472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901178810009897e-05,
+      "loss": 0.4809,
+      "step": 5473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9011356038985076e-05,
+      "loss": 0.5532,
+      "step": 5474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010923888351717e-05,
+      "loss": 0.5573,
+      "step": 5475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901049164820319e-05,
+      "loss": 0.5496,
+      "step": 5476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901005931854378e-05,
+      "loss": 0.5945,
+      "step": 5477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900962689937779e-05,
+      "loss": 0.5189,
+      "step": 5478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009194390709513e-05,
+      "loss": 0.5148,
+      "step": 5479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9008761792543243e-05,
+      "loss": 0.5207,
+      "step": 5480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9008329104883284e-05,
+      "loss": 0.5425,
+      "step": 5481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007896327733924e-05,
+      "loss": 0.5796,
+      "step": 5482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007463461099474e-05,
+      "loss": 0.5565,
+      "step": 5483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007030504984226e-05,
+      "loss": 0.4944,
+      "step": 5484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006597459392483e-05,
+      "loss": 0.5537,
+      "step": 5485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900616432432855e-05,
+      "loss": 0.5945,
+      "step": 5486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9005731099796725e-05,
+      "loss": 0.4849,
+      "step": 5487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9005297785801316e-05,
+      "loss": 0.5123,
+      "step": 5488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9004864382346623e-05,
+      "loss": 0.4907,
+      "step": 5489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9004430889436954e-05,
+      "loss": 0.5712,
+      "step": 5490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9003997307076614e-05,
+      "loss": 0.5772,
+      "step": 5491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9003563635269917e-05,
+      "loss": 0.5378,
+      "step": 5492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9003129874021162e-05,
+      "loss": 0.5036,
+      "step": 5493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002696023334663e-05,
+      "loss": 0.5588,
+      "step": 5494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002262083214728e-05,
+      "loss": 0.6374,
+      "step": 5495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9001828053665664e-05,
+      "loss": 0.5368,
+      "step": 5496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9001393934691792e-05,
+      "loss": 0.5455,
+      "step": 5497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000959726297417e-05,
+      "loss": 0.5714,
+      "step": 5498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000525428486857e-05,
+      "loss": 0.5616,
+      "step": 5499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000091041264426e-05,
+      "loss": 0.5382,
+      "step": 5500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8999656564634436e-05,
+      "loss": 0.5553,
+      "step": 5501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8999221998601202e-05,
+      "loss": 0.5214,
+      "step": 5502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8998787343169045e-05,
+      "loss": 0.5078,
+      "step": 5503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8998352598342285e-05,
+      "loss": 0.5698,
+      "step": 5504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997917764125233e-05,
+      "loss": 0.5582,
+      "step": 5505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997482840522218e-05,
+      "loss": 0.5967,
+      "step": 5506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997047827537552e-05,
+      "loss": 0.5164,
+      "step": 5507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996612725175562e-05,
+      "loss": 0.5863,
+      "step": 5508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996177533440567e-05,
+      "loss": 0.5555,
+      "step": 5509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8995742252336892e-05,
+      "loss": 0.5201,
+      "step": 5510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8995306881868862e-05,
+      "loss": 0.5179,
+      "step": 5511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89948714220408e-05,
+      "loss": 0.5066,
+      "step": 5512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899443587285703e-05,
+      "loss": 0.545,
+      "step": 5513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8994000234321885e-05,
+      "loss": 0.5686,
+      "step": 5514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8993564506439687e-05,
+      "loss": 0.4108,
+      "step": 5515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8993128689214767e-05,
+      "loss": 0.621,
+      "step": 5516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992692782651457e-05,
+      "loss": 0.4687,
+      "step": 5517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992256786754085e-05,
+      "loss": 0.595,
+      "step": 5518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991820701526976e-05,
+      "loss": 0.4743,
+      "step": 5519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991384526974472e-05,
+      "loss": 0.5999,
+      "step": 5520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89909482631009e-05,
+      "loss": 0.5022,
+      "step": 5521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89905119099106e-05,
+      "loss": 0.4466,
+      "step": 5522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89900754674079e-05,
+      "loss": 0.5551,
+      "step": 5523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989638935597136e-05,
+      "loss": 0.5377,
+      "step": 5524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989202314482645e-05,
+      "loss": 0.561,
+      "step": 5525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898876560406877e-05,
+      "loss": 0.6497,
+      "step": 5526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8988328804359843e-05,
+      "loss": 0.4834,
+      "step": 5527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898789191536021e-05,
+      "loss": 0.5495,
+      "step": 5528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8987454937074198e-05,
+      "loss": 0.5619,
+      "step": 5529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8987017869506164e-05,
+      "loss": 0.4946,
+      "step": 5530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898658071266044e-05,
+      "loss": 0.5447,
+      "step": 5531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986143466541372e-05,
+      "loss": 0.6296,
+      "step": 5532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89857061311533e-05,
+      "loss": 0.5062,
+      "step": 5533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8985268706500574e-05,
+      "loss": 0.549,
+      "step": 5534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898483119258754e-05,
+      "loss": 0.5088,
+      "step": 5535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984393589418534e-05,
+      "loss": 0.5708,
+      "step": 5536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983955896997917e-05,
+      "loss": 0.4813,
+      "step": 5537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983518115330025e-05,
+      "loss": 0.5031,
+      "step": 5538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983080244419217e-05,
+      "loss": 0.5986,
+      "step": 5539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8982642284269836e-05,
+      "loss": 0.5304,
+      "step": 5540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898220423488623e-05,
+      "loss": 0.4929,
+      "step": 5541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981766096272764e-05,
+      "loss": 0.5296,
+      "step": 5542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981327868433778e-05,
+      "loss": 0.5424,
+      "step": 5543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8980889551373627e-05,
+      "loss": 0.5877,
+      "step": 5544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898045114509667e-05,
+      "loss": 0.581,
+      "step": 5545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898001264960726e-05,
+      "loss": 0.5173,
+      "step": 5546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8979574064909753e-05,
+      "loss": 0.4533,
+      "step": 5547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8979135391008506e-05,
+      "loss": 0.5935,
+      "step": 5548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897869662790788e-05,
+      "loss": 0.5451,
+      "step": 5549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8978257775612223e-05,
+      "loss": 0.5266,
+      "step": 5550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977818834125906e-05,
+      "loss": 0.485,
+      "step": 5551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977379803453283e-05,
+      "loss": 0.5688,
+      "step": 5552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897694068359872e-05,
+      "loss": 0.4901,
+      "step": 5553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8976501474566576e-05,
+      "loss": 0.6,
+      "step": 5554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8976062176361213e-05,
+      "loss": 0.6573,
+      "step": 5555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8975622788987e-05,
+      "loss": 0.5358,
+      "step": 5556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89751833124483e-05,
+      "loss": 0.5988,
+      "step": 5557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974743746749475e-05,
+      "loss": 0.5206,
+      "step": 5558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974304091894896e-05,
+      "loss": 0.4679,
+      "step": 5559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897386434788893e-05,
+      "loss": 0.5549,
+      "step": 5560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973424514735945e-05,
+      "loss": 0.5329,
+      "step": 5561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972984592440308e-05,
+      "loss": 0.6001,
+      "step": 5562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897254458100639e-05,
+      "loss": 0.591,
+      "step": 5563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972104480438563e-05,
+      "loss": 0.6053,
+      "step": 5564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89716642907412e-05,
+      "loss": 0.5394,
+      "step": 5565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8971224011918673e-05,
+      "loss": 0.5813,
+      "step": 5566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897078364397536e-05,
+      "loss": 0.4447,
+      "step": 5567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970343186915626e-05,
+      "loss": 0.5099,
+      "step": 5568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8969902640743855e-05,
+      "loss": 0.4973,
+      "step": 5569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8969462005464418e-05,
+      "loss": 0.5028,
+      "step": 5570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8969021281081695e-05,
+      "loss": 0.4637,
+      "step": 5571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968580467600065e-05,
+      "loss": 0.5278,
+      "step": 5572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968139565023907e-05,
+      "loss": 0.5273,
+      "step": 5573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8967698573357597e-05,
+      "loss": 0.5018,
+      "step": 5574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8967257492605522e-05,
+      "loss": 0.5618,
+      "step": 5575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896681632277206e-05,
+      "loss": 0.48,
+      "step": 5576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8966375063861592e-05,
+      "loss": 0.6306,
+      "step": 5577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8965933715878506e-05,
+      "loss": 0.5666,
+      "step": 5578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896549227882718e-05,
+      "loss": 0.5575,
+      "step": 5579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896505075271201e-05,
+      "loss": 0.5148,
+      "step": 5580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896460913753737e-05,
+      "loss": 0.5642,
+      "step": 5581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964167433307653e-05,
+      "loss": 0.5184,
+      "step": 5582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8963725640027247e-05,
+      "loss": 0.5722,
+      "step": 5583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8963283757700538e-05,
+      "loss": 0.5841,
+      "step": 5584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962841786331923e-05,
+      "loss": 0.5251,
+      "step": 5585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962399725925783e-05,
+      "loss": 0.5499,
+      "step": 5586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961957576486516e-05,
+      "loss": 0.5216,
+      "step": 5587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961515338018512e-05,
+      "loss": 0.5046,
+      "step": 5588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961073010526166e-05,
+      "loss": 0.4826,
+      "step": 5589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896063059401387e-05,
+      "loss": 0.5107,
+      "step": 5590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8960188088486014e-05,
+      "loss": 0.5436,
+      "step": 5591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959745493947006e-05,
+      "loss": 0.5499,
+      "step": 5592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959302810401234e-05,
+      "loss": 0.4883,
+      "step": 5593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89588600378531e-05,
+      "loss": 0.5033,
+      "step": 5594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8958417176307e-05,
+      "loss": 0.5214,
+      "step": 5595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8957974225767328e-05,
+      "loss": 0.4977,
+      "step": 5596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8957531186238497e-05,
+      "loss": 0.5997,
+      "step": 5597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8957088057724898e-05,
+      "loss": 0.5191,
+      "step": 5598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895664484023094e-05,
+      "loss": 0.6982,
+      "step": 5599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956201533761013e-05,
+      "loss": 0.4756,
+      "step": 5600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895575813831954e-05,
+      "loss": 0.5257,
+      "step": 5601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955314653910912e-05,
+      "loss": 0.5437,
+      "step": 5602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895487108053954e-05,
+      "loss": 0.5649,
+      "step": 5603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8954427418209826e-05,
+      "loss": 0.6129,
+      "step": 5604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8953983666926182e-05,
+      "loss": 0.5521,
+      "step": 5605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8953539826693017e-05,
+      "loss": 0.6321,
+      "step": 5606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8953095897514735e-05,
+      "loss": 0.5084,
+      "step": 5607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895265187939575e-05,
+      "loss": 0.5203,
+      "step": 5608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895220777234047e-05,
+      "loss": 0.5746,
+      "step": 5609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895176357635331e-05,
+      "loss": 0.5073,
+      "step": 5610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8951319291438685e-05,
+      "loss": 0.4878,
+      "step": 5611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950874917600997e-05,
+      "loss": 0.4703,
+      "step": 5612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950430454844674e-05,
+      "loss": 0.488,
+      "step": 5613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894998590317412e-05,
+      "loss": 0.4983,
+      "step": 5614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8949541262593764e-05,
+      "loss": 0.6201,
+      "step": 5615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8949096533108012e-05,
+      "loss": 0.4992,
+      "step": 5616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8948651714721284e-05,
+      "loss": 0.6114,
+      "step": 5617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8948206807438005e-05,
+      "loss": 0.5656,
+      "step": 5618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894776181126259e-05,
+      "loss": 0.463,
+      "step": 5619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8947316726199453e-05,
+      "loss": 0.5612,
+      "step": 5620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8946871552253027e-05,
+      "loss": 0.5135,
+      "step": 5621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8946426289427733e-05,
+      "loss": 0.6112,
+      "step": 5622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945980937727987e-05,
+      "loss": 0.4403,
+      "step": 5623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945535497158222e-05,
+      "loss": 0.5102,
+      "step": 5624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945089967722853e-05,
+      "loss": 0.5811,
+      "step": 5625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944644349426314e-05,
+      "loss": 0.542,
+      "step": 5626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944198642273027e-05,
+      "loss": 0.5023,
+      "step": 5627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943752846267424e-05,
+      "loss": 0.5739,
+      "step": 5628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943306961413934e-05,
+      "loss": 0.508,
+      "step": 5629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8942860987716977e-05,
+      "loss": 0.5332,
+      "step": 5630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8942414925180995e-05,
+      "loss": 0.6043,
+      "step": 5631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941968773810413e-05,
+      "loss": 0.5714,
+      "step": 5632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894152253360966e-05,
+      "loss": 0.5525,
+      "step": 5633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894107620458318e-05,
+      "loss": 0.5665,
+      "step": 5634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.89406297867354e-05,
+      "loss": 0.4479,
+      "step": 5635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8940183280070755e-05,
+      "loss": 0.5953,
+      "step": 5636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8939736684593678e-05,
+      "loss": 0.6439,
+      "step": 5637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8939290000308607e-05,
+      "loss": 0.5531,
+      "step": 5638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938843227219986e-05,
+      "loss": 0.5187,
+      "step": 5639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938396365332243e-05,
+      "loss": 0.5483,
+      "step": 5640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937949414649822e-05,
+      "loss": 0.5218,
+      "step": 5641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937502375177166e-05,
+      "loss": 0.4888,
+      "step": 5642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937055246918712e-05,
+      "loss": 0.5306,
+      "step": 5643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8936608029878904e-05,
+      "loss": 0.519,
+      "step": 5644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8936160724062182e-05,
+      "loss": 0.4415,
+      "step": 5645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893571332947299e-05,
+      "loss": 0.5819,
+      "step": 5646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8935265846115772e-05,
+      "loss": 0.5125,
+      "step": 5647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893481827399498e-05,
+      "loss": 0.5801,
+      "step": 5648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893437061311505e-05,
+      "loss": 0.5074,
+      "step": 5649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933922863480435e-05,
+      "loss": 0.501,
+      "step": 5650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933475025095586e-05,
+      "loss": 0.5659,
+      "step": 5651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933027097964948e-05,
+      "loss": 0.5865,
+      "step": 5652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8932579082092964e-05,
+      "loss": 0.5548,
+      "step": 5653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.89321309774841e-05,
+      "loss": 0.5019,
+      "step": 5654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931682784142792e-05,
+      "loss": 0.6307,
+      "step": 5655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931234502073503e-05,
+      "loss": 0.4807,
+      "step": 5656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893078613128068e-05,
+      "loss": 0.5407,
+      "step": 5657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893033767176878e-05,
+      "loss": 0.5283,
+      "step": 5658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892988912354226e-05,
+      "loss": 0.5677,
+      "step": 5659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892944048660557e-05,
+      "loss": 0.5299,
+      "step": 5660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928991760963175e-05,
+      "loss": 0.5336,
+      "step": 5661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928542946619527e-05,
+      "loss": 0.5857,
+      "step": 5662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928094043579082e-05,
+      "loss": 0.6214,
+      "step": 5663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927645051846305e-05,
+      "loss": 0.4946,
+      "step": 5664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927195971425655e-05,
+      "loss": 0.4616,
+      "step": 5665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892674680232159e-05,
+      "loss": 0.519,
+      "step": 5666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892629754453858e-05,
+      "loss": 0.5925,
+      "step": 5667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892584819808108e-05,
+      "loss": 0.5577,
+      "step": 5668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925398762953555e-05,
+      "loss": 0.5039,
+      "step": 5669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892494923916047e-05,
+      "loss": 0.533,
+      "step": 5670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8924499626706293e-05,
+      "loss": 0.4663,
+      "step": 5671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892404992559549e-05,
+      "loss": 0.4113,
+      "step": 5672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892360013583253e-05,
+      "loss": 0.4957,
+      "step": 5673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923150257421876e-05,
+      "loss": 0.5841,
+      "step": 5674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922700290367998e-05,
+      "loss": 0.6354,
+      "step": 5675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922250234675374e-05,
+      "loss": 0.5022,
+      "step": 5676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892180009034846e-05,
+      "loss": 0.6137,
+      "step": 5677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921349857391745e-05,
+      "loss": 0.5579,
+      "step": 5678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892089953580969e-05,
+      "loss": 0.5616,
+      "step": 5679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920449125606773e-05,
+      "loss": 0.5399,
+      "step": 5680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919998626787467e-05,
+      "loss": 0.5909,
+      "step": 5681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919548039356243e-05,
+      "loss": 0.5362,
+      "step": 5682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919097363317583e-05,
+      "loss": 0.5931,
+      "step": 5683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918646598675966e-05,
+      "loss": 0.4498,
+      "step": 5684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918195745435866e-05,
+      "loss": 0.6314,
+      "step": 5685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8917744803601764e-05,
+      "loss": 0.4972,
+      "step": 5686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917293773178136e-05,
+      "loss": 0.4794,
+      "step": 5687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8916842654169464e-05,
+      "loss": 0.5331,
+      "step": 5688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891639144658023e-05,
+      "loss": 0.4246,
+      "step": 5689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8915940150414917e-05,
+      "loss": 0.565,
+      "step": 5690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891548876567801e-05,
+      "loss": 0.5326,
+      "step": 5691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8915037292373987e-05,
+      "loss": 0.5827,
+      "step": 5692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891458573050734e-05,
+      "loss": 0.5059,
+      "step": 5693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891413408008255e-05,
+      "loss": 0.6728,
+      "step": 5694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913682341104107e-05,
+      "loss": 0.5595,
+      "step": 5695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913230513576495e-05,
+      "loss": 0.454,
+      "step": 5696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8912778597504203e-05,
+      "loss": 0.5504,
+      "step": 5697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8912326592891726e-05,
+      "loss": 0.5201,
+      "step": 5698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911874499743547e-05,
+      "loss": 0.5495,
+      "step": 5699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891142231806416e-05,
+      "loss": 0.5206,
+      "step": 5700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891097004785806e-05,
+      "loss": 0.4997,
+      "step": 5701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910517689129734e-05,
+      "loss": 0.5198,
+      "step": 5702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.4786,
+      "step": 5703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890961270612439e-05,
+      "loss": 0.5389,
+      "step": 5704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8909160081856365e-05,
+      "loss": 0.554,
+      "step": 5705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908707369084093e-05,
+      "loss": 0.5852,
+      "step": 5706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908254567812078e-05,
+      "loss": 0.5424,
+      "step": 5707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907801678044814e-05,
+      "loss": 0.4591,
+      "step": 5708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907348699786805e-05,
+      "loss": 0.5602,
+      "step": 5709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8906895633042545e-05,
+      "loss": 0.536,
+      "step": 5710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890644247781654e-05,
+      "loss": 0.5078,
+      "step": 5711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905989234113288e-05,
+      "loss": 0.492,
+      "step": 5712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905535901937294e-05,
+      "loss": 0.6137,
+      "step": 5713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890508248129306e-05,
+      "loss": 0.5271,
+      "step": 5714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8904628972185086e-05,
+      "loss": 0.5125,
+      "step": 5715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8904175374617884e-05,
+      "loss": 0.6068,
+      "step": 5716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890372168859596e-05,
+      "loss": 0.651,
+      "step": 5717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890326791412382e-05,
+      "loss": 0.5508,
+      "step": 5718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902814051205967e-05,
+      "loss": 0.5067,
+      "step": 5719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902360099846913e-05,
+      "loss": 0.5217,
+      "step": 5720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8901906060051172e-05,
+      "loss": 0.5488,
+      "step": 5721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8901451931823247e-05,
+      "loss": 0.592,
+      "step": 5722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8900997715167653e-05,
+      "loss": 0.5641,
+      "step": 5723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.89005434100889e-05,
+      "loss": 0.5761,
+      "step": 5724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890008901659151e-05,
+      "loss": 0.5461,
+      "step": 5725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899634534679984e-05,
+      "loss": 0.4906,
+      "step": 5726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899179964358848e-05,
+      "loss": 0.5825,
+      "step": 5727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8898725305632605e-05,
+      "loss": 0.5246,
+      "step": 5728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8898270558505783e-05,
+      "loss": 0.6316,
+      "step": 5729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8897815722982897e-05,
+      "loss": 0.5919,
+      "step": 5730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8897360799068463e-05,
+      "loss": 0.5168,
+      "step": 5731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8896905786767003e-05,
+      "loss": 0.5076,
+      "step": 5732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8896450686083034e-05,
+      "loss": 0.5975,
+      "step": 5733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895995497021074e-05,
+      "loss": 0.5775,
+      "step": 5734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895540219585656e-05,
+      "loss": 0.5466,
+      "step": 5735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895084853781297e-05,
+      "loss": 0.5467,
+      "step": 5736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8894629399612516e-05,
+      "loss": 0.5331,
+      "step": 5737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889417385708384e-05,
+      "loss": 0.5733,
+      "step": 5738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8893718226199797e-05,
+      "loss": 0.5556,
+      "step": 5739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8893262506964913e-05,
+      "loss": 0.6124,
+      "step": 5740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892806699383713e-05,
+      "loss": 0.5729,
+      "step": 5741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892350803460728e-05,
+      "loss": 0.6087,
+      "step": 5742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889189481920048e-05,
+      "loss": 0.5466,
+      "step": 5743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889143874660751e-05,
+      "loss": 0.5595,
+      "step": 5744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889098258568634e-05,
+      "loss": 0.4396,
+      "step": 5745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8890526336441503e-05,
+      "loss": 0.5143,
+      "step": 5746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889006999887753e-05,
+      "loss": 0.4803,
+      "step": 5747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889613572998963e-05,
+      "loss": 0.525,
+      "step": 5748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889157058810322e-05,
+      "loss": 0.4632,
+      "step": 5749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8888700456316155e-05,
+      "loss": 0.4944,
+      "step": 5750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8888243765520996e-05,
+      "loss": 0.5888,
+      "step": 5751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8887786986429374e-05,
+      "loss": 0.4994,
+      "step": 5752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888733011904583e-05,
+      "loss": 0.4942,
+      "step": 5753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8886873163374905e-05,
+      "loss": 0.4811,
+      "step": 5754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888641611942114e-05,
+      "loss": 0.6427,
+      "step": 5755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8885958987189067e-05,
+      "loss": 0.5385,
+      "step": 5756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8885501766683236e-05,
+      "loss": 0.4902,
+      "step": 5757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888504445790819e-05,
+      "loss": 0.6324,
+      "step": 5758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888458706086846e-05,
+      "loss": 0.522,
+      "step": 5759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8884129575568605e-05,
+      "loss": 0.4828,
+      "step": 5760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8883672002013157e-05,
+      "loss": 0.5842,
+      "step": 5761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8883214340206667e-05,
+      "loss": 0.5134,
+      "step": 5762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882756590153683e-05,
+      "loss": 0.5034,
+      "step": 5763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882298751858748e-05,
+      "loss": 0.5111,
+      "step": 5764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881840825326416e-05,
+      "loss": 0.4435,
+      "step": 5765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881382810561232e-05,
+      "loss": 0.5471,
+      "step": 5766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888092470756775e-05,
+      "loss": 0.4219,
+      "step": 5767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880466516350512e-05,
+      "loss": 0.5615,
+      "step": 5768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888000823691408e-05,
+      "loss": 0.5579,
+      "step": 5769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879549869262997e-05,
+      "loss": 0.5179,
+      "step": 5770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879091413401825e-05,
+      "loss": 0.5369,
+      "step": 5771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887863286933511e-05,
+      "loss": 0.5865,
+      "step": 5772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887817423706742e-05,
+      "loss": 0.5754,
+      "step": 5773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88777155166033e-05,
+      "loss": 0.5861,
+      "step": 5774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8877256707947308e-05,
+      "loss": 0.556,
+      "step": 5775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876797811104006e-05,
+      "loss": 0.4677,
+      "step": 5776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876338826077947e-05,
+      "loss": 0.5218,
+      "step": 5777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8875879752873694e-05,
+      "loss": 0.4785,
+      "step": 5778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887542059149581e-05,
+      "loss": 0.4848,
+      "step": 5779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8874961341948856e-05,
+      "loss": 0.5293,
+      "step": 5780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8874502004237386e-05,
+      "loss": 0.5338,
+      "step": 5781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8874042578365973e-05,
+      "loss": 0.5814,
+      "step": 5782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8873583064339178e-05,
+      "loss": 0.4953,
+      "step": 5783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8873123462161566e-05,
+      "loss": 0.4702,
+      "step": 5784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88726637718377e-05,
+      "loss": 0.4852,
+      "step": 5785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8872203993372146e-05,
+      "loss": 0.6461,
+      "step": 5786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8871744126769476e-05,
+      "loss": 0.5735,
+      "step": 5787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8871284172034258e-05,
+      "loss": 0.5733,
+      "step": 5788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887082412917106e-05,
+      "loss": 0.5158,
+      "step": 5789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887036399818445e-05,
+      "loss": 0.5686,
+      "step": 5790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8869903779078996e-05,
+      "loss": 0.4792,
+      "step": 5791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886944347185928e-05,
+      "loss": 0.5427,
+      "step": 5792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886898307652987e-05,
+      "loss": 0.543,
+      "step": 5793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8868522593095334e-05,
+      "loss": 0.4884,
+      "step": 5794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8868062021560254e-05,
+      "loss": 0.4949,
+      "step": 5795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88676013619292e-05,
+      "loss": 0.5893,
+      "step": 5796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867140614206755e-05,
+      "loss": 0.5163,
+      "step": 5797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8866679778397487e-05,
+      "loss": 0.531,
+      "step": 5798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8866218854505982e-05,
+      "loss": 0.5828,
+      "step": 5799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8865757842536812e-05,
+      "loss": 0.5368,
+      "step": 5800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8865296742494563e-05,
+      "loss": 0.6154,
+      "step": 5801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864835554383815e-05,
+      "loss": 0.5716,
+      "step": 5802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864374278209145e-05,
+      "loss": 0.6508,
+      "step": 5803
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886391291397514e-05,
+      "loss": 0.6205,
+      "step": 5804
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8863451461686376e-05,
+      "loss": 0.5161,
+      "step": 5805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886298992134745e-05,
+      "loss": 0.4905,
+      "step": 5806
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886252829296293e-05,
+      "loss": 0.497,
+      "step": 5807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8862066576537418e-05,
+      "loss": 0.447,
+      "step": 5808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861604772075492e-05,
+      "loss": 0.5293,
+      "step": 5809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886114287958174e-05,
+      "loss": 0.4996,
+      "step": 5810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886068089906075e-05,
+      "loss": 0.5504,
+      "step": 5811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8860218830517115e-05,
+      "loss": 0.5092,
+      "step": 5812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8859756673955424e-05,
+      "loss": 0.527,
+      "step": 5813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8859294429380267e-05,
+      "loss": 0.5994,
+      "step": 5814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.885883209679624e-05,
+      "loss": 0.5748,
+      "step": 5815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8858369676207927e-05,
+      "loss": 0.4792,
+      "step": 5816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885790716761993e-05,
+      "loss": 0.4694,
+      "step": 5817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8857444571036837e-05,
+      "loss": 0.5021,
+      "step": 5818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885698188646325e-05,
+      "loss": 0.57,
+      "step": 5819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856519113903763e-05,
+      "loss": 0.4998,
+      "step": 5820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885605625336297e-05,
+      "loss": 0.4668,
+      "step": 5821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885559330484547e-05,
+      "loss": 0.5435,
+      "step": 5822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8855130268355868e-05,
+      "loss": 0.5144,
+      "step": 5823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885466714389876e-05,
+      "loss": 0.5265,
+      "step": 5824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8854203931478742e-05,
+      "loss": 0.5919,
+      "step": 5825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853740631100422e-05,
+      "loss": 0.5978,
+      "step": 5826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853277242768403e-05,
+      "loss": 0.5358,
+      "step": 5827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852813766487282e-05,
+      "loss": 0.5431,
+      "step": 5828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852350202261665e-05,
+      "loss": 0.6294,
+      "step": 5829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8851886550096162e-05,
+      "loss": 0.5614,
+      "step": 5830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8851422809995375e-05,
+      "loss": 0.4334,
+      "step": 5831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850958981963915e-05,
+      "loss": 0.5317,
+      "step": 5832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885049506600638e-05,
+      "loss": 0.4088,
+      "step": 5833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885003106212739e-05,
+      "loss": 0.493,
+      "step": 5834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884956697033155e-05,
+      "loss": 0.485,
+      "step": 5835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8849102790623464e-05,
+      "loss": 0.5158,
+      "step": 5836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8848638523007752e-05,
+      "loss": 0.5459,
+      "step": 5837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8848174167489025e-05,
+      "loss": 0.5697,
+      "step": 5838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884770972407189e-05,
+      "loss": 0.5003,
+      "step": 5839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884724519276097e-05,
+      "loss": 0.6176,
+      "step": 5840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8846780573560868e-05,
+      "loss": 0.5297,
+      "step": 5841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884631586647621e-05,
+      "loss": 0.4834,
+      "step": 5842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8845851071511605e-05,
+      "loss": 0.4849,
+      "step": 5843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884538618867168e-05,
+      "loss": 0.6101,
+      "step": 5844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884492121796104e-05,
+      "loss": 0.5631,
+      "step": 5845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884445615938431e-05,
+      "loss": 0.5341,
+      "step": 5846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843991012946116e-05,
+      "loss": 0.5603,
+      "step": 5847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884352577865107e-05,
+      "loss": 0.5151,
+      "step": 5848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843060456503802e-05,
+      "loss": 0.516,
+      "step": 5849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884259504650893e-05,
+      "loss": 0.541,
+      "step": 5850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8842129548671073e-05,
+      "loss": 0.5543,
+      "step": 5851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8841663962994863e-05,
+      "loss": 0.5007,
+      "step": 5852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8841198289484917e-05,
+      "loss": 0.4966,
+      "step": 5853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884073252814587e-05,
+      "loss": 0.6272,
+      "step": 5854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8840266678982343e-05,
+      "loss": 0.5306,
+      "step": 5855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8839800741998967e-05,
+      "loss": 0.45,
+      "step": 5856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8839334717200368e-05,
+      "loss": 0.5255,
+      "step": 5857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838868604591175e-05,
+      "loss": 0.5318,
+      "step": 5858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838402404176023e-05,
+      "loss": 0.4607,
+      "step": 5859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883793611595954e-05,
+      "loss": 0.5662,
+      "step": 5860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8837469739946355e-05,
+      "loss": 0.5516,
+      "step": 5861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8837003276141107e-05,
+      "loss": 0.4996,
+      "step": 5862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8836536724548426e-05,
+      "loss": 0.5606,
+      "step": 5863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883607008517295e-05,
+      "loss": 0.4565,
+      "step": 5864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8835603358019313e-05,
+      "loss": 0.4797,
+      "step": 5865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883513654309215e-05,
+      "loss": 0.4805,
+      "step": 5866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8834669640396098e-05,
+      "loss": 0.5842,
+      "step": 5867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88342026499358e-05,
+      "loss": 0.4004,
+      "step": 5868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883373557171589e-05,
+      "loss": 0.5356,
+      "step": 5869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883326840574101e-05,
+      "loss": 0.5094,
+      "step": 5870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88328011520158e-05,
+      "loss": 0.5629,
+      "step": 5871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8832333810544905e-05,
+      "loss": 0.6432,
+      "step": 5872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8831866381332964e-05,
+      "loss": 0.4915,
+      "step": 5873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8831398864384622e-05,
+      "loss": 0.5078,
+      "step": 5874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883093125970452e-05,
+      "loss": 0.5521,
+      "step": 5875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883046356729731e-05,
+      "loss": 0.5967,
+      "step": 5876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8829995787167634e-05,
+      "loss": 0.5188,
+      "step": 5877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882952791932014e-05,
+      "loss": 0.5941,
+      "step": 5878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882905996375947e-05,
+      "loss": 0.5163,
+      "step": 5879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8828591920490284e-05,
+      "loss": 0.6011,
+      "step": 5880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882812378951722e-05,
+      "loss": 0.5403,
+      "step": 5881
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8827655570844935e-05,
+      "loss": 0.5345,
+      "step": 5882
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8827187264478082e-05,
+      "loss": 0.6448,
+      "step": 5883
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882671887042131e-05,
+      "loss": 0.5643,
+      "step": 5884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882625038867927e-05,
+      "loss": 0.5471,
+      "step": 5885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882578181925662e-05,
+      "loss": 0.5345,
+      "step": 5886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882531316215801e-05,
+      "loss": 0.5323,
+      "step": 5887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824844417388102e-05,
+      "loss": 0.4796,
+      "step": 5888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824375584951545e-05,
+      "loss": 0.6297,
+      "step": 5889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8823906664853007e-05,
+      "loss": 0.6016,
+      "step": 5890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8823437657097137e-05,
+      "loss": 0.6212,
+      "step": 5891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8822968561688596e-05,
+      "loss": 0.5974,
+      "step": 5892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8822499378632047e-05,
+      "loss": 0.6019,
+      "step": 5893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882203010793215e-05,
+      "loss": 0.5383,
+      "step": 5894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821560749593564e-05,
+      "loss": 0.4739,
+      "step": 5895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821091303620953e-05,
+      "loss": 0.4803,
+      "step": 5896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8820621770018982e-05,
+      "loss": 0.5954,
+      "step": 5897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8820152148792313e-05,
+      "loss": 0.6276,
+      "step": 5898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8819682439945616e-05,
+      "loss": 0.5314,
+      "step": 5899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.5448,
+      "step": 5900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881874275941079e-05,
+      "loss": 0.5011,
+      "step": 5901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8818272787731997e-05,
+      "loss": 0.5013,
+      "step": 5902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881780272845184e-05,
+      "loss": 0.5508,
+      "step": 5903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817332581574998e-05,
+      "loss": 0.4851,
+      "step": 5904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8816862347106125e-05,
+      "loss": 0.5636,
+      "step": 5905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881639202504991e-05,
+      "loss": 0.6835,
+      "step": 5906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8815921615411013e-05,
+      "loss": 0.4766,
+      "step": 5907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8815451118194112e-05,
+      "loss": 0.5287,
+      "step": 5908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814980533403878e-05,
+      "loss": 0.5423,
+      "step": 5909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814509861044992e-05,
+      "loss": 0.5532,
+      "step": 5910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881403910112212e-05,
+      "loss": 0.5659,
+      "step": 5911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881356825363995e-05,
+      "loss": 0.5938,
+      "step": 5912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8813097318603152e-05,
+      "loss": 0.5615,
+      "step": 5913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812626296016403e-05,
+      "loss": 0.4618,
+      "step": 5914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812155185884385e-05,
+      "loss": 0.5546,
+      "step": 5915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8811683988211783e-05,
+      "loss": 0.498,
+      "step": 5916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8811212703003268e-05,
+      "loss": 0.5522,
+      "step": 5917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881074133026353e-05,
+      "loss": 0.5609,
+      "step": 5918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8810269869997247e-05,
+      "loss": 0.4966,
+      "step": 5919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8809798322209102e-05,
+      "loss": 0.5004,
+      "step": 5920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8809326686903785e-05,
+      "loss": 0.4996,
+      "step": 5921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808854964085975e-05,
+      "loss": 0.5539,
+      "step": 5922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880838315376036e-05,
+      "loss": 0.7096,
+      "step": 5923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880791125593163e-05,
+      "loss": 0.5857,
+      "step": 5924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880743927060447e-05,
+      "loss": 0.6853,
+      "step": 5925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880696719778357e-05,
+      "loss": 0.5906,
+      "step": 5926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880649503747362e-05,
+      "loss": 0.4326,
+      "step": 5927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806022789679302e-05,
+      "loss": 0.4793,
+      "step": 5928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880555045440532e-05,
+      "loss": 0.5664,
+      "step": 5929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805078031656366e-05,
+      "loss": 0.6056,
+      "step": 5930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880460552143712e-05,
+      "loss": 0.5465,
+      "step": 5931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8804132923752288e-05,
+      "loss": 0.5391,
+      "step": 5932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8803660238606557e-05,
+      "loss": 0.4863,
+      "step": 5933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8803187466004633e-05,
+      "loss": 0.5335,
+      "step": 5934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88027146059512e-05,
+      "loss": 0.5279,
+      "step": 5935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802241658450965e-05,
+      "loss": 0.5567,
+      "step": 5936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880176862350862e-05,
+      "loss": 0.4795,
+      "step": 5937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880129550112887e-05,
+      "loss": 0.5025,
+      "step": 5938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880082229131641e-05,
+      "loss": 0.6145,
+      "step": 5939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800348994075944e-05,
+      "loss": 0.4934,
+      "step": 5940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.879987560941217e-05,
+      "loss": 0.5967,
+      "step": 5941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799402137329798e-05,
+      "loss": 0.5696,
+      "step": 5942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8798928577833524e-05,
+      "loss": 0.5898,
+      "step": 5943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8798454930928054e-05,
+      "loss": 0.6345,
+      "step": 5944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8797981196618095e-05,
+      "loss": 0.5845,
+      "step": 5945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8797507374908353e-05,
+      "loss": 0.5276,
+      "step": 5946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8797033465803536e-05,
+      "loss": 0.5299,
+      "step": 5947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879655946930835e-05,
+      "loss": 0.5064,
+      "step": 5948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.87960853854275e-05,
+      "loss": 0.5754,
+      "step": 5949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.87956112141657e-05,
+      "loss": 0.5946,
+      "step": 5950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795136955527664e-05,
+      "loss": 0.5448,
+      "step": 5951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.87946626095181e-05,
+      "loss": 0.5374,
+      "step": 5952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794188176141715e-05,
+      "loss": 0.5461,
+      "step": 5953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879371365540323e-05,
+      "loss": 0.5156,
+      "step": 5954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8793239047307358e-05,
+      "loss": 0.5585,
+      "step": 5955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792764351858803e-05,
+      "loss": 0.5914,
+      "step": 5956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792289569062297e-05,
+      "loss": 0.4671,
+      "step": 5957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791814698922545e-05,
+      "loss": 0.5484,
+      "step": 5958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791339741444267e-05,
+      "loss": 0.4472,
+      "step": 5959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790864696632186e-05,
+      "loss": 0.5466,
+      "step": 5960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790389564491015e-05,
+      "loss": 0.5558,
+      "step": 5961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8789914345025474e-05,
+      "loss": 0.46,
+      "step": 5962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878943903824029e-05,
+      "loss": 0.5495,
+      "step": 5963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878896364414018e-05,
+      "loss": 0.5692,
+      "step": 5964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8788488162729868e-05,
+      "loss": 0.5391,
+      "step": 5965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8788012594014072e-05,
+      "loss": 0.508,
+      "step": 5966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8787536937997528e-05,
+      "loss": 0.4705,
+      "step": 5967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878706119468495e-05,
+      "loss": 0.4874,
+      "step": 5968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8786585364081073e-05,
+      "loss": 0.5612,
+      "step": 5969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8786109446190615e-05,
+      "loss": 0.5015,
+      "step": 5970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878563344101831e-05,
+      "loss": 0.4884,
+      "step": 5971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8785157348568884e-05,
+      "loss": 0.5984,
+      "step": 5972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878468116884707e-05,
+      "loss": 0.5205,
+      "step": 5973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784204901857595e-05,
+      "loss": 0.5208,
+      "step": 5974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8783728547605192e-05,
+      "loss": 0.5112,
+      "step": 5975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8783252106094592e-05,
+      "loss": 0.5378,
+      "step": 5976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782775577330526e-05,
+      "loss": 0.5214,
+      "step": 5977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782298961317732e-05,
+      "loss": 0.5417,
+      "step": 5978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878182225806094e-05,
+      "loss": 0.5836,
+      "step": 5979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8781345467564893e-05,
+      "loss": 0.5783,
+      "step": 5980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878086858983432e-05,
+      "loss": 0.5114,
+      "step": 5981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780391624873965e-05,
+      "loss": 0.5069,
+      "step": 5982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877991457268856e-05,
+      "loss": 0.5554,
+      "step": 5983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8779437433282848e-05,
+      "loss": 0.5547,
+      "step": 5984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778960206661567e-05,
+      "loss": 0.5115,
+      "step": 5985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778482892829458e-05,
+      "loss": 0.508,
+      "step": 5986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778005491791262e-05,
+      "loss": 0.5963,
+      "step": 5987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777528003551725e-05,
+      "loss": 0.5452,
+      "step": 5988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777050428115585e-05,
+      "loss": 0.5673,
+      "step": 5989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8776572765487594e-05,
+      "loss": 0.5918,
+      "step": 5990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877609501567249e-05,
+      "loss": 0.5138,
+      "step": 5991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775617178675017e-05,
+      "loss": 0.4274,
+      "step": 5992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775139254499932e-05,
+      "loss": 0.5237,
+      "step": 5993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8774661243151976e-05,
+      "loss": 0.4088,
+      "step": 5994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8774183144635894e-05,
+      "loss": 0.5148,
+      "step": 5995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773704958956446e-05,
+      "loss": 0.4753,
+      "step": 5996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773226686118374e-05,
+      "loss": 0.5431,
+      "step": 5997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877274832612643e-05,
+      "loss": 0.5564,
+      "step": 5998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8772269878985373e-05,
+      "loss": 0.5623,
+      "step": 5999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771791344699943e-05,
+      "loss": 0.5546,
+      "step": 6000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877131272327491e-05,
+      "loss": 0.5303,
+      "step": 6001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770834014715015e-05,
+      "loss": 0.5384,
+      "step": 6002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770355219025016e-05,
+      "loss": 0.5315,
+      "step": 6003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8769876336209676e-05,
+      "loss": 0.5879,
+      "step": 6004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876939736627375e-05,
+      "loss": 0.5794,
+      "step": 6005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876891830922199e-05,
+      "loss": 0.5045,
+      "step": 6006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8768439165059157e-05,
+      "loss": 0.5341,
+      "step": 6007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767959933790017e-05,
+      "loss": 0.5409,
+      "step": 6008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767480615419328e-05,
+      "loss": 0.5494,
+      "step": 6009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767001209951852e-05,
+      "loss": 0.5075,
+      "step": 6010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8766521717392344e-05,
+      "loss": 0.5702,
+      "step": 6011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8766042137745576e-05,
+      "loss": 0.5241,
+      "step": 6012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876556247101631e-05,
+      "loss": 0.587,
+      "step": 6013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8765082717209314e-05,
+      "loss": 0.5883,
+      "step": 6014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8764602876329346e-05,
+      "loss": 0.5053,
+      "step": 6015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8764122948381177e-05,
+      "loss": 0.4759,
+      "step": 6016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8763642933369578e-05,
+      "loss": 0.5242,
+      "step": 6017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876316283129931e-05,
+      "loss": 0.4416,
+      "step": 6018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876268264217515e-05,
+      "loss": 0.4641,
+      "step": 6019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762202366001863e-05,
+      "loss": 0.6837,
+      "step": 6020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8761722002784225e-05,
+      "loss": 0.524,
+      "step": 6021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8761241552527e-05,
+      "loss": 0.4999,
+      "step": 6022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876076101523497e-05,
+      "loss": 0.5533,
+      "step": 6023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87602803909129e-05,
+      "loss": 0.6582,
+      "step": 6024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8759799679565572e-05,
+      "loss": 0.5139,
+      "step": 6025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875931888119776e-05,
+      "loss": 0.4838,
+      "step": 6026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875883799581424e-05,
+      "loss": 0.4835,
+      "step": 6027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8758357023419782e-05,
+      "loss": 0.5462,
+      "step": 6028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8757875964019172e-05,
+      "loss": 0.487,
+      "step": 6029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875739481761719e-05,
+      "loss": 0.5511,
+      "step": 6030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875691358421861e-05,
+      "loss": 0.5909,
+      "step": 6031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8756432263828215e-05,
+      "loss": 0.5378,
+      "step": 6032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755950856450786e-05,
+      "loss": 0.4296,
+      "step": 6033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755469362091107e-05,
+      "loss": 0.6197,
+      "step": 6034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875498778075396e-05,
+      "loss": 0.5776,
+      "step": 6035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875450611244413e-05,
+      "loss": 0.4949,
+      "step": 6036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87540243571664e-05,
+      "loss": 0.5366,
+      "step": 6037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8753542514925556e-05,
+      "loss": 0.521,
+      "step": 6038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875306058572639e-05,
+      "loss": 0.55,
+      "step": 6039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8752578569573682e-05,
+      "loss": 0.5327,
+      "step": 6040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8752096466472224e-05,
+      "loss": 0.5314,
+      "step": 6041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8751614276426804e-05,
+      "loss": 0.4785,
+      "step": 6042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875113199944221e-05,
+      "loss": 0.5454,
+      "step": 6043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875064963552324e-05,
+      "loss": 0.5349,
+      "step": 6044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8750167184674682e-05,
+      "loss": 0.447,
+      "step": 6045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8749684646901326e-05,
+      "loss": 0.4821,
+      "step": 6046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874920202220797e-05,
+      "loss": 0.5187,
+      "step": 6047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8748719310599407e-05,
+      "loss": 0.6268,
+      "step": 6048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874823651208043e-05,
+      "loss": 0.4661,
+      "step": 6049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8747753626655833e-05,
+      "loss": 0.4666,
+      "step": 6050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874727065433042e-05,
+      "loss": 0.478,
+      "step": 6051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8746787595108987e-05,
+      "loss": 0.6019,
+      "step": 6052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874630444899633e-05,
+      "loss": 0.5359,
+      "step": 6053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8745821215997254e-05,
+      "loss": 0.5166,
+      "step": 6054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874533789611655e-05,
+      "loss": 0.5122,
+      "step": 6055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744854489359026e-05,
+      "loss": 0.5788,
+      "step": 6056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744370995729487e-05,
+      "loss": 0.5684,
+      "step": 6057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8743887415232728e-05,
+      "loss": 0.4726,
+      "step": 6058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8743403747873557e-05,
+      "loss": 0.6415,
+      "step": 6059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874291999365678e-05,
+      "loss": 0.5718,
+      "step": 6060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8742436152587206e-05,
+      "loss": 0.5882,
+      "step": 6061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874195222466963e-05,
+      "loss": 0.5012,
+      "step": 6062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874146820990887e-05,
+      "loss": 0.5306,
+      "step": 6063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874098410830973e-05,
+      "loss": 0.6042,
+      "step": 6064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8740499919877017e-05,
+      "loss": 0.4564,
+      "step": 6065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8740015644615547e-05,
+      "loss": 0.6079,
+      "step": 6066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739531282530127e-05,
+      "loss": 0.5457,
+      "step": 6067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739046833625568e-05,
+      "loss": 0.4517,
+      "step": 6068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8738562297906685e-05,
+      "loss": 0.4473,
+      "step": 6069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873807767537829e-05,
+      "loss": 0.5743,
+      "step": 6070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8737592966045196e-05,
+      "loss": 0.4945,
+      "step": 6071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8737108169912223e-05,
+      "loss": 0.4406,
+      "step": 6072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873662328698418e-05,
+      "loss": 0.4494,
+      "step": 6073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873613831726589e-05,
+      "loss": 0.6472,
+      "step": 6074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8735653260762165e-05,
+      "loss": 0.5457,
+      "step": 6075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8735168117477827e-05,
+      "loss": 0.4909,
+      "step": 6076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.87346828874177e-05,
+      "loss": 0.6292,
+      "step": 6077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734197570586594e-05,
+      "loss": 0.5031,
+      "step": 6078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733712166989336e-05,
+      "loss": 0.5509,
+      "step": 6079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733226676630752e-05,
+      "loss": 0.5307,
+      "step": 6080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8732741099515658e-05,
+      "loss": 0.4877,
+      "step": 6081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873225543564888e-05,
+      "loss": 0.7263,
+      "step": 6082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731769685035248e-05,
+      "loss": 0.5752,
+      "step": 6083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731283847679574e-05,
+      "loss": 0.5365,
+      "step": 6084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8730797923586696e-05,
+      "loss": 0.4726,
+      "step": 6085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873031191276144e-05,
+      "loss": 0.5298,
+      "step": 6086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8729825815208632e-05,
+      "loss": 0.5596,
+      "step": 6087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8729339630933098e-05,
+      "loss": 0.5039,
+      "step": 6088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728853359939672e-05,
+      "loss": 0.4983,
+      "step": 6089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728367002233187e-05,
+      "loss": 0.5143,
+      "step": 6090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8727880557818466e-05,
+      "loss": 0.4905,
+      "step": 6091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872739402670035e-05,
+      "loss": 0.5815,
+      "step": 6092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8726907408883662e-05,
+      "loss": 0.4951,
+      "step": 6093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8726420704373248e-05,
+      "loss": 0.6037,
+      "step": 6094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872593391317394e-05,
+      "loss": 0.4713,
+      "step": 6095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725447035290568e-05,
+      "loss": 0.4209,
+      "step": 6096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8724960070727974e-05,
+      "loss": 0.5352,
+      "step": 6097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872447301949099e-05,
+      "loss": 0.4281,
+      "step": 6098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723985881584463e-05,
+      "loss": 0.5117,
+      "step": 6099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723498657013224e-05,
+      "loss": 0.6189,
+      "step": 6100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723011345782116e-05,
+      "loss": 0.532,
+      "step": 6101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872252394789598e-05,
+      "loss": 0.4654,
+      "step": 6102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872203646335966e-05,
+      "loss": 0.5374,
+      "step": 6103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721548892178e-05,
+      "loss": 0.5598,
+      "step": 6104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872106123435584e-05,
+      "loss": 0.4956,
+      "step": 6105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872057348989802e-05,
+      "loss": 0.4958,
+      "step": 6106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872008565880939e-05,
+      "loss": 0.4962,
+      "step": 6107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719597741094802e-05,
+      "loss": 0.4784,
+      "step": 6108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719109736759093e-05,
+      "loss": 0.6119,
+      "step": 6109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8718621645807116e-05,
+      "loss": 0.433,
+      "step": 6110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871813346824372e-05,
+      "loss": 0.5742,
+      "step": 6111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8717645204073756e-05,
+      "loss": 0.5449,
+      "step": 6112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871715685330207e-05,
+      "loss": 0.5848,
+      "step": 6113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716668415933515e-05,
+      "loss": 0.5468,
+      "step": 6114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716179891972942e-05,
+      "loss": 0.5919,
+      "step": 6115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871569128142521e-05,
+      "loss": 0.4405,
+      "step": 6116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715202584295164e-05,
+      "loss": 0.5467,
+      "step": 6117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8714713800587665e-05,
+      "loss": 0.4789,
+      "step": 6118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8714224930307565e-05,
+      "loss": 0.4971,
+      "step": 6119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713735973459725e-05,
+      "loss": 0.511,
+      "step": 6120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713246930048998e-05,
+      "loss": 0.5867,
+      "step": 6121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712757800080245e-05,
+      "loss": 0.5283,
+      "step": 6122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712268583558324e-05,
+      "loss": 0.4528,
+      "step": 6123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711779280488095e-05,
+      "loss": 0.5231,
+      "step": 6124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711289890874415e-05,
+      "loss": 0.5477,
+      "step": 6125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871080041472215e-05,
+      "loss": 0.5605,
+      "step": 6126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8710310852036168e-05,
+      "loss": 0.5353,
+      "step": 6127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870982120282132e-05,
+      "loss": 0.4768,
+      "step": 6128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870933146708248e-05,
+      "loss": 0.5723,
+      "step": 6129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8708841644824507e-05,
+      "loss": 0.509,
+      "step": 6130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870835173605227e-05,
+      "loss": 0.5067,
+      "step": 6131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707861740770635e-05,
+      "loss": 0.5015,
+      "step": 6132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707371658984468e-05,
+      "loss": 0.6164,
+      "step": 6133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870688149069864e-05,
+      "loss": 0.5639,
+      "step": 6134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8706391235918022e-05,
+      "loss": 0.5712,
+      "step": 6135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870590089464748e-05,
+      "loss": 0.5273,
+      "step": 6136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705410466891885e-05,
+      "loss": 0.6072,
+      "step": 6137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8704919952656116e-05,
+      "loss": 0.5798,
+      "step": 6138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8704429351945035e-05,
+      "loss": 0.5565,
+      "step": 6139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8703938664763525e-05,
+      "loss": 0.4352,
+      "step": 6140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8703447891116458e-05,
+      "loss": 0.5145,
+      "step": 6141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702957031008703e-05,
+      "loss": 0.4871,
+      "step": 6142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702466084445146e-05,
+      "loss": 0.4746,
+      "step": 6143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8701975051430658e-05,
+      "loss": 0.504,
+      "step": 6144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870148393197012e-05,
+      "loss": 0.569,
+      "step": 6145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8700992726068407e-05,
+      "loss": 0.5153,
+      "step": 6146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.87005014337304e-05,
+      "loss": 0.4995,
+      "step": 6147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870001005496098e-05,
+      "loss": 0.5156,
+      "step": 6148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8699518589765035e-05,
+      "loss": 0.5459,
+      "step": 6149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8699027038147435e-05,
+      "loss": 0.4939,
+      "step": 6150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8698535400113075e-05,
+      "loss": 0.5114,
+      "step": 6151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869804367566683e-05,
+      "loss": 0.5343,
+      "step": 6152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8697551864813586e-05,
+      "loss": 0.4917,
+      "step": 6153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869705996755823e-05,
+      "loss": 0.5206,
+      "step": 6154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8696567983905655e-05,
+      "loss": 0.5612,
+      "step": 6155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8696075913860737e-05,
+      "loss": 0.4964,
+      "step": 6156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695583757428376e-05,
+      "loss": 0.5081,
+      "step": 6157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869509151461345e-05,
+      "loss": 0.5813,
+      "step": 6158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694599185420855e-05,
+      "loss": 0.3706,
+      "step": 6159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694106769855484e-05,
+      "loss": 0.4504,
+      "step": 6160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8693614267922222e-05,
+      "loss": 0.492,
+      "step": 6161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869312167962597e-05,
+      "loss": 0.4814,
+      "step": 6162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869262900497161e-05,
+      "loss": 0.5712,
+      "step": 6163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8692136243964048e-05,
+      "loss": 0.5863,
+      "step": 6164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869164339660817e-05,
+      "loss": 0.4304,
+      "step": 6165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8691150462908877e-05,
+      "loss": 0.5176,
+      "step": 6166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690657442871068e-05,
+      "loss": 0.4808,
+      "step": 6167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690164336499635e-05,
+      "loss": 0.541,
+      "step": 6168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8689671143799478e-05,
+      "loss": 0.4637,
+      "step": 6169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8689177864775497e-05,
+      "loss": 0.505,
+      "step": 6170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8688684499432597e-05,
+      "loss": 0.6122,
+      "step": 6171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868819104777567e-05,
+      "loss": 0.5343,
+      "step": 6172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8687697509809625e-05,
+      "loss": 0.5313,
+      "step": 6173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868720388553936e-05,
+      "loss": 0.5675,
+      "step": 6174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8686710174969786e-05,
+      "loss": 0.4882,
+      "step": 6175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.86862163781058e-05,
+      "loss": 0.5589,
+      "step": 6176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868572249495231e-05,
+      "loss": 0.6302,
+      "step": 6177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685228525514225e-05,
+      "loss": 0.4836,
+      "step": 6178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684734469796447e-05,
+      "loss": 0.4618,
+      "step": 6179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684240327803888e-05,
+      "loss": 0.5078,
+      "step": 6180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8683746099541456e-05,
+      "loss": 0.4871,
+      "step": 6181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868325178501406e-05,
+      "loss": 0.5566,
+      "step": 6182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682757384226612e-05,
+      "loss": 0.4969,
+      "step": 6183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682262897184018e-05,
+      "loss": 0.4669,
+      "step": 6184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8681768323891196e-05,
+      "loss": 0.5271,
+      "step": 6185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868127366435306e-05,
+      "loss": 0.5541,
+      "step": 6186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868077891857452e-05,
+      "loss": 0.4501,
+      "step": 6187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680284086560493e-05,
+      "loss": 0.6372,
+      "step": 6188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8679789168315892e-05,
+      "loss": 0.4892,
+      "step": 6189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867929416384564e-05,
+      "loss": 0.4885,
+      "step": 6190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8678799073154646e-05,
+      "loss": 0.6225,
+      "step": 6191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8678303896247837e-05,
+      "loss": 0.5301,
+      "step": 6192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8677808633130122e-05,
+      "loss": 0.6117,
+      "step": 6193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867731328380643e-05,
+      "loss": 0.5424,
+      "step": 6194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8676817848281677e-05,
+      "loss": 0.508,
+      "step": 6195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8676322326560786e-05,
+      "loss": 0.5018,
+      "step": 6196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867582671864868e-05,
+      "loss": 0.5114,
+      "step": 6197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675331024550283e-05,
+      "loss": 0.432,
+      "step": 6198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8674835244270518e-05,
+      "loss": 0.5158,
+      "step": 6199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8674339377814308e-05,
+      "loss": 0.5598,
+      "step": 6200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867384342518658e-05,
+      "loss": 0.6046,
+      "step": 6201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8673347386392268e-05,
+      "loss": 0.5185,
+      "step": 6202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867285126143629e-05,
+      "loss": 0.5855,
+      "step": 6203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8672355050323577e-05,
+      "loss": 0.5289,
+      "step": 6204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867185875305906e-05,
+      "loss": 0.5522,
+      "step": 6205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8671362369647673e-05,
+      "loss": 0.5771,
+      "step": 6206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670865900094342e-05,
+      "loss": 0.4669,
+      "step": 6207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670369344403998e-05,
+      "loss": 0.504,
+      "step": 6208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669872702581577e-05,
+      "loss": 0.6203,
+      "step": 6209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669375974632013e-05,
+      "loss": 0.6157,
+      "step": 6210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8668879160560238e-05,
+      "loss": 0.5207,
+      "step": 6211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8668382260371185e-05,
+      "loss": 0.5088,
+      "step": 6212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8667885274069796e-05,
+      "loss": 0.4856,
+      "step": 6213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866738820166101e-05,
+      "loss": 0.4471,
+      "step": 6214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8666891043149757e-05,
+      "loss": 0.4987,
+      "step": 6215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8666393798540977e-05,
+      "loss": 0.4242,
+      "step": 6216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665896467839613e-05,
+      "loss": 0.539,
+      "step": 6217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665399051050608e-05,
+      "loss": 0.5806,
+      "step": 6218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86649015481789e-05,
+      "loss": 0.4276,
+      "step": 6219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664403959229432e-05,
+      "loss": 0.4992,
+      "step": 6220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8663906284207142e-05,
+      "loss": 0.3829,
+      "step": 6221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8663408523116984e-05,
+      "loss": 0.6386,
+      "step": 6222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8662910675963894e-05,
+      "loss": 0.4943,
+      "step": 6223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866241274275282e-05,
+      "loss": 0.4728,
+      "step": 6224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866191472348871e-05,
+      "loss": 0.5702,
+      "step": 6225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8661416618176515e-05,
+      "loss": 0.5432,
+      "step": 6226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866091842682117e-05,
+      "loss": 0.4821,
+      "step": 6227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8660420149427643e-05,
+      "loss": 0.5246,
+      "step": 6228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8659921786000867e-05,
+      "loss": 0.483,
+      "step": 6229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86594233365458e-05,
+      "loss": 0.4748,
+      "step": 6230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8658924801067397e-05,
+      "loss": 0.5084,
+      "step": 6231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8658426179570607e-05,
+      "loss": 0.4823,
+      "step": 6232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865792747206038e-05,
+      "loss": 0.6378,
+      "step": 6233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8657428678541674e-05,
+      "loss": 0.6372,
+      "step": 6234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656929799019444e-05,
+      "loss": 0.5236,
+      "step": 6235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656430833498643e-05,
+      "loss": 0.4975,
+      "step": 6236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8655931781984236e-05,
+      "loss": 0.5097,
+      "step": 6237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865543264448117e-05,
+      "loss": 0.5108,
+      "step": 6238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865493342099441e-05,
+      "loss": 0.5641,
+      "step": 6239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8654434111528915e-05,
+      "loss": 0.5401,
+      "step": 6240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865393471608964e-05,
+      "loss": 0.4868,
+      "step": 6241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865343523468155e-05,
+      "loss": 0.5554,
+      "step": 6242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865293566730961e-05,
+      "loss": 0.4828,
+      "step": 6243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8652436013978774e-05,
+      "loss": 0.5141,
+      "step": 6244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8651936274694013e-05,
+      "loss": 0.5333,
+      "step": 6245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865143644946029e-05,
+      "loss": 0.562,
+      "step": 6246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865093653828257e-05,
+      "loss": 0.4631,
+      "step": 6247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8650436541165818e-05,
+      "loss": 0.4556,
+      "step": 6248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8649936458115006e-05,
+      "loss": 0.5434,
+      "step": 6249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8649436289135093e-05,
+      "loss": 0.5958,
+      "step": 6250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8648936034231053e-05,
+      "loss": 0.5375,
+      "step": 6251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8648435693407857e-05,
+      "loss": 0.6197,
+      "step": 6252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864793526667047e-05,
+      "loss": 0.6115,
+      "step": 6253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8647434754023868e-05,
+      "loss": 0.5363,
+      "step": 6254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646934155473025e-05,
+      "loss": 0.5544,
+      "step": 6255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646433471022908e-05,
+      "loss": 0.5484,
+      "step": 6256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8645932700678493e-05,
+      "loss": 0.5381,
+      "step": 6257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8645431844444754e-05,
+      "loss": 0.581,
+      "step": 6258
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864493090232667e-05,
+      "loss": 0.6085,
+      "step": 6259
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8644429874329214e-05,
+      "loss": 0.5392,
+      "step": 6260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8643928760457367e-05,
+      "loss": 0.44,
+      "step": 6261
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8643427560716105e-05,
+      "loss": 0.5929,
+      "step": 6262
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86429262751104e-05,
+      "loss": 0.5761,
+      "step": 6263
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8642424903645247e-05,
+      "loss": 0.5461,
+      "step": 6264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8641923446325615e-05,
+      "loss": 0.4811,
+      "step": 6265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8641421903156485e-05,
+      "loss": 0.5433,
+      "step": 6266
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640920274142845e-05,
+      "loss": 0.5982,
+      "step": 6267
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640418559289676e-05,
+      "loss": 0.4733,
+      "step": 6268
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8639916758601963e-05,
+      "loss": 0.5057,
+      "step": 6269
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.863941487208469e-05,
+      "loss": 0.6992,
+      "step": 6270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.863891289974284e-05,
+      "loss": 0.5632,
+      "step": 6271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8638410841581408e-05,
+      "loss": 0.5001,
+      "step": 6272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863790869760537e-05,
+      "loss": 0.5155,
+      "step": 6273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8637406467819724e-05,
+      "loss": 0.5391,
+      "step": 6274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8636904152229455e-05,
+      "loss": 0.5063,
+      "step": 6275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8636401750839552e-05,
+      "loss": 0.564,
+      "step": 6276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863589926365501e-05,
+      "loss": 0.4529,
+      "step": 6277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863539669068082e-05,
+      "loss": 0.4776,
+      "step": 6278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634894031921965e-05,
+      "loss": 0.4475,
+      "step": 6279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634391287383457e-05,
+      "loss": 0.4528,
+      "step": 6280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863388845707027e-05,
+      "loss": 0.4875,
+      "step": 6281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8633385540987413e-05,
+      "loss": 0.5474,
+      "step": 6282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632882539139876e-05,
+      "loss": 0.592,
+      "step": 6283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632379451532658e-05,
+      "loss": 0.478,
+      "step": 6284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8631876278170758e-05,
+      "loss": 0.589,
+      "step": 6285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8631373019059172e-05,
+      "loss": 0.6186,
+      "step": 6286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630869674202898e-05,
+      "loss": 0.5121,
+      "step": 6287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863036624360694e-05,
+      "loss": 0.491,
+      "step": 6288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8629862727276297e-05,
+      "loss": 0.5748,
+      "step": 6289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8629359125215972e-05,
+      "loss": 0.4582,
+      "step": 6290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862885543743097e-05,
+      "loss": 0.5684,
+      "step": 6291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862835166392629e-05,
+      "loss": 0.5913,
+      "step": 6292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8627847804706938e-05,
+      "loss": 0.4855,
+      "step": 6293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862734385977792e-05,
+      "loss": 0.5227,
+      "step": 6294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8626839829144244e-05,
+      "loss": 0.5076,
+      "step": 6295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862633571281091e-05,
+      "loss": 0.5794,
+      "step": 6296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625831510782936e-05,
+      "loss": 0.5923,
+      "step": 6297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625327223065327e-05,
+      "loss": 0.5356,
+      "step": 6298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8624822849663088e-05,
+      "loss": 0.5972,
+      "step": 6299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8624318390581234e-05,
+      "loss": 0.4933,
+      "step": 6300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8623813845824775e-05,
+      "loss": 0.4435,
+      "step": 6301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8623309215398726e-05,
+      "loss": 0.5368,
+      "step": 6302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86228044993081e-05,
+      "loss": 0.4747,
+      "step": 6303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86222996975579e-05,
+      "loss": 0.5111,
+      "step": 6304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8621794810153155e-05,
+      "loss": 0.5169,
+      "step": 6305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8621289837098874e-05,
+      "loss": 0.5215,
+      "step": 6306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620784778400072e-05,
+      "loss": 0.5717,
+      "step": 6307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620279634061774e-05,
+      "loss": 0.6178,
+      "step": 6308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8619774404088987e-05,
+      "loss": 0.4718,
+      "step": 6309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861926908848674e-05,
+      "loss": 0.541,
+      "step": 6310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618763687260047e-05,
+      "loss": 0.562,
+      "step": 6311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618258200413927e-05,
+      "loss": 0.5779,
+      "step": 6312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861775262795341e-05,
+      "loss": 0.4335,
+      "step": 6313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861724696988351e-05,
+      "loss": 0.5083,
+      "step": 6314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8616741226209256e-05,
+      "loss": 0.5651,
+      "step": 6315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8616235396935667e-05,
+      "loss": 0.5873,
+      "step": 6316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861572948206777e-05,
+      "loss": 0.571,
+      "step": 6317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8615223481610595e-05,
+      "loss": 0.5596,
+      "step": 6318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8614717395569163e-05,
+      "loss": 0.5777,
+      "step": 6319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86142112239485e-05,
+      "loss": 0.5216,
+      "step": 6320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861370496675364e-05,
+      "loss": 0.505,
+      "step": 6321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613198623989614e-05,
+      "loss": 0.5721,
+      "step": 6322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861269219566144e-05,
+      "loss": 0.5712,
+      "step": 6323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8612185681774163e-05,
+      "loss": 0.4808,
+      "step": 6324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8611679082332808e-05,
+      "loss": 0.4404,
+      "step": 6325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8611172397342406e-05,
+      "loss": 0.5099,
+      "step": 6326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610665626807995e-05,
+      "loss": 0.5733,
+      "step": 6327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610158770734607e-05,
+      "loss": 0.5084,
+      "step": 6328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609651829127276e-05,
+      "loss": 0.5511,
+      "step": 6329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609144801991037e-05,
+      "loss": 0.5662,
+      "step": 6330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860863768933093e-05,
+      "loss": 0.4833,
+      "step": 6331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8608130491151994e-05,
+      "loss": 0.4881,
+      "step": 6332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860762320745926e-05,
+      "loss": 0.5086,
+      "step": 6333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860711583825778e-05,
+      "loss": 0.5212,
+      "step": 6334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8606608383552583e-05,
+      "loss": 0.5284,
+      "step": 6335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8606100843348714e-05,
+      "loss": 0.445,
+      "step": 6336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605593217651216e-05,
+      "loss": 0.5403,
+      "step": 6337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605085506465128e-05,
+      "loss": 0.5658,
+      "step": 6338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.86045777097955e-05,
+      "loss": 0.4898,
+      "step": 6339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8604069827647367e-05,
+      "loss": 0.5106,
+      "step": 6340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8603561860025785e-05,
+      "loss": 0.6303,
+      "step": 6341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8603053806935793e-05,
+      "loss": 0.5607,
+      "step": 6342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8602545668382443e-05,
+      "loss": 0.6123,
+      "step": 6343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8602037444370777e-05,
+      "loss": 0.4994,
+      "step": 6344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860152913490585e-05,
+      "loss": 0.4704,
+      "step": 6345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601020739992708e-05,
+      "loss": 0.5049,
+      "step": 6346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.86005122596364e-05,
+      "loss": 0.4997,
+      "step": 6347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860000369384198e-05,
+      "loss": 0.5631,
+      "step": 6348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85994950426145e-05,
+      "loss": 0.5378,
+      "step": 6349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859898630595901e-05,
+      "loss": 0.5573,
+      "step": 6350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8598477483880567e-05,
+      "loss": 0.4257,
+      "step": 6351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859796857638423e-05,
+      "loss": 0.5987,
+      "step": 6352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8597459583475045e-05,
+      "loss": 0.5083,
+      "step": 6353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859695050515807e-05,
+      "loss": 0.5221,
+      "step": 6354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859644134143837e-05,
+      "loss": 0.5634,
+      "step": 6355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8595932092320998e-05,
+      "loss": 0.5053,
+      "step": 6356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859542275781101e-05,
+      "loss": 0.528,
+      "step": 6357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859491333791347e-05,
+      "loss": 0.5696,
+      "step": 6358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8594403832633436e-05,
+      "loss": 0.4966,
+      "step": 6359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593894241975973e-05,
+      "loss": 0.4625,
+      "step": 6360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859338456594614e-05,
+      "loss": 0.519,
+      "step": 6361
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8592874804549002e-05,
+      "loss": 0.514,
+      "step": 6362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859236495778962e-05,
+      "loss": 0.509,
+      "step": 6363
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859185502567306e-05,
+      "loss": 0.5197,
+      "step": 6364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8591345008204394e-05,
+      "loss": 0.5335,
+      "step": 6365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859083490538868e-05,
+      "loss": 0.546,
+      "step": 6366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859032471723099e-05,
+      "loss": 0.5787,
+      "step": 6367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858981444373639e-05,
+      "loss": 0.5265,
+      "step": 6368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8589304084909953e-05,
+      "loss": 0.5339,
+      "step": 6369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588793640756743e-05,
+      "loss": 0.6006,
+      "step": 6370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588283111281836e-05,
+      "loss": 0.4293,
+      "step": 6371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85877724964903e-05,
+      "loss": 0.5523,
+      "step": 6372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8587261796387208e-05,
+      "loss": 0.5323,
+      "step": 6373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586751010977634e-05,
+      "loss": 0.6417,
+      "step": 6374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586240140266655e-05,
+      "loss": 0.5307,
+      "step": 6375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8585729184259343e-05,
+      "loss": 0.5121,
+      "step": 6376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8585218142960776e-05,
+      "loss": 0.5865,
+      "step": 6377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584707016376024e-05,
+      "loss": 0.4623,
+      "step": 6378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584195804510176e-05,
+      "loss": 0.5233,
+      "step": 6379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85836845073683e-05,
+      "loss": 0.4764,
+      "step": 6380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583173124955477e-05,
+      "loss": 0.6329,
+      "step": 6381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8582661657276795e-05,
+      "loss": 0.4383,
+      "step": 6382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8582150104337325e-05,
+      "loss": 0.508,
+      "step": 6383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858163846614216e-05,
+      "loss": 0.5128,
+      "step": 6384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858112674269637e-05,
+      "loss": 0.525,
+      "step": 6385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8580614934005045e-05,
+      "loss": 0.5304,
+      "step": 6386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8580103040073268e-05,
+      "loss": 0.5275,
+      "step": 6387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8579591060906127e-05,
+      "loss": 0.5104,
+      "step": 6388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8579078996508705e-05,
+      "loss": 0.4765,
+      "step": 6389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857856684688609e-05,
+      "loss": 0.4842,
+      "step": 6390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857805461204337e-05,
+      "loss": 0.4101,
+      "step": 6391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577542291985634e-05,
+      "loss": 0.4835,
+      "step": 6392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577029886717967e-05,
+      "loss": 0.5581,
+      "step": 6393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8576517396245466e-05,
+      "loss": 0.5652,
+      "step": 6394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8576004820573217e-05,
+      "loss": 0.601,
+      "step": 6395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8575492159706314e-05,
+      "loss": 0.607,
+      "step": 6396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857497941364985e-05,
+      "loss": 0.5067,
+      "step": 6397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857446658240892e-05,
+      "loss": 0.4811,
+      "step": 6398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8573953665988617e-05,
+      "loss": 0.6245,
+      "step": 6399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8573440664394036e-05,
+      "loss": 0.5557,
+      "step": 6400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8572927577630273e-05,
+      "loss": 0.6296,
+      "step": 6401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8572414405702422e-05,
+      "loss": 0.5107,
+      "step": 6402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857190114861559e-05,
+      "loss": 0.5433,
+      "step": 6403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857138780637487e-05,
+      "loss": 0.4832,
+      "step": 6404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8570874378985357e-05,
+      "loss": 0.5236,
+      "step": 6405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8570360866452158e-05,
+      "loss": 0.431,
+      "step": 6406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856984726878037e-05,
+      "loss": 0.6585,
+      "step": 6407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85693335859751e-05,
+      "loss": 0.4934,
+      "step": 6408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8568819818041452e-05,
+      "loss": 0.4713,
+      "step": 6409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856830596498452e-05,
+      "loss": 0.5226,
+      "step": 6410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856779202680942e-05,
+      "loss": 0.5073,
+      "step": 6411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856727800352125e-05,
+      "loss": 0.5556,
+      "step": 6412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856676389512512e-05,
+      "loss": 0.4804,
+      "step": 6413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8566249701626134e-05,
+      "loss": 0.4753,
+      "step": 6414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8565735423029406e-05,
+      "loss": 0.704,
+      "step": 6415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8565221059340035e-05,
+      "loss": 0.5416,
+      "step": 6416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8564706610563143e-05,
+      "loss": 0.5524,
+      "step": 6417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8564192076703826e-05,
+      "loss": 0.4955,
+      "step": 6418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856367745776721e-05,
+      "loss": 0.4677,
+      "step": 6419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85631627537584e-05,
+      "loss": 0.5049,
+      "step": 6420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8562647964682506e-05,
+      "loss": 0.4341,
+      "step": 6421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856213309054465e-05,
+      "loss": 0.4915,
+      "step": 6422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8561618131349942e-05,
+      "loss": 0.5319,
+      "step": 6423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8561103087103497e-05,
+      "loss": 0.4511,
+      "step": 6424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560587957810434e-05,
+      "loss": 0.6246,
+      "step": 6425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856007274347587e-05,
+      "loss": 0.5851,
+      "step": 6426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855955744410492e-05,
+      "loss": 0.4015,
+      "step": 6427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8559042059702705e-05,
+      "loss": 0.4241,
+      "step": 6428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8558526590274347e-05,
+      "loss": 0.4778,
+      "step": 6429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8558011035824965e-05,
+      "loss": 0.501,
+      "step": 6430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557495396359682e-05,
+      "loss": 0.6158,
+      "step": 6431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8556979671883618e-05,
+      "loss": 0.5209,
+      "step": 6432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8556463862401896e-05,
+      "loss": 0.4039,
+      "step": 6433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855594796791964e-05,
+      "loss": 0.499,
+      "step": 6434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8555431988441977e-05,
+      "loss": 0.5879,
+      "step": 6435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554915923974034e-05,
+      "loss": 0.5079,
+      "step": 6436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554399774520934e-05,
+      "loss": 0.5685,
+      "step": 6437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8553883540087806e-05,
+      "loss": 0.4827,
+      "step": 6438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8553367220679778e-05,
+      "loss": 0.4763,
+      "step": 6439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855285081630198e-05,
+      "loss": 0.5408,
+      "step": 6440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855233432695954e-05,
+      "loss": 0.483,
+      "step": 6441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551817752657594e-05,
+      "loss": 0.5855,
+      "step": 6442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551301093401265e-05,
+      "loss": 0.5283,
+      "step": 6443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8550784349195694e-05,
+      "loss": 0.621,
+      "step": 6444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855026752004601e-05,
+      "loss": 0.5676,
+      "step": 6445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549750605957348e-05,
+      "loss": 0.5238,
+      "step": 6446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549233606934844e-05,
+      "loss": 0.5639,
+      "step": 6447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548716522983632e-05,
+      "loss": 0.4613,
+      "step": 6448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548199354108853e-05,
+      "loss": 0.5357,
+      "step": 6449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854768210031564e-05,
+      "loss": 0.5747,
+      "step": 6450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547164761609128e-05,
+      "loss": 0.4981,
+      "step": 6451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8546647337994465e-05,
+      "loss": 0.5359,
+      "step": 6452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8546129829476788e-05,
+      "loss": 0.4736,
+      "step": 6453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8545612236061235e-05,
+      "loss": 0.6099,
+      "step": 6454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854509455775295e-05,
+      "loss": 0.4788,
+      "step": 6455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8544576794557077e-05,
+      "loss": 0.6707,
+      "step": 6456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854405894647876e-05,
+      "loss": 0.5155,
+      "step": 6457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854354101352314e-05,
+      "loss": 0.5087,
+      "step": 6458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8543022995695366e-05,
+      "loss": 0.4968,
+      "step": 6459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854250489300058e-05,
+      "loss": 0.4718,
+      "step": 6460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8541986705443927e-05,
+      "loss": 0.4849,
+      "step": 6461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8541468433030562e-05,
+      "loss": 0.6304,
+      "step": 6462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854095007576563e-05,
+      "loss": 0.5084,
+      "step": 6463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854043163365428e-05,
+      "loss": 0.5097,
+      "step": 6464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8539913106701663e-05,
+      "loss": 0.5838,
+      "step": 6465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.853939449491293e-05,
+      "loss": 0.5476,
+      "step": 6466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8538875798293226e-05,
+      "loss": 0.4953,
+      "step": 6467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853835701684772e-05,
+      "loss": 0.4677,
+      "step": 6468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853783815058155e-05,
+      "loss": 0.5876,
+      "step": 6469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8537319199499876e-05,
+      "loss": 0.5209,
+      "step": 6470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8536800163607858e-05,
+      "loss": 0.5731,
+      "step": 6471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8536281042910644e-05,
+      "loss": 0.5498,
+      "step": 6472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85357618374134e-05,
+      "loss": 0.4888,
+      "step": 6473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8535242547121274e-05,
+      "loss": 0.5138,
+      "step": 6474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853472317203943e-05,
+      "loss": 0.5494,
+      "step": 6475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534203712173028e-05,
+      "loss": 0.5812,
+      "step": 6476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8533684167527225e-05,
+      "loss": 0.6014,
+      "step": 6477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8533164538107186e-05,
+      "loss": 0.513,
+      "step": 6478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853264482391807e-05,
+      "loss": 0.4834,
+      "step": 6479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8532125024965043e-05,
+      "loss": 0.5265,
+      "step": 6480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8531605141253265e-05,
+      "loss": 0.5962,
+      "step": 6481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8531085172787903e-05,
+      "loss": 0.5934,
+      "step": 6482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530565119574123e-05,
+      "loss": 0.4819,
+      "step": 6483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853004498161709e-05,
+      "loss": 0.4775,
+      "step": 6484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8529524758921973e-05,
+      "loss": 0.4591,
+      "step": 6485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8529004451493934e-05,
+      "loss": 0.5693,
+      "step": 6486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852848405933815e-05,
+      "loss": 0.5973,
+      "step": 6487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527963582459785e-05,
+      "loss": 0.3928,
+      "step": 6488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527443020864013e-05,
+      "loss": 0.5555,
+      "step": 6489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526922374556e-05,
+      "loss": 0.5385,
+      "step": 6490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526401643540924e-05,
+      "loss": 0.4989,
+      "step": 6491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8525880827823955e-05,
+      "loss": 0.463,
+      "step": 6492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8525359927410263e-05,
+      "loss": 0.5169,
+      "step": 6493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8524838942305034e-05,
+      "loss": 0.537,
+      "step": 6494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8524317872513434e-05,
+      "loss": 0.6039,
+      "step": 6495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852379671804064e-05,
+      "loss": 0.5671,
+      "step": 6496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8523275478891833e-05,
+      "loss": 0.5898,
+      "step": 6497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522754155072186e-05,
+      "loss": 0.5164,
+      "step": 6498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522232746586882e-05,
+      "loss": 0.5953,
+      "step": 6499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521711253441104e-05,
+      "loss": 0.3659,
+      "step": 6500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521189675640023e-05,
+      "loss": 0.4007,
+      "step": 6501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8520668013188828e-05,
+      "loss": 0.3946,
+      "step": 6502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85201462660927e-05,
+      "loss": 0.498,
+      "step": 6503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851962443435682e-05,
+      "loss": 0.4275,
+      "step": 6504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851910251798637e-05,
+      "loss": 0.4509,
+      "step": 6505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851858051698654e-05,
+      "loss": 0.3681,
+      "step": 6506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8518058431362513e-05,
+      "loss": 0.464,
+      "step": 6507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8517536261119473e-05,
+      "loss": 0.4236,
+      "step": 6508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8517014006262613e-05,
+      "loss": 0.3967,
+      "step": 6509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8516491666797116e-05,
+      "loss": 0.3668,
+      "step": 6510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515969242728172e-05,
+      "loss": 0.3938,
+      "step": 6511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515446734060973e-05,
+      "loss": 0.4019,
+      "step": 6512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851492414080071e-05,
+      "loss": 0.4674,
+      "step": 6513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8514401462952572e-05,
+      "loss": 0.4057,
+      "step": 6514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8513878700521752e-05,
+      "loss": 0.4148,
+      "step": 6515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8513355853513442e-05,
+      "loss": 0.399,
+      "step": 6516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851283292193284e-05,
+      "loss": 0.3917,
+      "step": 6517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8512309905785135e-05,
+      "loss": 0.4036,
+      "step": 6518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8511786805075526e-05,
+      "loss": 0.344,
+      "step": 6519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8511263619809212e-05,
+      "loss": 0.3949,
+      "step": 6520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8510740349991386e-05,
+      "loss": 0.3982,
+      "step": 6521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8510216995627248e-05,
+      "loss": 0.3853,
+      "step": 6522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8509693556722e-05,
+      "loss": 0.3828,
+      "step": 6523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8509170033280833e-05,
+      "loss": 0.3671,
+      "step": 6524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508646425308958e-05,
+      "loss": 0.4172,
+      "step": 6525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508122732811567e-05,
+      "loss": 0.3612,
+      "step": 6526
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850759895579387e-05,
+      "loss": 0.4315,
+      "step": 6527
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850707509426107e-05,
+      "loss": 0.4385,
+      "step": 6528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850655114821837e-05,
+      "loss": 0.4103,
+      "step": 6529
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850602711767097e-05,
+      "loss": 0.4238,
+      "step": 6530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8505503002624083e-05,
+      "loss": 0.4307,
+      "step": 6531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850497880308291e-05,
+      "loss": 0.3687,
+      "step": 6532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850445451905266e-05,
+      "loss": 0.4218,
+      "step": 6533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8503930150538544e-05,
+      "loss": 0.3839,
+      "step": 6534
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8503405697545767e-05,
+      "loss": 0.4077,
+      "step": 6535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850288116007954e-05,
+      "loss": 0.4007,
+      "step": 6536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8502356538145077e-05,
+      "loss": 0.3789,
+      "step": 6537
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8501831831747585e-05,
+      "loss": 0.3848,
+      "step": 6538
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8501307040892282e-05,
+      "loss": 0.4305,
+      "step": 6539
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500782165584375e-05,
+      "loss": 0.3554,
+      "step": 6540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500257205829083e-05,
+      "loss": 0.3818,
+      "step": 6541
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499732161631616e-05,
+      "loss": 0.4129,
+      "step": 6542
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499207032997198e-05,
+      "loss": 0.4358,
+      "step": 6543
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8498681819931038e-05,
+      "loss": 0.4026,
+      "step": 6544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8498156522438356e-05,
+      "loss": 0.475,
+      "step": 6545
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8497631140524374e-05,
+      "loss": 0.3634,
+      "step": 6546
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8497105674194302e-05,
+      "loss": 0.4099,
+      "step": 6547
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849658012345337e-05,
+      "loss": 0.4017,
+      "step": 6548
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849605448830679e-05,
+      "loss": 0.367,
+      "step": 6549
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849552876875979e-05,
+      "loss": 0.4397,
+      "step": 6550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8495002964817595e-05,
+      "loss": 0.4564,
+      "step": 6551
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8494477076485423e-05,
+      "loss": 0.4931,
+      "step": 6552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8493951103768497e-05,
+      "loss": 0.4305,
+      "step": 6553
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8493425046672047e-05,
+      "loss": 0.3967,
+      "step": 6554
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8492898905201297e-05,
+      "loss": 0.2729,
+      "step": 6555
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8492372679361467e-05,
+      "loss": 0.39,
+      "step": 6556
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8491846369157796e-05,
+      "loss": 0.4311,
+      "step": 6557
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8491319974595505e-05,
+      "loss": 0.4344,
+      "step": 6558
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8490793495679825e-05,
+      "loss": 0.3948,
+      "step": 6559
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8490266932415988e-05,
+      "loss": 0.3558,
+      "step": 6560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8489740284809224e-05,
+      "loss": 0.4384,
+      "step": 6561
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8489213552864762e-05,
+      "loss": 0.5066,
+      "step": 6562
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848868673658784e-05,
+      "loss": 0.3984,
+      "step": 6563
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8488159835983682e-05,
+      "loss": 0.4475,
+      "step": 6564
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8487632851057533e-05,
+      "loss": 0.4556,
+      "step": 6565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8487105781814624e-05,
+      "loss": 0.3716,
+      "step": 6566
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848657862826019e-05,
+      "loss": 0.3685,
+      "step": 6567
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8486051390399466e-05,
+      "loss": 0.4538,
+      "step": 6568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8485524068237693e-05,
+      "loss": 0.3746,
+      "step": 6569
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848499666178011e-05,
+      "loss": 0.4845,
+      "step": 6570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8484469171031948e-05,
+      "loss": 0.3877,
+      "step": 6571
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8483941595998463e-05,
+      "loss": 0.3507,
+      "step": 6572
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8483413936684882e-05,
+      "loss": 0.3925,
+      "step": 6573
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8482886193096448e-05,
+      "loss": 0.3902,
+      "step": 6574
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8482358365238414e-05,
+      "loss": 0.4266,
+      "step": 6575
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8481830453116015e-05,
+      "loss": 0.4645,
+      "step": 6576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8481302456734494e-05,
+      "loss": 0.3765,
+      "step": 6577
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84807743760991e-05,
+      "loss": 0.411,
+      "step": 6578
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8480246211215077e-05,
+      "loss": 0.4392,
+      "step": 6579
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8479717962087674e-05,
+      "loss": 0.3524,
+      "step": 6580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8479189628722142e-05,
+      "loss": 0.3729,
+      "step": 6581
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8478661211123717e-05,
+      "loss": 0.3985,
+      "step": 6582
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847813270929766e-05,
+      "loss": 0.4267,
+      "step": 6583
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847760412324922e-05,
+      "loss": 0.3865,
+      "step": 6584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8477075452983643e-05,
+      "loss": 0.4687,
+      "step": 6585
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8476546698506185e-05,
+      "loss": 0.3438,
+      "step": 6586
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8476017859822095e-05,
+      "loss": 0.4895,
+      "step": 6587
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847548893693663e-05,
+      "loss": 0.4173,
+      "step": 6588
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8474959929855046e-05,
+      "loss": 0.3779,
+      "step": 6589
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847443083858259e-05,
+      "loss": 0.4232,
+      "step": 6590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8473901663124522e-05,
+      "loss": 0.5074,
+      "step": 6591
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8473372403486105e-05,
+      "loss": 0.451,
+      "step": 6592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8472843059672588e-05,
+      "loss": 0.3517,
+      "step": 6593
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847231363168924e-05,
+      "loss": 0.4596,
+      "step": 6594
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847178411954131e-05,
+      "loss": 0.3577,
+      "step": 6595
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847125452323406e-05,
+      "loss": 0.3987,
+      "step": 6596
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8470724842772753e-05,
+      "loss": 0.3995,
+      "step": 6597
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8470195078162653e-05,
+      "loss": 0.3763,
+      "step": 6598
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846966522940902e-05,
+      "loss": 0.4249,
+      "step": 6599
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469135296517115e-05,
+      "loss": 0.4336,
+      "step": 6600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468605279492207e-05,
+      "loss": 0.4336,
+      "step": 6601
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468075178339563e-05,
+      "loss": 0.4011,
+      "step": 6602
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8467544993064442e-05,
+      "loss": 0.3879,
+      "step": 6603
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846701472367212e-05,
+      "loss": 0.4064,
+      "step": 6604
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8466484370167855e-05,
+      "loss": 0.4433,
+      "step": 6605
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846595393255692e-05,
+      "loss": 0.4295,
+      "step": 6606
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8465423410844584e-05,
+      "loss": 0.4261,
+      "step": 6607
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846489280503612e-05,
+      "loss": 0.4037,
+      "step": 6608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8464362115136797e-05,
+      "loss": 0.3662,
+      "step": 6609
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8463831341151887e-05,
+      "loss": 0.4461,
+      "step": 6610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8463300483086657e-05,
+      "loss": 0.401,
+      "step": 6611
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846276954094639e-05,
+      "loss": 0.409,
+      "step": 6612
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846223851473636e-05,
+      "loss": 0.3955,
+      "step": 6613
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461707404461833e-05,
+      "loss": 0.4226,
+      "step": 6614
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461176210128095e-05,
+      "loss": 0.3596,
+      "step": 6615
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8460644931740414e-05,
+      "loss": 0.3744,
+      "step": 6616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8460113569304073e-05,
+      "loss": 0.3226,
+      "step": 6617
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8459582122824354e-05,
+      "loss": 0.4266,
+      "step": 6618
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8459050592306532e-05,
+      "loss": 0.3645,
+      "step": 6619
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8458518977755883e-05,
+      "loss": 0.3273,
+      "step": 6620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8457987279177695e-05,
+      "loss": 0.4273,
+      "step": 6621
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845745549657725e-05,
+      "loss": 0.4447,
+      "step": 6622
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8456923629959825e-05,
+      "loss": 0.4052,
+      "step": 6623
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845639167933071e-05,
+      "loss": 0.4226,
+      "step": 6624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8455859644695186e-05,
+      "loss": 0.3717,
+      "step": 6625
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8455327526058536e-05,
+      "loss": 0.3748,
+      "step": 6626
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454795323426054e-05,
+      "loss": 0.4048,
+      "step": 6627
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845426303680302e-05,
+      "loss": 0.3553,
+      "step": 6628
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845373066619472e-05,
+      "loss": 0.5048,
+      "step": 6629
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8453198211606455e-05,
+      "loss": 0.4172,
+      "step": 6630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8452665673043494e-05,
+      "loss": 0.4121,
+      "step": 6631
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845213305051115e-05,
+      "loss": 0.4116,
+      "step": 6632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8451600344014694e-05,
+      "loss": 0.3939,
+      "step": 6633
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8451067553559433e-05,
+      "loss": 0.3952,
+      "step": 6634
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845053467915065e-05,
+      "loss": 0.4115,
+      "step": 6635
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8450001720793644e-05,
+      "loss": 0.3568,
+      "step": 6636
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844946867849371e-05,
+      "loss": 0.4063,
+      "step": 6637
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844893555225614e-05,
+      "loss": 0.4566,
+      "step": 6638
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844840234208623e-05,
+      "loss": 0.4577,
+      "step": 6639
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8447869047989274e-05,
+      "loss": 0.3906,
+      "step": 6640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844733566997058e-05,
+      "loss": 0.3667,
+      "step": 6641
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844680220803544e-05,
+      "loss": 0.4355,
+      "step": 6642
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8446268662189154e-05,
+      "loss": 0.4258,
+      "step": 6643
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844573503243702e-05,
+      "loss": 0.4648,
+      "step": 6644
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445201318784343e-05,
+      "loss": 0.4054,
+      "step": 6645
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8444667521236425e-05,
+      "loss": 0.3757,
+      "step": 6646
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8444133639798566e-05,
+      "loss": 0.3836,
+      "step": 6647
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844359967447607e-05,
+      "loss": 0.4546,
+      "step": 6648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844306562527424e-05,
+      "loss": 0.3432,
+      "step": 6649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8442531492198384e-05,
+      "loss": 0.4296,
+      "step": 6650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844199727525381e-05,
+      "loss": 0.3563,
+      "step": 6651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8441462974445824e-05,
+      "loss": 0.438,
+      "step": 6652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8440928589779732e-05,
+      "loss": 0.4176,
+      "step": 6653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844039412126084e-05,
+      "loss": 0.4105,
+      "step": 6654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439859568894464e-05,
+      "loss": 0.4513,
+      "step": 6655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439324932685905e-05,
+      "loss": 0.407,
+      "step": 6656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8438790212640486e-05,
+      "loss": 0.3759,
+      "step": 6657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8438255408763514e-05,
+      "loss": 0.34,
+      "step": 6658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8437720521060293e-05,
+      "loss": 0.3644,
+      "step": 6659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843718554953615e-05,
+      "loss": 0.3834,
+      "step": 6660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843665049419639e-05,
+      "loss": 0.3874,
+      "step": 6661
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8436115355046337e-05,
+      "loss": 0.3069,
+      "step": 6662
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8435580132091297e-05,
+      "loss": 0.3625,
+      "step": 6663
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8435044825336597e-05,
+      "loss": 0.4376,
+      "step": 6664
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434509434787544e-05,
+      "loss": 0.4325,
+      "step": 6665
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8433973960449466e-05,
+      "loss": 0.4207,
+      "step": 6666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8433438402327676e-05,
+      "loss": 0.5426,
+      "step": 6667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8432902760427498e-05,
+      "loss": 0.3735,
+      "step": 6668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8432367034754256e-05,
+      "loss": 0.4017,
+      "step": 6669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8431831225313264e-05,
+      "loss": 0.3616,
+      "step": 6670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8431295332109852e-05,
+      "loss": 0.4259,
+      "step": 6671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843075935514934e-05,
+      "loss": 0.3599,
+      "step": 6672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843022329443705e-05,
+      "loss": 0.4194,
+      "step": 6673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842968714997831e-05,
+      "loss": 0.3555,
+      "step": 6674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8429150921778448e-05,
+      "loss": 0.4843,
+      "step": 6675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8428614609842787e-05,
+      "loss": 0.4321,
+      "step": 6676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842807821417666e-05,
+      "loss": 0.3316,
+      "step": 6677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842754173478539e-05,
+      "loss": 0.4211,
+      "step": 6678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842700517167431e-05,
+      "loss": 0.3949,
+      "step": 6679
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842646852484875e-05,
+      "loss": 0.4572,
+      "step": 6680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842593179431404e-05,
+      "loss": 0.4096,
+      "step": 6681
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8425394980075513e-05,
+      "loss": 0.4569,
+      "step": 6682
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.84248580821385e-05,
+      "loss": 0.4517,
+      "step": 6683
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8424321100508336e-05,
+      "loss": 0.3434,
+      "step": 6684
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8423784035190354e-05,
+      "loss": 0.3561,
+      "step": 6685
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842324688618989e-05,
+      "loss": 0.3786,
+      "step": 6686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8422709653512286e-05,
+      "loss": 0.3625,
+      "step": 6687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8422172337162865e-05,
+      "loss": 0.4745,
+      "step": 6688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842163493714698e-05,
+      "loss": 0.2902,
+      "step": 6689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421097453469958e-05,
+      "loss": 0.4105,
+      "step": 6690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8420559886137144e-05,
+      "loss": 0.4483,
+      "step": 6691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842002223515388e-05,
+      "loss": 0.4514,
+      "step": 6692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.84194845005255e-05,
+      "loss": 0.4144,
+      "step": 6693
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841894668225735e-05,
+      "loss": 0.4031,
+      "step": 6694
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8418408780354776e-05,
+      "loss": 0.4022,
+      "step": 6695
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8417870794823117e-05,
+      "loss": 0.4486,
+      "step": 6696
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8417332725667717e-05,
+      "loss": 0.3425,
+      "step": 6697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841679457289393e-05,
+      "loss": 0.365,
+      "step": 6698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8416256336507086e-05,
+      "loss": 0.4649,
+      "step": 6699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8415718016512545e-05,
+      "loss": 0.4977,
+      "step": 6700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841517961291565e-05,
+      "loss": 0.3987,
+      "step": 6701
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841464112572175e-05,
+      "loss": 0.3824,
+      "step": 6702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8414102554936196e-05,
+      "loss": 0.3681,
+      "step": 6703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8413563900564337e-05,
+      "loss": 0.465,
+      "step": 6704
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8413025162611525e-05,
+      "loss": 0.4335,
+      "step": 6705
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8412486341083107e-05,
+      "loss": 0.4282,
+      "step": 6706
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841194743598444e-05,
+      "loss": 0.3579,
+      "step": 6707
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411408447320878e-05,
+      "loss": 0.4536,
+      "step": 6708
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8410869375097772e-05,
+      "loss": 0.3633,
+      "step": 6709
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8410330219320484e-05,
+      "loss": 0.3843,
+      "step": 6710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8409790979994362e-05,
+      "loss": 0.3707,
+      "step": 6711
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840925165712477e-05,
+      "loss": 0.4368,
+      "step": 6712
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408712250717058e-05,
+      "loss": 0.4302,
+      "step": 6713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408172760776588e-05,
+      "loss": 0.4485,
+      "step": 6714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8407633187308722e-05,
+      "loss": 0.4468,
+      "step": 6715
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8407093530318817e-05,
+      "loss": 0.4194,
+      "step": 6716
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8406553789812235e-05,
+      "loss": 0.3995,
+      "step": 6717
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8406013965794338e-05,
+      "loss": 0.4171,
+      "step": 6718
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840547405827049e-05,
+      "loss": 0.4805,
+      "step": 6719
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840493406724605e-05,
+      "loss": 0.4548,
+      "step": 6720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8404393992726386e-05,
+      "loss": 0.4434,
+      "step": 6721
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840385383471686e-05,
+      "loss": 0.3754,
+      "step": 6722
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8403313593222844e-05,
+      "loss": 0.4648,
+      "step": 6723
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8402773268249702e-05,
+      "loss": 0.4371,
+      "step": 6724
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.84022328598028e-05,
+      "loss": 0.4062,
+      "step": 6725
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8401692367887505e-05,
+      "loss": 0.4646,
+      "step": 6726
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.840115179250919e-05,
+      "loss": 0.438,
+      "step": 6727
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8400611133673222e-05,
+      "loss": 0.405,
+      "step": 6728
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8400070391384975e-05,
+      "loss": 0.3691,
+      "step": 6729
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839952956564982e-05,
+      "loss": 0.401,
+      "step": 6730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398988656473128e-05,
+      "loss": 0.4246,
+      "step": 6731
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398447663860272e-05,
+      "loss": 0.3631,
+      "step": 6732
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839790658781663e-05,
+      "loss": 0.3399,
+      "step": 6733
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8397365428347572e-05,
+      "loss": 0.5016,
+      "step": 6734
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839682418545848e-05,
+      "loss": 0.4542,
+      "step": 6735
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8396282859154727e-05,
+      "loss": 0.3741,
+      "step": 6736
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8395741449441685e-05,
+      "loss": 0.3826,
+      "step": 6737
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8395199956324745e-05,
+      "loss": 0.3109,
+      "step": 6738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839465837980928e-05,
+      "loss": 0.4515,
+      "step": 6739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8394116719900665e-05,
+      "loss": 0.3973,
+      "step": 6740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8393574976604287e-05,
+      "loss": 0.3712,
+      "step": 6741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8393033149925528e-05,
+      "loss": 0.3567,
+      "step": 6742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8392491239869765e-05,
+      "loss": 0.3638,
+      "step": 6743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839194924644239e-05,
+      "loss": 0.3884,
+      "step": 6744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8391407169648777e-05,
+      "loss": 0.4459,
+      "step": 6745
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8390865009494322e-05,
+      "loss": 0.3667,
+      "step": 6746
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8390322765984402e-05,
+      "loss": 0.4188,
+      "step": 6747
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8389780439124408e-05,
+      "loss": 0.3402,
+      "step": 6748
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8389238028919725e-05,
+      "loss": 0.4183,
+      "step": 6749
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8388695535375745e-05,
+      "loss": 0.4575,
+      "step": 6750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838815295849785e-05,
+      "loss": 0.3722,
+      "step": 6751
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8387610298291442e-05,
+      "loss": 0.3811,
+      "step": 6752
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.83870675547619e-05,
+      "loss": 0.4937,
+      "step": 6753
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8386524727914622e-05,
+      "loss": 0.3951,
+      "step": 6754
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8385981817754997e-05,
+      "loss": 0.5103,
+      "step": 6755
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8385438824288424e-05,
+      "loss": 0.3888,
+      "step": 6756
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838489574752029e-05,
+      "loss": 0.3683,
+      "step": 6757
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8384352587455996e-05,
+      "loss": 0.4271,
+      "step": 6758
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383809344100934e-05,
+      "loss": 0.3783,
+      "step": 6759
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.83832660174605e-05,
+      "loss": 0.4154,
+      "step": 6760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8382722607540097e-05,
+      "loss": 0.4021,
+      "step": 6761
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8382179114345115e-05,
+      "loss": 0.4093,
+      "step": 6762
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838163553788096e-05,
+      "loss": 0.4531,
+      "step": 6763
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838109187815303e-05,
+      "loss": 0.3721,
+      "step": 6764
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8380548135166725e-05,
+      "loss": 0.3417,
+      "step": 6765
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8380004308927445e-05,
+      "loss": 0.3913,
+      "step": 6766
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8379460399440595e-05,
+      "loss": 0.3366,
+      "step": 6767
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837891640671158e-05,
+      "loss": 0.3781,
+      "step": 6768
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378372330745797e-05,
+      "loss": 0.3241,
+      "step": 6769
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837782817154866e-05,
+      "loss": 0.4209,
+      "step": 6770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8377283929125566e-05,
+      "loss": 0.4725,
+      "step": 6771
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8376739603481928e-05,
+      "loss": 0.3428,
+      "step": 6772
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837619519462315e-05,
+      "loss": 0.4054,
+      "step": 6773
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8375650702554643e-05,
+      "loss": 0.416,
+      "step": 6774
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837510612728181e-05,
+      "loss": 0.3974,
+      "step": 6775
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837456146881007e-05,
+      "loss": 0.4474,
+      "step": 6776
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8374016727144827e-05,
+      "loss": 0.378,
+      "step": 6777
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8373471902291497e-05,
+      "loss": 0.3908,
+      "step": 6778
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8372926994255486e-05,
+      "loss": 0.3635,
+      "step": 6779
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837238200304221e-05,
+      "loss": 0.3459,
+      "step": 6780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8371836928657085e-05,
+      "loss": 0.3832,
+      "step": 6781
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8371291771105527e-05,
+      "loss": 0.4057,
+      "step": 6782
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8370746530392946e-05,
+      "loss": 0.43,
+      "step": 6783
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8370201206524764e-05,
+      "loss": 0.5039,
+      "step": 6784
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369655799506394e-05,
+      "loss": 0.3931,
+      "step": 6785
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369110309343257e-05,
+      "loss": 0.3985,
+      "step": 6786
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.836856473604077e-05,
+      "loss": 0.3473,
+      "step": 6787
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.836801907960436e-05,
+      "loss": 0.3685,
+      "step": 6788
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.836747334003943e-05,
+      "loss": 0.3811,
+      "step": 6789
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.836692751735142e-05,
+      "loss": 0.3907,
+      "step": 6790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8366381611545745e-05,
+      "loss": 0.4461,
+      "step": 6791
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8365835622627826e-05,
+      "loss": 0.4492,
+      "step": 6792
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8365289550603093e-05,
+      "loss": 0.4662,
+      "step": 6793
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8364743395476964e-05,
+      "loss": 0.4431,
+      "step": 6794
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8364197157254863e-05,
+      "loss": 0.3909,
+      "step": 6795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8363650835942225e-05,
+      "loss": 0.4198,
+      "step": 6796
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8363104431544474e-05,
+      "loss": 0.463,
+      "step": 6797
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8362557944067038e-05,
+      "loss": 0.471,
+      "step": 6798
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8362011373515338e-05,
+      "loss": 0.3398,
+      "step": 6799
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8361464719894818e-05,
+      "loss": 0.4694,
+      "step": 6800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8360917983210894e-05,
+      "loss": 0.4728,
+      "step": 6801
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836037116346901e-05,
+      "loss": 0.4174,
+      "step": 6802
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835982426067459e-05,
+      "loss": 0.3783,
+      "step": 6803
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8359277274833066e-05,
+      "loss": 0.391,
+      "step": 6804
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835873020594988e-05,
+      "loss": 0.3242,
+      "step": 6805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835818305403046e-05,
+      "loss": 0.37,
+      "step": 6806
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835763581908024e-05,
+      "loss": 0.4036,
+      "step": 6807
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8357088501104662e-05,
+      "loss": 0.4064,
+      "step": 6808
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835654110010916e-05,
+      "loss": 0.3969,
+      "step": 6809
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8355993616099175e-05,
+      "loss": 0.3915,
+      "step": 6810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835544604908014e-05,
+      "loss": 0.3928,
+      "step": 6811
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.83548983990575e-05,
+      "loss": 0.3654,
+      "step": 6812
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8354350666036692e-05,
+      "loss": 0.4368,
+      "step": 6813
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835380285002316e-05,
+      "loss": 0.4435,
+      "step": 6814
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8353254951022342e-05,
+      "loss": 0.3827,
+      "step": 6815
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8352706969039686e-05,
+      "loss": 0.4431,
+      "step": 6816
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8352158904080632e-05,
+      "loss": 0.4183,
+      "step": 6817
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835161075615062e-05,
+      "loss": 0.434,
+      "step": 6818
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835106252525511e-05,
+      "loss": 0.3817,
+      "step": 6819
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8350514211399533e-05,
+      "loss": 0.3656,
+      "step": 6820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8349965814589342e-05,
+      "loss": 0.3917,
+      "step": 6821
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834941733482999e-05,
+      "loss": 0.3936,
+      "step": 6822
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348868772126916e-05,
+      "loss": 0.4573,
+      "step": 6823
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348320126485573e-05,
+      "loss": 0.4531,
+      "step": 6824
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8347771397911414e-05,
+      "loss": 0.38,
+      "step": 6825
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834722258640989e-05,
+      "loss": 0.4423,
+      "step": 6826
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834667369198645e-05,
+      "loss": 0.3887,
+      "step": 6827
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834612471464655e-05,
+      "loss": 0.4811,
+      "step": 6828
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8345575654395637e-05,
+      "loss": 0.4415,
+      "step": 6829
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8345026511239172e-05,
+      "loss": 0.3517,
+      "step": 6830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8344477285182608e-05,
+      "loss": 0.4177,
+      "step": 6831
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.83439279762314e-05,
+      "loss": 0.4484,
+      "step": 6832
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834337858439101e-05,
+      "loss": 0.4468,
+      "step": 6833
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834282910966689e-05,
+      "loss": 0.3852,
+      "step": 6834
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.83422795520645e-05,
+      "loss": 0.3505,
+      "step": 6835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.83417299115893e-05,
+      "loss": 0.4518,
+      "step": 6836
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8341180188246745e-05,
+      "loss": 0.4171,
+      "step": 6837
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834063038204231e-05,
+      "loss": 0.3711,
+      "step": 6838
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8340080492981442e-05,
+      "loss": 0.3575,
+      "step": 6839
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833953052106961e-05,
+      "loss": 0.4596,
+      "step": 6840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338980466312275e-05,
+      "loss": 0.3837,
+      "step": 6841
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338430328714905e-05,
+      "loss": 0.4352,
+      "step": 6842
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337880108282965e-05,
+      "loss": 0.3612,
+      "step": 6843
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337329805021916e-05,
+      "loss": 0.4177,
+      "step": 6844
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833677941893723e-05,
+      "loss": 0.411,
+      "step": 6845
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8336228950034372e-05,
+      "loss": 0.4565,
+      "step": 6846
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833567839831881e-05,
+      "loss": 0.4322,
+      "step": 6847
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8335127763796016e-05,
+      "loss": 0.501,
+      "step": 6848
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8334577046471452e-05,
+      "loss": 0.3857,
+      "step": 6849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8334026246350604e-05,
+      "loss": 0.3725,
+      "step": 6850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833347536343893e-05,
+      "loss": 0.4185,
+      "step": 6851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8332924397741906e-05,
+      "loss": 0.3766,
+      "step": 6852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8332373349265007e-05,
+      "loss": 0.4626,
+      "step": 6853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8331822218013707e-05,
+      "loss": 0.4272,
+      "step": 6854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833127100399348e-05,
+      "loss": 0.3941,
+      "step": 6855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8330719707209806e-05,
+      "loss": 0.3995,
+      "step": 6856
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8330168327668155e-05,
+      "loss": 0.3373,
+      "step": 6857
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329616865374008e-05,
+      "loss": 0.3756,
+      "step": 6858
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329065320332842e-05,
+      "loss": 0.4151,
+      "step": 6859
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8328513692550138e-05,
+      "loss": 0.4233,
+      "step": 6860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8327961982031376e-05,
+      "loss": 0.4401,
+      "step": 6861
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8327410188782032e-05,
+      "loss": 0.4627,
+      "step": 6862
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832685831280759e-05,
+      "loss": 0.4449,
+      "step": 6863
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832630635411354e-05,
+      "loss": 0.4055,
+      "step": 6864
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8325754312705354e-05,
+      "loss": 0.4319,
+      "step": 6865
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832520218858852e-05,
+      "loss": 0.3899,
+      "step": 6866
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8324649981768525e-05,
+      "loss": 0.3713,
+      "step": 6867
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8324097692250854e-05,
+      "loss": 0.4086,
+      "step": 6868
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8323545320040992e-05,
+      "loss": 0.3906,
+      "step": 6869
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8322992865144428e-05,
+      "loss": 0.3662,
+      "step": 6870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8322440327566643e-05,
+      "loss": 0.3601,
+      "step": 6871
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832188770731314e-05,
+      "loss": 0.4595,
+      "step": 6872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321335004389395e-05,
+      "loss": 0.3506,
+      "step": 6873
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832078221880091e-05,
+      "loss": 0.4514,
+      "step": 6874
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8320229350553165e-05,
+      "loss": 0.3664,
+      "step": 6875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831967639965166e-05,
+      "loss": 0.3447,
+      "step": 6876
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831912336610189e-05,
+      "loss": 0.354,
+      "step": 6877
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831857024990934e-05,
+      "loss": 0.4239,
+      "step": 6878
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831801705107951e-05,
+      "loss": 0.4372,
+      "step": 6879
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317463769617898e-05,
+      "loss": 0.4024,
+      "step": 6880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316910405529995e-05,
+      "loss": 0.4005,
+      "step": 6881
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.83163569588213e-05,
+      "loss": 0.4136,
+      "step": 6882
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8315803429497312e-05,
+      "loss": 0.3705,
+      "step": 6883
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831524981756353e-05,
+      "loss": 0.3917,
+      "step": 6884
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.4334,
+      "step": 6885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8314142345888584e-05,
+      "loss": 0.3876,
+      "step": 6886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831358848615842e-05,
+      "loss": 0.3856,
+      "step": 6887
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8313034543840464e-05,
+      "loss": 0.3938,
+      "step": 6888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831248051894022e-05,
+      "loss": 0.3344,
+      "step": 6889
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831192641146319e-05,
+      "loss": 0.3515,
+      "step": 6890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8311372221414882e-05,
+      "loss": 0.3965,
+      "step": 6891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.83108179488008e-05,
+      "loss": 0.4654,
+      "step": 6892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8310263593626448e-05,
+      "loss": 0.4226,
+      "step": 6893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8309709155897332e-05,
+      "loss": 0.3887,
+      "step": 6894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8309154635618967e-05,
+      "loss": 0.3849,
+      "step": 6895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8308600032796854e-05,
+      "loss": 0.4137,
+      "step": 6896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830804534743651e-05,
+      "loss": 0.4037,
+      "step": 6897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8307490579543435e-05,
+      "loss": 0.4041,
+      "step": 6898
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830693572912315e-05,
+      "loss": 0.4728,
+      "step": 6899
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830638079618116e-05,
+      "loss": 0.3606,
+      "step": 6900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830582578072298e-05,
+      "loss": 0.431,
+      "step": 6901
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830527068275413e-05,
+      "loss": 0.3884,
+      "step": 6902
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8304715502280115e-05,
+      "loss": 0.3942,
+      "step": 6903
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8304160239306453e-05,
+      "loss": 0.3919,
+      "step": 6904
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830360489383866e-05,
+      "loss": 0.4667,
+      "step": 6905
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8303049465882254e-05,
+      "loss": 0.3789,
+      "step": 6906
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8302493955442753e-05,
+      "loss": 0.3474,
+      "step": 6907
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8301938362525676e-05,
+      "loss": 0.4012,
+      "step": 6908
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830138268713654e-05,
+      "loss": 0.3483,
+      "step": 6909
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8300826929280863e-05,
+      "loss": 0.3981,
+      "step": 6910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830027108896417e-05,
+      "loss": 0.426,
+      "step": 6911
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8299715166191986e-05,
+      "loss": 0.3864,
+      "step": 6912
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8299159160969827e-05,
+      "loss": 0.4484,
+      "step": 6913
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8298603073303217e-05,
+      "loss": 0.4785,
+      "step": 6914
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8298046903197685e-05,
+      "loss": 0.4723,
+      "step": 6915
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829749065065875e-05,
+      "loss": 0.3695,
+      "step": 6916
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8296934315691945e-05,
+      "loss": 0.4072,
+      "step": 6917
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8296377898302788e-05,
+      "loss": 0.3873,
+      "step": 6918
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8295821398496814e-05,
+      "loss": 0.3828,
+      "step": 6919
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8295264816279545e-05,
+      "loss": 0.3914,
+      "step": 6920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8294708151656517e-05,
+      "loss": 0.4181,
+      "step": 6921
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8294151404633254e-05,
+      "loss": 0.4789,
+      "step": 6922
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829359457521529e-05,
+      "loss": 0.4341,
+      "step": 6923
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8293037663408158e-05,
+      "loss": 0.4056,
+      "step": 6924
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8292480669217387e-05,
+      "loss": 0.4288,
+      "step": 6925
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8291923592648512e-05,
+      "loss": 0.3854,
+      "step": 6926
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829136643370707e-05,
+      "loss": 0.474,
+      "step": 6927
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8290809192398583e-05,
+      "loss": 0.4254,
+      "step": 6928
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82902518687286e-05,
+      "loss": 0.4313,
+      "step": 6929
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8289694462702656e-05,
+      "loss": 0.3967,
+      "step": 6930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828913697432628e-05,
+      "loss": 0.3543,
+      "step": 6931
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288579403605022e-05,
+      "loss": 0.4082,
+      "step": 6932
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288021750544415e-05,
+      "loss": 0.4176,
+      "step": 6933
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8287464015149996e-05,
+      "loss": 0.3861,
+      "step": 6934
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828690619742731e-05,
+      "loss": 0.4086,
+      "step": 6935
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8286348297381893e-05,
+      "loss": 0.3953,
+      "step": 6936
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828579031501929e-05,
+      "loss": 0.3878,
+      "step": 6937
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828523225034505e-05,
+      "loss": 0.4261,
+      "step": 6938
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828467410336471e-05,
+      "loss": 0.3714,
+      "step": 6939
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8284115874083816e-05,
+      "loss": 0.4101,
+      "step": 6940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8283557562507917e-05,
+      "loss": 0.3463,
+      "step": 6941
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8282999168642554e-05,
+      "loss": 0.3353,
+      "step": 6942
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8282440692493272e-05,
+      "loss": 0.3497,
+      "step": 6943
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8281882134065626e-05,
+      "loss": 0.4207,
+      "step": 6944
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8281323493365165e-05,
+      "loss": 0.3769,
+      "step": 6945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8280764770397432e-05,
+      "loss": 0.4078,
+      "step": 6946
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8280205965167984e-05,
+      "loss": 0.3846,
+      "step": 6947
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8279647077682366e-05,
+      "loss": 0.4211,
+      "step": 6948
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8279088107946136e-05,
+      "loss": 0.3778,
+      "step": 6949
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8278529055964843e-05,
+      "loss": 0.4922,
+      "step": 6950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277969921744044e-05,
+      "loss": 0.4019,
+      "step": 6951
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277410705289288e-05,
+      "loss": 0.3606,
+      "step": 6952
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8276851406606132e-05,
+      "loss": 0.4232,
+      "step": 6953
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8276292025700136e-05,
+      "loss": 0.365,
+      "step": 6954
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8275732562576857e-05,
+      "loss": 0.362,
+      "step": 6955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8275173017241847e-05,
+      "loss": 0.4356,
+      "step": 6956
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8274613389700667e-05,
+      "loss": 0.3906,
+      "step": 6957
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827405367995888e-05,
+      "loss": 0.4202,
+      "step": 6958
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8273493888022042e-05,
+      "loss": 0.435,
+      "step": 6959
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8272934013895717e-05,
+      "loss": 0.3501,
+      "step": 6960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8272374057585464e-05,
+      "loss": 0.5121,
+      "step": 6961
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8271814019096847e-05,
+      "loss": 0.3519,
+      "step": 6962
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827125389843543e-05,
+      "loss": 0.4855,
+      "step": 6963
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8270693695606775e-05,
+      "loss": 0.3636,
+      "step": 6964
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8270133410616453e-05,
+      "loss": 0.3906,
+      "step": 6965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826957304347002e-05,
+      "loss": 0.4262,
+      "step": 6966
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8269012594173054e-05,
+      "loss": 0.416,
+      "step": 6967
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8268452062731116e-05,
+      "loss": 0.4013,
+      "step": 6968
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826789144914977e-05,
+      "loss": 0.4799,
+      "step": 6969
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8267330753434596e-05,
+      "loss": 0.382,
+      "step": 6970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8266769975591157e-05,
+      "loss": 0.3857,
+      "step": 6971
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8266209115625025e-05,
+      "loss": 0.4807,
+      "step": 6972
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8265648173541772e-05,
+      "loss": 0.3994,
+      "step": 6973
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826508714934697e-05,
+      "loss": 0.419,
+      "step": 6974
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8264526043046193e-05,
+      "loss": 0.3818,
+      "step": 6975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263964854645017e-05,
+      "loss": 0.4149,
+      "step": 6976
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263403584149008e-05,
+      "loss": 0.3809,
+      "step": 6977
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8262842231563753e-05,
+      "loss": 0.4384,
+      "step": 6978
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8262280796894826e-05,
+      "loss": 0.451,
+      "step": 6979
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82617192801478e-05,
+      "loss": 0.4175,
+      "step": 6980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826115768132825e-05,
+      "loss": 0.3516,
+      "step": 6981
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8260596000441767e-05,
+      "loss": 0.4354,
+      "step": 6982
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826003423749392e-05,
+      "loss": 0.3055,
+      "step": 6983
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8259472392490297e-05,
+      "loss": 0.4378,
+      "step": 6984
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8258910465436476e-05,
+      "loss": 0.336,
+      "step": 6985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8258348456338037e-05,
+      "loss": 0.4118,
+      "step": 6986
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8257786365200564e-05,
+      "loss": 0.4135,
+      "step": 6987
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8257224192029648e-05,
+      "loss": 0.3951,
+      "step": 6988
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825666193683086e-05,
+      "loss": 0.4113,
+      "step": 6989
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82560995996098e-05,
+      "loss": 0.4184,
+      "step": 6990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8255537180372044e-05,
+      "loss": 0.3646,
+      "step": 6991
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8254974679123183e-05,
+      "loss": 0.3963,
+      "step": 6992
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8254412095868804e-05,
+      "loss": 0.4162,
+      "step": 6993
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82538494306145e-05,
+      "loss": 0.4367,
+      "step": 6994
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8253286683365855e-05,
+      "loss": 0.4174,
+      "step": 6995
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8252723854128463e-05,
+      "loss": 0.4262,
+      "step": 6996
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825216094290791e-05,
+      "loss": 0.353,
+      "step": 6997
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825159794970979e-05,
+      "loss": 0.4786,
+      "step": 6998
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82510348745397e-05,
+      "loss": 0.4409,
+      "step": 6999
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8250471717403237e-05,
+      "loss": 0.3516,
+      "step": 7000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824990847830598e-05,
+      "loss": 0.3522,
+      "step": 7001
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8249345157253538e-05,
+      "loss": 0.3368,
+      "step": 7002
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82487817542515e-05,
+      "loss": 0.4305,
+      "step": 7003
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8248218269305467e-05,
+      "loss": 0.4425,
+      "step": 7004
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8247654702421037e-05,
+      "loss": 0.4318,
+      "step": 7005
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8247091053603808e-05,
+      "loss": 0.4363,
+      "step": 7006
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246527322859374e-05,
+      "loss": 0.4939,
+      "step": 7007
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824596351019334e-05,
+      "loss": 0.3824,
+      "step": 7008
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824539961561131e-05,
+      "loss": 0.3635,
+      "step": 7009
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824483563911888e-05,
+      "loss": 0.3769,
+      "step": 7010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8244271580721653e-05,
+      "loss": 0.4183,
+      "step": 7011
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8243707440425232e-05,
+      "loss": 0.3702,
+      "step": 7012
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824314321823523e-05,
+      "loss": 0.3718,
+      "step": 7013
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8242578914157242e-05,
+      "loss": 0.3606,
+      "step": 7014
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8242014528196875e-05,
+      "loss": 0.3827,
+      "step": 7015
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824145006035974e-05,
+      "loss": 0.4578,
+      "step": 7016
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824088551065144e-05,
+      "loss": 0.4044,
+      "step": 7017
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8240320879077586e-05,
+      "loss": 0.3636,
+      "step": 7018
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823975616564379e-05,
+      "loss": 0.3563,
+      "step": 7019
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8239191370355654e-05,
+      "loss": 0.4252,
+      "step": 7020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8238626493218798e-05,
+      "loss": 0.4691,
+      "step": 7021
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8238061534238822e-05,
+      "loss": 0.3527,
+      "step": 7022
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823749649342135e-05,
+      "loss": 0.4008,
+      "step": 7023
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823693137077199e-05,
+      "loss": 0.3649,
+      "step": 7024
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8236366166296356e-05,
+      "loss": 0.3909,
+      "step": 7025
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8235800880000058e-05,
+      "loss": 0.5072,
+      "step": 7026
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8235235511888722e-05,
+      "loss": 0.3815,
+      "step": 7027
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8234670061967956e-05,
+      "loss": 0.3469,
+      "step": 7028
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8234104530243384e-05,
+      "loss": 0.4063,
+      "step": 7029
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8233538916720616e-05,
+      "loss": 0.4384,
+      "step": 7030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823297322140528e-05,
+      "loss": 0.4444,
+      "step": 7031
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8232407444302985e-05,
+      "loss": 0.3818,
+      "step": 7032
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823184158541936e-05,
+      "loss": 0.4037,
+      "step": 7033
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8231275644760024e-05,
+      "loss": 0.3538,
+      "step": 7034
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8230709622330596e-05,
+      "loss": 0.3874,
+      "step": 7035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8230143518136706e-05,
+      "loss": 0.3893,
+      "step": 7036
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822957733218397e-05,
+      "loss": 0.3759,
+      "step": 7037
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8229011064478016e-05,
+      "loss": 0.3905,
+      "step": 7038
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8228444715024472e-05,
+      "loss": 0.3587,
+      "step": 7039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822787828382896e-05,
+      "loss": 0.2948,
+      "step": 7040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8227311770897104e-05,
+      "loss": 0.3326,
+      "step": 7041
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8226745176234538e-05,
+      "loss": 0.3848,
+      "step": 7042
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822617849984689e-05,
+      "loss": 0.4653,
+      "step": 7043
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822561174173979e-05,
+      "loss": 0.4116,
+      "step": 7044
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8225044901918862e-05,
+      "loss": 0.4762,
+      "step": 7045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8224477980389742e-05,
+      "loss": 0.3718,
+      "step": 7046
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8223910977158063e-05,
+      "loss": 0.3708,
+      "step": 7047
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8223343892229453e-05,
+      "loss": 0.3769,
+      "step": 7048
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822277672560955e-05,
+      "loss": 0.454,
+      "step": 7049
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822220947730399e-05,
+      "loss": 0.4306,
+      "step": 7050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8221642147318396e-05,
+      "loss": 0.358,
+      "step": 7051
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8221074735658418e-05,
+      "loss": 0.3508,
+      "step": 7052
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8220507242329687e-05,
+      "loss": 0.4323,
+      "step": 7053
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821993966733784e-05,
+      "loss": 0.3533,
+      "step": 7054
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8219372010688516e-05,
+      "loss": 0.458,
+      "step": 7055
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218804272387354e-05,
+      "loss": 0.4002,
+      "step": 7056
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218236452439992e-05,
+      "loss": 0.4154,
+      "step": 7057
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8217668550852074e-05,
+      "loss": 0.3928,
+      "step": 7058
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8217100567629242e-05,
+      "loss": 0.4219,
+      "step": 7059
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8216532502777136e-05,
+      "loss": 0.3968,
+      "step": 7060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8215964356301402e-05,
+      "loss": 0.3975,
+      "step": 7061
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821539612820768e-05,
+      "loss": 0.3736,
+      "step": 7062
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8214827818501616e-05,
+      "loss": 0.3684,
+      "step": 7063
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821425942718886e-05,
+      "loss": 0.374,
+      "step": 7064
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821369095427505e-05,
+      "loss": 0.3795,
+      "step": 7065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8213122399765843e-05,
+      "loss": 0.4233,
+      "step": 7066
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821255376366688e-05,
+      "loss": 0.4277,
+      "step": 7067
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821198504598381e-05,
+      "loss": 0.4121,
+      "step": 7068
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8211416246722287e-05,
+      "loss": 0.4007,
+      "step": 7069
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8210847365887957e-05,
+      "loss": 0.4119,
+      "step": 7070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8210278403486476e-05,
+      "loss": 0.3968,
+      "step": 7071
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8209709359523494e-05,
+      "loss": 0.4258,
+      "step": 7072
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8209140234004664e-05,
+      "loss": 0.4043,
+      "step": 7073
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820857102693564e-05,
+      "loss": 0.4012,
+      "step": 7074
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820800173832207e-05,
+      "loss": 0.3331,
+      "step": 7075
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820743236816962e-05,
+      "loss": 0.434,
+      "step": 7076
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820686291648394e-05,
+      "loss": 0.38,
+      "step": 7077
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820629338327069e-05,
+      "loss": 0.4185,
+      "step": 7078
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205723768535525e-05,
+      "loss": 0.4661,
+      "step": 7079
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205154072284105e-05,
+      "loss": 0.3721,
+      "step": 7080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820458429452209e-05,
+      "loss": 0.379,
+      "step": 7081
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8204014435255136e-05,
+      "loss": 0.4322,
+      "step": 7082
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820344449448891e-05,
+      "loss": 0.3529,
+      "step": 7083
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8202874472229075e-05,
+      "loss": 0.4151,
+      "step": 7084
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8202304368481285e-05,
+      "loss": 0.3809,
+      "step": 7085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820173418325121e-05,
+      "loss": 0.3724,
+      "step": 7086
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820116391654451e-05,
+      "loss": 0.3653,
+      "step": 7087
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8200593568366854e-05,
+      "loss": 0.382,
+      "step": 7088
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8200023138723907e-05,
+      "loss": 0.4272,
+      "step": 7089
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8199452627621337e-05,
+      "loss": 0.3713,
+      "step": 7090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8198882035064806e-05,
+      "loss": 0.3857,
+      "step": 7091
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819831136105999e-05,
+      "loss": 0.3771,
+      "step": 7092
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819774060561255e-05,
+      "loss": 0.4677,
+      "step": 7093
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819716976872816e-05,
+      "loss": 0.3665,
+      "step": 7094
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8196598850412496e-05,
+      "loss": 0.3862,
+      "step": 7095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8196027850671222e-05,
+      "loss": 0.4377,
+      "step": 7096
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8195456769510014e-05,
+      "loss": 0.4283,
+      "step": 7097
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819488560693454e-05,
+      "loss": 0.3792,
+      "step": 7098
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8194314362950482e-05,
+      "loss": 0.436,
+      "step": 7099
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193743037563508e-05,
+      "loss": 0.3557,
+      "step": 7100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193171630779294e-05,
+      "loss": 0.4266,
+      "step": 7101
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8192600142603522e-05,
+      "loss": 0.3786,
+      "step": 7102
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8192028573041866e-05,
+      "loss": 0.4515,
+      "step": 7103
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8191456922100002e-05,
+      "loss": 0.3788,
+      "step": 7104
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819088518978361e-05,
+      "loss": 0.4463,
+      "step": 7105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8190313376098373e-05,
+      "loss": 0.4122,
+      "step": 7106
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818974148104997e-05,
+      "loss": 0.4384,
+      "step": 7107
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8189169504644077e-05,
+      "loss": 0.3726,
+      "step": 7108
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8188597446886383e-05,
+      "loss": 0.3816,
+      "step": 7109
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8188025307782565e-05,
+      "loss": 0.4508,
+      "step": 7110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818745308733831e-05,
+      "loss": 0.3986,
+      "step": 7111
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8186880785559303e-05,
+      "loss": 0.4963,
+      "step": 7112
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818630840245123e-05,
+      "loss": 0.348,
+      "step": 7113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8185735938019773e-05,
+      "loss": 0.354,
+      "step": 7114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8185163392270623e-05,
+      "loss": 0.4989,
+      "step": 7115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8184590765209464e-05,
+      "loss": 0.3639,
+      "step": 7116
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818401805684199e-05,
+      "loss": 0.361,
+      "step": 7117
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8183445267173883e-05,
+      "loss": 0.384,
+      "step": 7118
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8182872396210837e-05,
+      "loss": 0.4215,
+      "step": 7119
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8182299443958548e-05,
+      "loss": 0.3142,
+      "step": 7120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8181726410422696e-05,
+      "loss": 0.3787,
+      "step": 7121
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8181153295608986e-05,
+      "loss": 0.3326,
+      "step": 7122
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81805800995231e-05,
+      "loss": 0.4699,
+      "step": 7123
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8180006822170743e-05,
+      "loss": 0.4188,
+      "step": 7124
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8179433463557603e-05,
+      "loss": 0.4236,
+      "step": 7125
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8178860023689374e-05,
+      "loss": 0.4707,
+      "step": 7126
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817828650257176e-05,
+      "loss": 0.4395,
+      "step": 7127
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8177712900210455e-05,
+      "loss": 0.3776,
+      "step": 7128
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8177139216611155e-05,
+      "loss": 0.3833,
+      "step": 7129
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8176565451779563e-05,
+      "loss": 0.4352,
+      "step": 7130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8175991605721374e-05,
+      "loss": 0.4328,
+      "step": 7131
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817541767844229e-05,
+      "loss": 0.446,
+      "step": 7132
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8174843669948016e-05,
+      "loss": 0.3873,
+      "step": 7133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8174269580244252e-05,
+      "loss": 0.3854,
+      "step": 7134
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8173695409336703e-05,
+      "loss": 0.4054,
+      "step": 7135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8173121157231065e-05,
+      "loss": 0.4896,
+      "step": 7136
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8172546823933053e-05,
+      "loss": 0.3972,
+      "step": 7137
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817197240944837e-05,
+      "loss": 0.4422,
+      "step": 7138
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8171397913782717e-05,
+      "loss": 0.3423,
+      "step": 7139
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81708233369418e-05,
+      "loss": 0.4039,
+      "step": 7140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8170248678931336e-05,
+      "loss": 0.3499,
+      "step": 7141
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816967393975703e-05,
+      "loss": 0.3765,
+      "step": 7142
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816909911942459e-05,
+      "loss": 0.4221,
+      "step": 7143
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8168524217939724e-05,
+      "loss": 0.3969,
+      "step": 7144
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8167949235308148e-05,
+      "loss": 0.4409,
+      "step": 7145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816737417153557e-05,
+      "loss": 0.3609,
+      "step": 7146
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8166799026627705e-05,
+      "loss": 0.4009,
+      "step": 7147
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8166223800590268e-05,
+      "loss": 0.4825,
+      "step": 7148
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816564849342897e-05,
+      "loss": 0.4173,
+      "step": 7149
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8165073105149528e-05,
+      "loss": 0.4566,
+      "step": 7150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8164497635757655e-05,
+      "loss": 0.3753,
+      "step": 7151
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8163922085259068e-05,
+      "loss": 0.3959,
+      "step": 7152
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816334645365949e-05,
+      "loss": 0.3726,
+      "step": 7153
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816277074096464e-05,
+      "loss": 0.5,
+      "step": 7154
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8162194947180228e-05,
+      "loss": 0.4134,
+      "step": 7155
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8161619072311983e-05,
+      "loss": 0.3366,
+      "step": 7156
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816104311636562e-05,
+      "loss": 0.5022,
+      "step": 7157
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816046707934686e-05,
+      "loss": 0.4533,
+      "step": 7158
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8159890961261433e-05,
+      "loss": 0.3725,
+      "step": 7159
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8159314762115054e-05,
+      "loss": 0.3926,
+      "step": 7160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8158738481913453e-05,
+      "loss": 0.4224,
+      "step": 7161
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8158162120662348e-05,
+      "loss": 0.405,
+      "step": 7162
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8157585678367467e-05,
+      "loss": 0.342,
+      "step": 7163
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8157009155034545e-05,
+      "loss": 0.3454,
+      "step": 7164
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8156432550669298e-05,
+      "loss": 0.3888,
+      "step": 7165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815585586527746e-05,
+      "loss": 0.3954,
+      "step": 7166
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8155279098864755e-05,
+      "loss": 0.3833,
+      "step": 7167
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815470225143692e-05,
+      "loss": 0.3658,
+      "step": 7168
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8154125322999678e-05,
+      "loss": 0.4211,
+      "step": 7169
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8153548313558766e-05,
+      "loss": 0.3664,
+      "step": 7170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815297122311991e-05,
+      "loss": 0.3569,
+      "step": 7171
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815239405168885e-05,
+      "loss": 0.3973,
+      "step": 7172
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8151816799271316e-05,
+      "loss": 0.3506,
+      "step": 7173
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815123946587304e-05,
+      "loss": 0.4338,
+      "step": 7174
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815066205149976e-05,
+      "loss": 0.3935,
+      "step": 7175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8150084556157216e-05,
+      "loss": 0.4282,
+      "step": 7176
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8149506979851135e-05,
+      "loss": 0.4416,
+      "step": 7177
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8148929322587264e-05,
+      "loss": 0.3797,
+      "step": 7178
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8148351584371337e-05,
+      "loss": 0.3879,
+      "step": 7179
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8147773765209092e-05,
+      "loss": 0.4338,
+      "step": 7180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8147195865106273e-05,
+      "loss": 0.4437,
+      "step": 7181
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814661788406862e-05,
+      "loss": 0.367,
+      "step": 7182
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8146039822101873e-05,
+      "loss": 0.3639,
+      "step": 7183
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8145461679211774e-05,
+      "loss": 0.4382,
+      "step": 7184
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8144883455404072e-05,
+      "loss": 0.383,
+      "step": 7185
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8144305150684504e-05,
+      "loss": 0.3852,
+      "step": 7186
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8143726765058818e-05,
+      "loss": 0.4182,
+      "step": 7187
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814314829853276e-05,
+      "loss": 0.4172,
+      "step": 7188
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8142569751112073e-05,
+      "loss": 0.4378,
+      "step": 7189
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814199112280251e-05,
+      "loss": 0.3051,
+      "step": 7190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814141241360982e-05,
+      "loss": 0.2992,
+      "step": 7191
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814083362353974e-05,
+      "loss": 0.4164,
+      "step": 7192
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8140254752598037e-05,
+      "loss": 0.3364,
+      "step": 7193
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8139675800790447e-05,
+      "loss": 0.4005,
+      "step": 7194
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8139096768122727e-05,
+      "loss": 0.4075,
+      "step": 7195
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8138517654600634e-05,
+      "loss": 0.3948,
+      "step": 7196
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8137938460229912e-05,
+      "loss": 0.4099,
+      "step": 7197
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813735918501632e-05,
+      "loss": 0.3734,
+      "step": 7198
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8136779828965614e-05,
+      "loss": 0.385,
+      "step": 7199
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8136200392083543e-05,
+      "loss": 0.3446,
+      "step": 7200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8135620874375866e-05,
+      "loss": 0.4442,
+      "step": 7201
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8135041275848345e-05,
+      "loss": 0.3599,
+      "step": 7202
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813446159650673e-05,
+      "loss": 0.4422,
+      "step": 7203
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813388183635679e-05,
+      "loss": 0.4486,
+      "step": 7204
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813330199540427e-05,
+      "loss": 0.4009,
+      "step": 7205
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132722073654943e-05,
+      "loss": 0.3955,
+      "step": 7206
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132142071114562e-05,
+      "loss": 0.4164,
+      "step": 7207
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8131561987788892e-05,
+      "loss": 0.3654,
+      "step": 7208
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8130981823683696e-05,
+      "loss": 0.3972,
+      "step": 7209
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8130401578804736e-05,
+      "loss": 0.3934,
+      "step": 7210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812982125315778e-05,
+      "loss": 0.3494,
+      "step": 7211
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8129240846748587e-05,
+      "loss": 0.3748,
+      "step": 7212
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8128660359582927e-05,
+      "loss": 0.3667,
+      "step": 7213
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8128079791666563e-05,
+      "loss": 0.3652,
+      "step": 7214
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8127499143005266e-05,
+      "loss": 0.3569,
+      "step": 7215
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8126918413604803e-05,
+      "loss": 0.4433,
+      "step": 7216
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812633760347094e-05,
+      "loss": 0.3635,
+      "step": 7217
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8125756712609454e-05,
+      "loss": 0.3954,
+      "step": 7218
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8125175741026112e-05,
+      "loss": 0.4814,
+      "step": 7219
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812459468872668e-05,
+      "loss": 0.3613,
+      "step": 7220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8124013555716938e-05,
+      "loss": 0.4204,
+      "step": 7221
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8123432342002655e-05,
+      "loss": 0.4321,
+      "step": 7222
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8122851047589606e-05,
+      "loss": 0.4559,
+      "step": 7223
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8122269672483566e-05,
+      "loss": 0.4009,
+      "step": 7224
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812168821669031e-05,
+      "loss": 0.4844,
+      "step": 7225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8121106680215614e-05,
+      "loss": 0.3859,
+      "step": 7226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8120525063065254e-05,
+      "loss": 0.4003,
+      "step": 7227
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8119943365245008e-05,
+      "loss": 0.3715,
+      "step": 7228
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8119361586760656e-05,
+      "loss": 0.3713,
+      "step": 7229
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8118779727617977e-05,
+      "loss": 0.4849,
+      "step": 7230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811819778782275e-05,
+      "loss": 0.3574,
+      "step": 7231
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811761576738076e-05,
+      "loss": 0.374,
+      "step": 7232
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8117033666297787e-05,
+      "loss": 0.4171,
+      "step": 7233
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811645148457961e-05,
+      "loss": 0.4521,
+      "step": 7234
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8115869222232015e-05,
+      "loss": 0.3436,
+      "step": 7235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8115286879260787e-05,
+      "loss": 0.4664,
+      "step": 7236
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8114704455671713e-05,
+      "loss": 0.4253,
+      "step": 7237
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8114121951470572e-05,
+      "loss": 0.4626,
+      "step": 7238
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8113539366663157e-05,
+      "loss": 0.3774,
+      "step": 7239
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112956701255254e-05,
+      "loss": 0.3667,
+      "step": 7240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112373955252647e-05,
+      "loss": 0.3856,
+      "step": 7241
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811179112866113e-05,
+      "loss": 0.4211,
+      "step": 7242
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8111208221486494e-05,
+      "loss": 0.3957,
+      "step": 7243
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8110625233734524e-05,
+      "loss": 0.3652,
+      "step": 7244
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8110042165411016e-05,
+      "loss": 0.4212,
+      "step": 7245
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8109459016521758e-05,
+      "loss": 0.3663,
+      "step": 7246
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810887578707255e-05,
+      "loss": 0.424,
+      "step": 7247
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8108292477069182e-05,
+      "loss": 0.4086,
+      "step": 7248
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8107709086517444e-05,
+      "loss": 0.4446,
+      "step": 7249
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810712561542314e-05,
+      "loss": 0.3653,
+      "step": 7250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810654206379206e-05,
+      "loss": 0.411,
+      "step": 7251
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8105958431630003e-05,
+      "loss": 0.3632,
+      "step": 7252
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810537471894277e-05,
+      "loss": 0.457,
+      "step": 7253
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810479092573615e-05,
+      "loss": 0.3394,
+      "step": 7254
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8104207052015952e-05,
+      "loss": 0.4829,
+      "step": 7255
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8103623097787975e-05,
+      "loss": 0.4175,
+      "step": 7256
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8103039063058014e-05,
+      "loss": 0.4177,
+      "step": 7257
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810245494783188e-05,
+      "loss": 0.412,
+      "step": 7258
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810187075211537e-05,
+      "loss": 0.3781,
+      "step": 7259
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8101286475914285e-05,
+      "loss": 0.3362,
+      "step": 7260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8100702119234432e-05,
+      "loss": 0.3652,
+      "step": 7261
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810011768208162e-05,
+      "loss": 0.4168,
+      "step": 7262
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809953316446165e-05,
+      "loss": 0.3546,
+      "step": 7263
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809894856638033e-05,
+      "loss": 0.452,
+      "step": 7264
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8098363887843467e-05,
+      "loss": 0.4111,
+      "step": 7265
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809777912885687e-05,
+      "loss": 0.4464,
+      "step": 7266
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8097194289426347e-05,
+      "loss": 0.4264,
+      "step": 7267
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809660936955771e-05,
+      "loss": 0.4074,
+      "step": 7268
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8096024369256766e-05,
+      "loss": 0.3702,
+      "step": 7269
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809543928852933e-05,
+      "loss": 0.4019,
+      "step": 7270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8094854127381212e-05,
+      "loss": 0.4755,
+      "step": 7271
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8094268885818223e-05,
+      "loss": 0.4246,
+      "step": 7272
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8093683563846187e-05,
+      "loss": 0.4122,
+      "step": 7273
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8093098161470903e-05,
+      "loss": 0.403,
+      "step": 7274
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.80925126786982e-05,
+      "loss": 0.3584,
+      "step": 7275
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8091927115533887e-05,
+      "loss": 0.4012,
+      "step": 7276
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8091341471983783e-05,
+      "loss": 0.3865,
+      "step": 7277
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809075574805371e-05,
+      "loss": 0.3904,
+      "step": 7278
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.3896,
+      "step": 7279
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808958405907691e-05,
+      "loss": 0.3527,
+      "step": 7280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8088998094041827e-05,
+      "loss": 0.3691,
+      "step": 7281
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8088412048650053e-05,
+      "loss": 0.3801,
+      "step": 7282
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8087825922907406e-05,
+      "loss": 0.4309,
+      "step": 7283
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808723971681971e-05,
+      "loss": 0.3289,
+      "step": 7284
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8086653430392784e-05,
+      "loss": 0.3327,
+      "step": 7285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808606706363246e-05,
+      "loss": 0.3496,
+      "step": 7286
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808548061654456e-05,
+      "loss": 0.4753,
+      "step": 7287
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8084894089134907e-05,
+      "loss": 0.3887,
+      "step": 7288
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808430748140933e-05,
+      "loss": 0.4493,
+      "step": 7289
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8083720793373656e-05,
+      "loss": 0.3822,
+      "step": 7290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8083134025033713e-05,
+      "loss": 0.4132,
+      "step": 7291
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808254717639533e-05,
+      "loss": 0.3996,
+      "step": 7292
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8081960247464336e-05,
+      "loss": 0.4515,
+      "step": 7293
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8081373238246565e-05,
+      "loss": 0.5183,
+      "step": 7294
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080786148747842e-05,
+      "loss": 0.39,
+      "step": 7295
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080198978974007e-05,
+      "loss": 0.3826,
+      "step": 7296
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8079611728930888e-05,
+      "loss": 0.4221,
+      "step": 7297
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807902439862432e-05,
+      "loss": 0.4073,
+      "step": 7298
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8078436988060136e-05,
+      "loss": 0.382,
+      "step": 7299
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8077849497244175e-05,
+      "loss": 0.3891,
+      "step": 7300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807726192618227e-05,
+      "loss": 0.3502,
+      "step": 7301
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807667427488026e-05,
+      "loss": 0.4255,
+      "step": 7302
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807608654334398e-05,
+      "loss": 0.4389,
+      "step": 7303
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8075498731579275e-05,
+      "loss": 0.3793,
+      "step": 7304
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8074910839591972e-05,
+      "loss": 0.4453,
+      "step": 7305
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8074322867387922e-05,
+      "loss": 0.4852,
+      "step": 7306
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8073734814972966e-05,
+      "loss": 0.3789,
+      "step": 7307
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807314668235294e-05,
+      "loss": 0.4625,
+      "step": 7308
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807255846953369e-05,
+      "loss": 0.3948,
+      "step": 7309
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8071970176521057e-05,
+      "loss": 0.3141,
+      "step": 7310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8071381803320886e-05,
+      "loss": 0.5332,
+      "step": 7311
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8070793349939025e-05,
+      "loss": 0.378,
+      "step": 7312
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8070204816381318e-05,
+      "loss": 0.3693,
+      "step": 7313
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806961620265361e-05,
+      "loss": 0.4776,
+      "step": 7314
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8069027508761747e-05,
+      "loss": 0.3991,
+      "step": 7315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8068438734711583e-05,
+      "loss": 0.4388,
+      "step": 7316
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8067849880508957e-05,
+      "loss": 0.3473,
+      "step": 7317
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8067260946159732e-05,
+      "loss": 0.3723,
+      "step": 7318
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806667193166975e-05,
+      "loss": 0.4341,
+      "step": 7319
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806608283704486e-05,
+      "loss": 0.3883,
+      "step": 7320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806549366229092e-05,
+      "loss": 0.3605,
+      "step": 7321
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8064904407413782e-05,
+      "loss": 0.3669,
+      "step": 7322
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8064315072419298e-05,
+      "loss": 0.3682,
+      "step": 7323
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806372565731332e-05,
+      "loss": 0.3995,
+      "step": 7324
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8063136162101715e-05,
+      "loss": 0.4244,
+      "step": 7325
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806254658679032e-05,
+      "loss": 0.4235,
+      "step": 7326
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806195693138501e-05,
+      "loss": 0.4052,
+      "step": 7327
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8061367195891632e-05,
+      "loss": 0.4834,
+      "step": 7328
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8060777380316047e-05,
+      "loss": 0.4422,
+      "step": 7329
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8060187484664112e-05,
+      "loss": 0.3769,
+      "step": 7330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8059597508941692e-05,
+      "loss": 0.3896,
+      "step": 7331
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8059007453154645e-05,
+      "loss": 0.3965,
+      "step": 7332
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8058417317308835e-05,
+      "loss": 0.4101,
+      "step": 7333
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805782710141012e-05,
+      "loss": 0.4473,
+      "step": 7334
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805723680546437e-05,
+      "loss": 0.5058,
+      "step": 7335
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056646429477443e-05,
+      "loss": 0.367,
+      "step": 7336
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056055973455205e-05,
+      "loss": 0.4125,
+      "step": 7337
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8055465437403526e-05,
+      "loss": 0.3937,
+      "step": 7338
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8054874821328267e-05,
+      "loss": 0.4162,
+      "step": 7339
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8054284125235296e-05,
+      "loss": 0.4149,
+      "step": 7340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805369334913048e-05,
+      "loss": 0.4706,
+      "step": 7341
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8053102493019695e-05,
+      "loss": 0.4976,
+      "step": 7342
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80525115569088e-05,
+      "loss": 0.4295,
+      "step": 7343
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8051920540803676e-05,
+      "loss": 0.4456,
+      "step": 7344
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8051329444710187e-05,
+      "loss": 0.4158,
+      "step": 7345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8050738268634206e-05,
+      "loss": 0.3693,
+      "step": 7346
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8050147012581608e-05,
+      "loss": 0.3473,
+      "step": 7347
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804955567655827e-05,
+      "loss": 0.4176,
+      "step": 7348
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8048964260570056e-05,
+      "loss": 0.3713,
+      "step": 7349
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8048372764622847e-05,
+      "loss": 0.4638,
+      "step": 7350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8047781188722524e-05,
+      "loss": 0.4886,
+      "step": 7351
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8047189532874955e-05,
+      "loss": 0.4177,
+      "step": 7352
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8046597797086023e-05,
+      "loss": 0.4376,
+      "step": 7353
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8046005981361605e-05,
+      "loss": 0.4191,
+      "step": 7354
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8045414085707578e-05,
+      "loss": 0.4546,
+      "step": 7355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8044822110129825e-05,
+      "loss": 0.3727,
+      "step": 7356
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8044230054634223e-05,
+      "loss": 0.3225,
+      "step": 7357
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804363791922666e-05,
+      "loss": 0.4263,
+      "step": 7358
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8043045703913012e-05,
+      "loss": 0.3977,
+      "step": 7359
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8042453408699164e-05,
+      "loss": 0.3765,
+      "step": 7360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8041861033591003e-05,
+      "loss": 0.4336,
+      "step": 7361
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8041268578594405e-05,
+      "loss": 0.3545,
+      "step": 7362
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8040676043715268e-05,
+      "loss": 0.4275,
+      "step": 7363
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804008342895947e-05,
+      "loss": 0.4126,
+      "step": 7364
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80394907343329e-05,
+      "loss": 0.3865,
+      "step": 7365
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8038897959841445e-05,
+      "loss": 0.4069,
+      "step": 7366
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8038305105490994e-05,
+      "loss": 0.4158,
+      "step": 7367
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8037712171287438e-05,
+      "loss": 0.3993,
+      "step": 7368
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8037119157236667e-05,
+      "loss": 0.4021,
+      "step": 7369
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803652606334457e-05,
+      "loss": 0.3694,
+      "step": 7370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803593288961704e-05,
+      "loss": 0.4352,
+      "step": 7371
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8035339636059968e-05,
+      "loss": 0.3487,
+      "step": 7372
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8034746302679253e-05,
+      "loss": 0.3467,
+      "step": 7373
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8034152889480783e-05,
+      "loss": 0.4052,
+      "step": 7374
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8033559396470455e-05,
+      "loss": 0.3764,
+      "step": 7375
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8032965823654166e-05,
+      "loss": 0.4119,
+      "step": 7376
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803237217103781e-05,
+      "loss": 0.4613,
+      "step": 7377
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8031778438627286e-05,
+      "loss": 0.3585,
+      "step": 7378
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8031184626428496e-05,
+      "loss": 0.3857,
+      "step": 7379
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8030590734447333e-05,
+      "loss": 0.4337,
+      "step": 7380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8029996762689697e-05,
+      "loss": 0.4572,
+      "step": 7381
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8029402711161493e-05,
+      "loss": 0.3658,
+      "step": 7382
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802880857986862e-05,
+      "loss": 0.3989,
+      "step": 7383
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802821436881698e-05,
+      "loss": 0.319,
+      "step": 7384
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8027620078012474e-05,
+      "loss": 0.4121,
+      "step": 7385
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802702570746101e-05,
+      "loss": 0.3496,
+      "step": 7386
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802643125716849e-05,
+      "loss": 0.4705,
+      "step": 7387
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802583672714082e-05,
+      "loss": 0.4133,
+      "step": 7388
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8025242117383902e-05,
+      "loss": 0.478,
+      "step": 7389
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802464742790365e-05,
+      "loss": 0.3759,
+      "step": 7390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024052658705967e-05,
+      "loss": 0.4681,
+      "step": 7391
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8023457809796766e-05,
+      "loss": 0.3882,
+      "step": 7392
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8022862881181947e-05,
+      "loss": 0.3742,
+      "step": 7393
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802226787286743e-05,
+      "loss": 0.4266,
+      "step": 7394
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802167278485912e-05,
+      "loss": 0.369,
+      "step": 7395
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8021077617162937e-05,
+      "loss": 0.4299,
+      "step": 7396
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8020482369784778e-05,
+      "loss": 0.3895,
+      "step": 7397
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801988704273057e-05,
+      "loss": 0.4096,
+      "step": 7398
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8019291636006222e-05,
+      "loss": 0.3958,
+      "step": 7399
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801869614961765e-05,
+      "loss": 0.4575,
+      "step": 7400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8018100583570765e-05,
+      "loss": 0.4041,
+      "step": 7401
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801750493787149e-05,
+      "loss": 0.4013,
+      "step": 7402
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8016909212525738e-05,
+      "loss": 0.4529,
+      "step": 7403
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801631340753943e-05,
+      "loss": 0.3802,
+      "step": 7404
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801571752291848e-05,
+      "loss": 0.3584,
+      "step": 7405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8015121558668814e-05,
+      "loss": 0.3784,
+      "step": 7406
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801452551479635e-05,
+      "loss": 0.3822,
+      "step": 7407
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8013929391307002e-05,
+      "loss": 0.4086,
+      "step": 7408
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8013333188206702e-05,
+      "loss": 0.4024,
+      "step": 7409
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8012736905501368e-05,
+      "loss": 0.3834,
+      "step": 7410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8012140543196926e-05,
+      "loss": 0.4,
+      "step": 7411
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8011544101299295e-05,
+      "loss": 0.4506,
+      "step": 7412
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8010947579814407e-05,
+      "loss": 0.4028,
+      "step": 7413
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8010350978748183e-05,
+      "loss": 0.4433,
+      "step": 7414
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8009754298106553e-05,
+      "loss": 0.4318,
+      "step": 7415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800915753789544e-05,
+      "loss": 0.409,
+      "step": 7416
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8008560698120777e-05,
+      "loss": 0.4959,
+      "step": 7417
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8007963778788492e-05,
+      "loss": 0.3707,
+      "step": 7418
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8007366779904512e-05,
+      "loss": 0.4066,
+      "step": 7419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8006769701474767e-05,
+      "loss": 0.4116,
+      "step": 7420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8006172543505193e-05,
+      "loss": 0.4281,
+      "step": 7421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8005575306001722e-05,
+      "loss": 0.3891,
+      "step": 7422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8004977988970284e-05,
+      "loss": 0.4205,
+      "step": 7423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8004380592416814e-05,
+      "loss": 0.3997,
+      "step": 7424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8003783116347245e-05,
+      "loss": 0.4246,
+      "step": 7425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8003185560767517e-05,
+      "loss": 0.3921,
+      "step": 7426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800258792568356e-05,
+      "loss": 0.4353,
+      "step": 7427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8001990211101316e-05,
+      "loss": 0.4089,
+      "step": 7428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800139241702672e-05,
+      "loss": 0.4025,
+      "step": 7429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8000794543465706e-05,
+      "loss": 0.3521,
+      "step": 7430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8000196590424224e-05,
+      "loss": 0.43,
+      "step": 7431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799959855790821e-05,
+      "loss": 0.3892,
+      "step": 7432
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.79990004459236e-05,
+      "loss": 0.3682,
+      "step": 7433
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799840225447634e-05,
+      "loss": 0.4257,
+      "step": 7434
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7997803983572372e-05,
+      "loss": 0.3308,
+      "step": 7435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799720563321764e-05,
+      "loss": 0.3634,
+      "step": 7436
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799660720341809e-05,
+      "loss": 0.4886,
+      "step": 7437
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7996008694179655e-05,
+      "loss": 0.4548,
+      "step": 7438
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.79954101055083e-05,
+      "loss": 0.4834,
+      "step": 7439
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7994811437409955e-05,
+      "loss": 0.3482,
+      "step": 7440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7994212689890572e-05,
+      "loss": 0.4473,
+      "step": 7441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79936138629561e-05,
+      "loss": 0.449,
+      "step": 7442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7993014956612492e-05,
+      "loss": 0.3644,
+      "step": 7443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799241597086569e-05,
+      "loss": 0.3526,
+      "step": 7444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7991816905721654e-05,
+      "loss": 0.4085,
+      "step": 7445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7991217761186322e-05,
+      "loss": 0.4216,
+      "step": 7446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7990618537265656e-05,
+      "loss": 0.4166,
+      "step": 7447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7990019233965605e-05,
+      "loss": 0.4111,
+      "step": 7448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7989419851292123e-05,
+      "loss": 0.3975,
+      "step": 7449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988820389251166e-05,
+      "loss": 0.3871,
+      "step": 7450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988220847848684e-05,
+      "loss": 0.421,
+      "step": 7451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7987621227090642e-05,
+      "loss": 0.4124,
+      "step": 7452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7987021526982988e-05,
+      "loss": 0.3591,
+      "step": 7453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7986421747531684e-05,
+      "loss": 0.3305,
+      "step": 7454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7985821888742687e-05,
+      "loss": 0.4957,
+      "step": 7455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7985221950621956e-05,
+      "loss": 0.429,
+      "step": 7456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7984621933175447e-05,
+      "loss": 0.4385,
+      "step": 7457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7984021836409128e-05,
+      "loss": 0.4979,
+      "step": 7458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7983421660328958e-05,
+      "loss": 0.4151,
+      "step": 7459
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7982821404940893e-05,
+      "loss": 0.4017,
+      "step": 7460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7982221070250906e-05,
+      "loss": 0.3921,
+      "step": 7461
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7981620656264955e-05,
+      "loss": 0.4304,
+      "step": 7462
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7981020162989e-05,
+      "loss": 0.4504,
+      "step": 7463
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798041959042902e-05,
+      "loss": 0.4754,
+      "step": 7464
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7979818938590967e-05,
+      "loss": 0.4046,
+      "step": 7465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7979218207480817e-05,
+      "loss": 0.4158,
+      "step": 7466
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797861739710453e-05,
+      "loss": 0.3629,
+      "step": 7467
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7978016507468084e-05,
+      "loss": 0.4029,
+      "step": 7468
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7977415538577442e-05,
+      "loss": 0.3557,
+      "step": 7469
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7976814490438574e-05,
+      "loss": 0.3735,
+      "step": 7470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7976213363057452e-05,
+      "loss": 0.4679,
+      "step": 7471
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797561215644005e-05,
+      "loss": 0.4298,
+      "step": 7472
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7975010870592336e-05,
+      "loss": 0.3564,
+      "step": 7473
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7974409505520288e-05,
+      "loss": 0.331,
+      "step": 7474
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7973808061229873e-05,
+      "loss": 0.4555,
+      "step": 7475
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797320653772707e-05,
+      "loss": 0.3459,
+      "step": 7476
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797260493501786e-05,
+      "loss": 0.3927,
+      "step": 7477
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7972003253108212e-05,
+      "loss": 0.4215,
+      "step": 7478
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7971401492004107e-05,
+      "loss": 0.4128,
+      "step": 7479
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797079965171152e-05,
+      "loss": 0.3969,
+      "step": 7480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7970197732236427e-05,
+      "loss": 0.3969,
+      "step": 7481
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7969595733584815e-05,
+      "loss": 0.4767,
+      "step": 7482
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796899365576266e-05,
+      "loss": 0.4217,
+      "step": 7483
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7968391498775945e-05,
+      "loss": 0.391,
+      "step": 7484
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7967789262630647e-05,
+      "loss": 0.4086,
+      "step": 7485
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7967186947332756e-05,
+      "loss": 0.391,
+      "step": 7486
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7966584552888252e-05,
+      "loss": 0.4038,
+      "step": 7487
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796598207930312e-05,
+      "loss": 0.4349,
+      "step": 7488
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7965379526583343e-05,
+      "loss": 0.4328,
+      "step": 7489
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7964776894734904e-05,
+      "loss": 0.3499,
+      "step": 7490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79641741837638e-05,
+      "loss": 0.3831,
+      "step": 7491
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7963571393676007e-05,
+      "loss": 0.3975,
+      "step": 7492
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796296852447752e-05,
+      "loss": 0.4424,
+      "step": 7493
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796236557617433e-05,
+      "loss": 0.41,
+      "step": 7494
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7961762548772413e-05,
+      "loss": 0.4478,
+      "step": 7495
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7961159442277777e-05,
+      "loss": 0.3943,
+      "step": 7496
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7960556256696403e-05,
+      "loss": 0.4247,
+      "step": 7497
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7959952992034284e-05,
+      "loss": 0.4017,
+      "step": 7498
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7959349648297415e-05,
+      "loss": 0.403,
+      "step": 7499
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7958746225491792e-05,
+      "loss": 0.4521,
+      "step": 7500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7958142723623403e-05,
+      "loss": 0.3528,
+      "step": 7501
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795753914269825e-05,
+      "loss": 0.4498,
+      "step": 7502
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7956935482722325e-05,
+      "loss": 0.3348,
+      "step": 7503
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7956331743701624e-05,
+      "loss": 0.4059,
+      "step": 7504
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7955727925642147e-05,
+      "loss": 0.3232,
+      "step": 7505
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7955124028549895e-05,
+      "loss": 0.4455,
+      "step": 7506
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795452005243086e-05,
+      "loss": 0.3032,
+      "step": 7507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7953915997291048e-05,
+      "loss": 0.4344,
+      "step": 7508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7953311863136453e-05,
+      "loss": 0.4182,
+      "step": 7509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7952707649973084e-05,
+      "loss": 0.4278,
+      "step": 7510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795210335780694e-05,
+      "loss": 0.4302,
+      "step": 7511
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7951498986644025e-05,
+      "loss": 0.4969,
+      "step": 7512
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7950894536490338e-05,
+      "loss": 0.4151,
+      "step": 7513
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7950290007351893e-05,
+      "loss": 0.4391,
+      "step": 7514
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794968539923469e-05,
+      "loss": 0.3736,
+      "step": 7515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7949080712144735e-05,
+      "loss": 0.4363,
+      "step": 7516
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7948475946088034e-05,
+      "loss": 0.4352,
+      "step": 7517
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79478711010706e-05,
+      "loss": 0.3817,
+      "step": 7518
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7947266177098432e-05,
+      "loss": 0.4209,
+      "step": 7519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794666117417755e-05,
+      "loss": 0.4164,
+      "step": 7520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7946056092313954e-05,
+      "loss": 0.3684,
+      "step": 7521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7945450931513665e-05,
+      "loss": 0.4763,
+      "step": 7522
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794484569178269e-05,
+      "loss": 0.467,
+      "step": 7523
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794424037312704e-05,
+      "loss": 0.3487,
+      "step": 7524
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794363497555273e-05,
+      "loss": 0.438,
+      "step": 7525
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794302949906577e-05,
+      "loss": 0.3846,
+      "step": 7526
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7942423943672186e-05,
+      "loss": 0.4291,
+      "step": 7527
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7941818309377982e-05,
+      "loss": 0.3949,
+      "step": 7528
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794121259618918e-05,
+      "loss": 0.3817,
+      "step": 7529
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7940606804111794e-05,
+      "loss": 0.33,
+      "step": 7530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7940000933151845e-05,
+      "loss": 0.3758,
+      "step": 7531
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793939498331535e-05,
+      "loss": 0.4203,
+      "step": 7532
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7938788954608333e-05,
+      "loss": 0.4935,
+      "step": 7533
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7938182847036808e-05,
+      "loss": 0.4411,
+      "step": 7534
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79375766606068e-05,
+      "loss": 0.372,
+      "step": 7535
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7936970395324326e-05,
+      "loss": 0.3621,
+      "step": 7536
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7936364051195415e-05,
+      "loss": 0.4056,
+      "step": 7537
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7935757628226086e-05,
+      "loss": 0.4529,
+      "step": 7538
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7935151126422368e-05,
+      "loss": 0.5135,
+      "step": 7539
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793454454579028e-05,
+      "loss": 0.5188,
+      "step": 7540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793393788633585e-05,
+      "loss": 0.3694,
+      "step": 7541
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793333114806511e-05,
+      "loss": 0.3143,
+      "step": 7542
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7932724330984078e-05,
+      "loss": 0.4241,
+      "step": 7543
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793211743509879e-05,
+      "loss": 0.4108,
+      "step": 7544
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7931510460415272e-05,
+      "loss": 0.3913,
+      "step": 7545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7930903406939554e-05,
+      "loss": 0.3583,
+      "step": 7546
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7930296274677663e-05,
+      "loss": 0.3809,
+      "step": 7547
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7929689063635636e-05,
+      "loss": 0.3634,
+      "step": 7548
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7929081773819504e-05,
+      "loss": 0.4602,
+      "step": 7549
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7928474405235295e-05,
+      "loss": 0.3896,
+      "step": 7550
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7927866957889047e-05,
+      "loss": 0.382,
+      "step": 7551
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7927259431786793e-05,
+      "loss": 0.3667,
+      "step": 7552
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7926651826934568e-05,
+      "loss": 0.3902,
+      "step": 7553
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7926044143338413e-05,
+      "loss": 0.3804,
+      "step": 7554
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7925436381004356e-05,
+      "loss": 0.3838,
+      "step": 7555
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792482853993844e-05,
+      "loss": 0.3666,
+      "step": 7556
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7924220620146707e-05,
+      "loss": 0.3175,
+      "step": 7557
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7923612621635185e-05,
+      "loss": 0.3453,
+      "step": 7558
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792300454440992e-05,
+      "loss": 0.4143,
+      "step": 7559
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7922396388476956e-05,
+      "loss": 0.4387,
+      "step": 7560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7921788153842334e-05,
+      "loss": 0.4349,
+      "step": 7561
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792117984051209e-05,
+      "loss": 0.3904,
+      "step": 7562
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7920571448492273e-05,
+      "loss": 0.3375,
+      "step": 7563
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919962977788922e-05,
+      "loss": 0.4086,
+      "step": 7564
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919354428408086e-05,
+      "loss": 0.4189,
+      "step": 7565
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.791874580035581e-05,
+      "loss": 0.4183,
+      "step": 7566
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7918137093638135e-05,
+      "loss": 0.3892,
+      "step": 7567
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7917528308261113e-05,
+      "loss": 0.3716,
+      "step": 7568
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7916919444230793e-05,
+      "loss": 0.401,
+      "step": 7569
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7916310501553216e-05,
+      "loss": 0.4066,
+      "step": 7570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7915701480234438e-05,
+      "loss": 0.4105,
+      "step": 7571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7915092380280512e-05,
+      "loss": 0.3666,
+      "step": 7572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791448320169748e-05,
+      "loss": 0.3989,
+      "step": 7573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7913873944491397e-05,
+      "loss": 0.4117,
+      "step": 7574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7913264608668316e-05,
+      "loss": 0.3775,
+      "step": 7575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7912655194234293e-05,
+      "loss": 0.3639,
+      "step": 7576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791204570119538e-05,
+      "loss": 0.3759,
+      "step": 7577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911436129557627e-05,
+      "loss": 0.434,
+      "step": 7578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7910826479327094e-05,
+      "loss": 0.4703,
+      "step": 7579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791021675050984e-05,
+      "loss": 0.4793,
+      "step": 7580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7909606943111913e-05,
+      "loss": 0.3614,
+      "step": 7581
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790899705713938e-05,
+      "loss": 0.3504,
+      "step": 7582
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.79083870925983e-05,
+      "loss": 0.4141,
+      "step": 7583
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907777049494726e-05,
+      "loss": 0.3481,
+      "step": 7584
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907166927834717e-05,
+      "loss": 0.3965,
+      "step": 7585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7906556727624343e-05,
+      "loss": 0.405,
+      "step": 7586
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790594644886966e-05,
+      "loss": 0.3866,
+      "step": 7587
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790533609157673e-05,
+      "loss": 0.5371,
+      "step": 7588
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7904725655751623e-05,
+      "loss": 0.4578,
+      "step": 7589
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790411514140039e-05,
+      "loss": 0.3665,
+      "step": 7590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790350454852911e-05,
+      "loss": 0.4038,
+      "step": 7591
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790289387714384e-05,
+      "loss": 0.4309,
+      "step": 7592
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7902283127250647e-05,
+      "loss": 0.3512,
+      "step": 7593
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7901672298855604e-05,
+      "loss": 0.4064,
+      "step": 7594
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790106139196477e-05,
+      "loss": 0.5148,
+      "step": 7595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7900450406584223e-05,
+      "loss": 0.3583,
+      "step": 7596
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899839342720028e-05,
+      "loss": 0.4282,
+      "step": 7597
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899228200378257e-05,
+      "loss": 0.3226,
+      "step": 7598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7898616979564976e-05,
+      "loss": 0.4182,
+      "step": 7599
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7898005680286262e-05,
+      "loss": 0.3499,
+      "step": 7600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789739430254819e-05,
+      "loss": 0.3956,
+      "step": 7601
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7896782846356827e-05,
+      "loss": 0.4074,
+      "step": 7602
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7896171311718252e-05,
+      "loss": 0.4088,
+      "step": 7603
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895559698638534e-05,
+      "loss": 0.3943,
+      "step": 7604
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789494800712376e-05,
+      "loss": 0.4554,
+      "step": 7605
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7894336237179994e-05,
+      "loss": 0.3307,
+      "step": 7606
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789372438881332e-05,
+      "loss": 0.3984,
+      "step": 7607
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7893112462029818e-05,
+      "loss": 0.4353,
+      "step": 7608
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7892500456835563e-05,
+      "loss": 0.3711,
+      "step": 7609
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7891888373236636e-05,
+      "loss": 0.3976,
+      "step": 7610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7891276211239118e-05,
+      "loss": 0.4747,
+      "step": 7611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7890663970849086e-05,
+      "loss": 0.4094,
+      "step": 7612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789005165207263e-05,
+      "loss": 0.4614,
+      "step": 7613
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7889439254915824e-05,
+      "loss": 0.4701,
+      "step": 7614
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788882677938476e-05,
+      "loss": 0.4492,
+      "step": 7615
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7888214225485518e-05,
+      "loss": 0.461,
+      "step": 7616
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788760159322418e-05,
+      "loss": 0.4062,
+      "step": 7617
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788698888260684e-05,
+      "loss": 0.3309,
+      "step": 7618
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7886376093639577e-05,
+      "loss": 0.4354,
+      "step": 7619
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7885763226328484e-05,
+      "loss": 0.4152,
+      "step": 7620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7885150280679647e-05,
+      "loss": 0.345,
+      "step": 7621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7884537256699155e-05,
+      "loss": 0.4405,
+      "step": 7622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78839241543931e-05,
+      "loss": 0.3327,
+      "step": 7623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7883310973767565e-05,
+      "loss": 0.4255,
+      "step": 7624
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7882697714828647e-05,
+      "loss": 0.3967,
+      "step": 7625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7882084377582442e-05,
+      "loss": 0.3822,
+      "step": 7626
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7881470962035037e-05,
+      "loss": 0.4095,
+      "step": 7627
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7880857468192528e-05,
+      "loss": 0.3965,
+      "step": 7628
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788024389606101e-05,
+      "loss": 0.381,
+      "step": 7629
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7879630245646574e-05,
+      "loss": 0.4622,
+      "step": 7630
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7879016516955324e-05,
+      "loss": 0.5164,
+      "step": 7631
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787840270999335e-05,
+      "loss": 0.3901,
+      "step": 7632
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7877788824766752e-05,
+      "loss": 0.4021,
+      "step": 7633
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7877174861281628e-05,
+      "loss": 0.4324,
+      "step": 7634
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7876560819544077e-05,
+      "loss": 0.4283,
+      "step": 7635
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78759466995602e-05,
+      "loss": 0.476,
+      "step": 7636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7875332501336097e-05,
+      "loss": 0.446,
+      "step": 7637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7874718224877872e-05,
+      "loss": 0.3426,
+      "step": 7638
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787410387019162e-05,
+      "loss": 0.3346,
+      "step": 7639
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7873489437283457e-05,
+      "loss": 0.47,
+      "step": 7640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787287492615947e-05,
+      "loss": 0.4182,
+      "step": 7641
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7872260336825778e-05,
+      "loss": 0.4147,
+      "step": 7642
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7871645669288477e-05,
+      "loss": 0.4098,
+      "step": 7643
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7871030923553678e-05,
+      "loss": 0.4578,
+      "step": 7644
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7870416099627488e-05,
+      "loss": 0.4155,
+      "step": 7645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786980119751601e-05,
+      "loss": 0.3898,
+      "step": 7646
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7869186217225357e-05,
+      "loss": 0.4554,
+      "step": 7647
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7868571158761643e-05,
+      "loss": 0.4064,
+      "step": 7648
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7867956022130967e-05,
+      "loss": 0.4461,
+      "step": 7649
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7867340807339445e-05,
+      "loss": 0.4368,
+      "step": 7650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786672551439319e-05,
+      "loss": 0.503,
+      "step": 7651
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7866110143298312e-05,
+      "loss": 0.4213,
+      "step": 7652
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786549469406093e-05,
+      "loss": 0.4431,
+      "step": 7653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7864879166687144e-05,
+      "loss": 0.5517,
+      "step": 7654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7864263561183085e-05,
+      "loss": 0.3917,
+      "step": 7655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7863647877554856e-05,
+      "loss": 0.3835,
+      "step": 7656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7863032115808583e-05,
+      "loss": 0.433,
+      "step": 7657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7862416275950375e-05,
+      "loss": 0.4384,
+      "step": 7658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7861800357986355e-05,
+      "loss": 0.351,
+      "step": 7659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786118436192264e-05,
+      "loss": 0.3712,
+      "step": 7660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7860568287765348e-05,
+      "loss": 0.4128,
+      "step": 7661
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7859952135520603e-05,
+      "loss": 0.3398,
+      "step": 7662
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785933590519452e-05,
+      "loss": 0.4052,
+      "step": 7663
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785871959679323e-05,
+      "loss": 0.3746,
+      "step": 7664
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785810321032284e-05,
+      "loss": 0.3683,
+      "step": 7665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7857486745789493e-05,
+      "loss": 0.4705,
+      "step": 7666
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7856870203199294e-05,
+      "loss": 0.4662,
+      "step": 7667
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785625358255838e-05,
+      "loss": 0.4597,
+      "step": 7668
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7855636883872874e-05,
+      "loss": 0.4859,
+      "step": 7669
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7855020107148898e-05,
+      "loss": 0.4507,
+      "step": 7670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7854403252392587e-05,
+      "loss": 0.3972,
+      "step": 7671
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785378631961006e-05,
+      "loss": 0.4166,
+      "step": 7672
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785316930880745e-05,
+      "loss": 0.4349,
+      "step": 7673
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785255221999089e-05,
+      "loss": 0.3697,
+      "step": 7674
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7851935053166503e-05,
+      "loss": 0.3506,
+      "step": 7675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7851317808340424e-05,
+      "loss": 0.3679,
+      "step": 7676
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7850700485518786e-05,
+      "loss": 0.4359,
+      "step": 7677
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785008308470772e-05,
+      "loss": 0.4329,
+      "step": 7678
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7849465605913357e-05,
+      "loss": 0.3932,
+      "step": 7679
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7848848049141835e-05,
+      "loss": 0.3653,
+      "step": 7680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7848230414399284e-05,
+      "loss": 0.3731,
+      "step": 7681
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7847612701691846e-05,
+      "loss": 0.3836,
+      "step": 7682
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7846994911025653e-05,
+      "loss": 0.3586,
+      "step": 7683
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7846377042406843e-05,
+      "loss": 0.438,
+      "step": 7684
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7845759095841553e-05,
+      "loss": 0.411,
+      "step": 7685
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7845141071335924e-05,
+      "loss": 0.4465,
+      "step": 7686
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7844522968896094e-05,
+      "loss": 0.4267,
+      "step": 7687
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843904788528207e-05,
+      "loss": 0.3919,
+      "step": 7688
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843286530238397e-05,
+      "loss": 0.4608,
+      "step": 7689
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7842668194032812e-05,
+      "loss": 0.3288,
+      "step": 7690
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784204977991759e-05,
+      "loss": 0.4046,
+      "step": 7691
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784143128789888e-05,
+      "loss": 0.4485,
+      "step": 7692
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784081271798282e-05,
+      "loss": 0.4058,
+      "step": 7693
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7840194070175558e-05,
+      "loss": 0.4508,
+      "step": 7694
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7839575344483237e-05,
+      "loss": 0.5271,
+      "step": 7695
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7838956540912007e-05,
+      "loss": 0.3987,
+      "step": 7696
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7838337659468017e-05,
+      "loss": 0.5023,
+      "step": 7697
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783771870015741e-05,
+      "loss": 0.4936,
+      "step": 7698
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7837099662986335e-05,
+      "loss": 0.3532,
+      "step": 7699
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7836480547960945e-05,
+      "loss": 0.396,
+      "step": 7700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783586135508739e-05,
+      "loss": 0.3868,
+      "step": 7701
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7835242084371817e-05,
+      "loss": 0.4122,
+      "step": 7702
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7834622735820385e-05,
+      "loss": 0.4148,
+      "step": 7703
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783400330943924e-05,
+      "loss": 0.3833,
+      "step": 7704
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7833383805234536e-05,
+      "loss": 0.3784,
+      "step": 7705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7832764223212427e-05,
+      "loss": 0.4918,
+      "step": 7706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7832144563379078e-05,
+      "loss": 0.3443,
+      "step": 7707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783152482574063e-05,
+      "loss": 0.4623,
+      "step": 7708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7830905010303247e-05,
+      "loss": 0.386,
+      "step": 7709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7830285117073086e-05,
+      "loss": 0.3629,
+      "step": 7710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7829665146056308e-05,
+      "loss": 0.3559,
+      "step": 7711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7829045097259062e-05,
+      "loss": 0.409,
+      "step": 7712
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782842497068752e-05,
+      "loss": 0.3919,
+      "step": 7713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7827804766347834e-05,
+      "loss": 0.3586,
+      "step": 7714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782718448424617e-05,
+      "loss": 0.4026,
+      "step": 7715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782656412438868e-05,
+      "loss": 0.3713,
+      "step": 7716
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7825943686781543e-05,
+      "loss": 0.3717,
+      "step": 7717
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782532317143091e-05,
+      "loss": 0.4855,
+      "step": 7718
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7824702578342944e-05,
+      "loss": 0.4065,
+      "step": 7719
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782408190752382e-05,
+      "loss": 0.3857,
+      "step": 7720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78234611589797e-05,
+      "loss": 0.5001,
+      "step": 7721
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7822840332716745e-05,
+      "loss": 0.3659,
+      "step": 7722
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782221942874113e-05,
+      "loss": 0.4349,
+      "step": 7723
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782159844705902e-05,
+      "loss": 0.401,
+      "step": 7724
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7820977387676583e-05,
+      "loss": 0.4565,
+      "step": 7725
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782035625059999e-05,
+      "loss": 0.4288,
+      "step": 7726
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7819735035835407e-05,
+      "loss": 0.413,
+      "step": 7727
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7819113743389012e-05,
+      "loss": 0.4992,
+      "step": 7728
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7818492373266973e-05,
+      "loss": 0.4733,
+      "step": 7729
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781787092547546e-05,
+      "loss": 0.3331,
+      "step": 7730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7817249400020657e-05,
+      "loss": 0.448,
+      "step": 7731
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7816627796908728e-05,
+      "loss": 0.4354,
+      "step": 7732
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781600611614585e-05,
+      "loss": 0.4253,
+      "step": 7733
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78153843577382e-05,
+      "loss": 0.4921,
+      "step": 7734
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781476252169196e-05,
+      "loss": 0.3717,
+      "step": 7735
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7814140608013296e-05,
+      "loss": 0.4381,
+      "step": 7736
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7813518616708396e-05,
+      "loss": 0.4419,
+      "step": 7737
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7812896547783433e-05,
+      "loss": 0.4316,
+      "step": 7738
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781227440124459e-05,
+      "loss": 0.3959,
+      "step": 7739
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7811652177098048e-05,
+      "loss": 0.4437,
+      "step": 7740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7811029875349982e-05,
+      "loss": 0.4407,
+      "step": 7741
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781040749600658e-05,
+      "loss": 0.3664,
+      "step": 7742
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7809785039074025e-05,
+      "loss": 0.4854,
+      "step": 7743
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78091625045585e-05,
+      "loss": 0.3764,
+      "step": 7744
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7808539892466186e-05,
+      "loss": 0.3969,
+      "step": 7745
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7807917202803268e-05,
+      "loss": 0.4309,
+      "step": 7746
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7807294435575938e-05,
+      "loss": 0.4155,
+      "step": 7747
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7806671590790377e-05,
+      "loss": 0.4141,
+      "step": 7748
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7806048668452774e-05,
+      "loss": 0.4019,
+      "step": 7749
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7805425668569318e-05,
+      "loss": 0.4165,
+      "step": 7750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804802591146196e-05,
+      "loss": 0.4169,
+      "step": 7751
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804179436189598e-05,
+      "loss": 0.4027,
+      "step": 7752
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7803556203705718e-05,
+      "loss": 0.4272,
+      "step": 7753
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7802932893700742e-05,
+      "loss": 0.3332,
+      "step": 7754
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7802309506180867e-05,
+      "loss": 0.4791,
+      "step": 7755
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780168604115228e-05,
+      "loss": 0.4135,
+      "step": 7756
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780106249862118e-05,
+      "loss": 0.4043,
+      "step": 7757
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780043887859376e-05,
+      "loss": 0.4155,
+      "step": 7758
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7799815181076213e-05,
+      "loss": 0.4868,
+      "step": 7759
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7799191406074736e-05,
+      "loss": 0.4454,
+      "step": 7760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7798567553595526e-05,
+      "loss": 0.4153,
+      "step": 7761
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779794362364478e-05,
+      "loss": 0.4238,
+      "step": 7762
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7797319616228695e-05,
+      "loss": 0.4207,
+      "step": 7763
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7796695531353476e-05,
+      "loss": 0.4208,
+      "step": 7764
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779607136902531e-05,
+      "loss": 0.4463,
+      "step": 7765
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7795447129250415e-05,
+      "loss": 0.3669,
+      "step": 7766
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7794822812034977e-05,
+      "loss": 0.3875,
+      "step": 7767
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7794198417385203e-05,
+      "loss": 0.4016,
+      "step": 7768
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77935739453073e-05,
+      "loss": 0.4333,
+      "step": 7769
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7792949395807466e-05,
+      "loss": 0.3556,
+      "step": 7770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7792324768891903e-05,
+      "loss": 0.3906,
+      "step": 7771
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779170006456683e-05,
+      "loss": 0.4124,
+      "step": 7772
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7791075282838434e-05,
+      "loss": 0.3955,
+      "step": 7773
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7790450423712936e-05,
+      "loss": 0.398,
+      "step": 7774
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7789825487196538e-05,
+      "loss": 0.4351,
+      "step": 7775
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7789200473295447e-05,
+      "loss": 0.4331,
+      "step": 7776
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7788575382015873e-05,
+      "loss": 0.4118,
+      "step": 7777
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7787950213364028e-05,
+      "loss": 0.3657,
+      "step": 7778
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778732496734612e-05,
+      "loss": 0.4707,
+      "step": 7779
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778669964396836e-05,
+      "loss": 0.4231,
+      "step": 7780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778607424323696e-05,
+      "loss": 0.3785,
+      "step": 7781
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7785448765158135e-05,
+      "loss": 0.4209,
+      "step": 7782
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77848232097381e-05,
+      "loss": 0.3254,
+      "step": 7783
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778419757698306e-05,
+      "loss": 0.3536,
+      "step": 7784
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7783571866899242e-05,
+      "loss": 0.4008,
+      "step": 7785
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7782946079492854e-05,
+      "loss": 0.3982,
+      "step": 7786
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7782320214770115e-05,
+      "loss": 0.4319,
+      "step": 7787
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7781694272737243e-05,
+      "loss": 0.3743,
+      "step": 7788
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7781068253400454e-05,
+      "loss": 0.3995,
+      "step": 7789
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778044215676597e-05,
+      "loss": 0.3893,
+      "step": 7790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779815982840006e-05,
+      "loss": 0.4156,
+      "step": 7791
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779189731628788e-05,
+      "loss": 0.4471,
+      "step": 7792
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7778563403138536e-05,
+      "loss": 0.3775,
+      "step": 7793
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777793699737547e-05,
+      "loss": 0.4354,
+      "step": 7794
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7777310514345816e-05,
+      "loss": 0.4721,
+      "step": 7795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7776683954055793e-05,
+      "loss": 0.4376,
+      "step": 7796
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7776057316511625e-05,
+      "loss": 0.3683,
+      "step": 7797
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7775430601719542e-05,
+      "loss": 0.3828,
+      "step": 7798
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7774803809685767e-05,
+      "loss": 0.439,
+      "step": 7799
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7774176940416528e-05,
+      "loss": 0.3937,
+      "step": 7800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7773549993918048e-05,
+      "loss": 0.4071,
+      "step": 7801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7772922970196563e-05,
+      "loss": 0.4448,
+      "step": 7802
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777229586925829e-05,
+      "loss": 0.3267,
+      "step": 7803
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7771668691109473e-05,
+      "loss": 0.4977,
+      "step": 7804
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7771041435756332e-05,
+      "loss": 0.3628,
+      "step": 7805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7770414103205103e-05,
+      "loss": 0.4084,
+      "step": 7806
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776978669346202e-05,
+      "loss": 0.3936,
+      "step": 7807
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7769159206533308e-05,
+      "loss": 0.3855,
+      "step": 7808
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7768531642425203e-05,
+      "loss": 0.389,
+      "step": 7809
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7767904001143942e-05,
+      "loss": 0.3422,
+      "step": 7810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776727628269576e-05,
+      "loss": 0.418,
+      "step": 7811
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7766648487086894e-05,
+      "loss": 0.3872,
+      "step": 7812
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7766020614323574e-05,
+      "loss": 0.4281,
+      "step": 7813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7765392664412047e-05,
+      "loss": 0.4806,
+      "step": 7814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7764764637358545e-05,
+      "loss": 0.4418,
+      "step": 7815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7764136533169303e-05,
+      "loss": 0.4234,
+      "step": 7816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776350835185057e-05,
+      "loss": 0.3848,
+      "step": 7817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7762880093408582e-05,
+      "loss": 0.4107,
+      "step": 7818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7762251757849577e-05,
+      "loss": 0.4137,
+      "step": 7819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77616233451798e-05,
+      "loss": 0.424,
+      "step": 7820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7760994855405497e-05,
+      "loss": 0.4348,
+      "step": 7821
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7760366288532904e-05,
+      "loss": 0.3447,
+      "step": 7822
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775973764456827e-05,
+      "loss": 0.3988,
+      "step": 7823
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7759108923517843e-05,
+      "loss": 0.4431,
+      "step": 7824
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775848012538786e-05,
+      "loss": 0.4858,
+      "step": 7825
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775785125018458e-05,
+      "loss": 0.4355,
+      "step": 7826
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7757222297914234e-05,
+      "loss": 0.4684,
+      "step": 7827
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775659326858308e-05,
+      "loss": 0.3925,
+      "step": 7828
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7755964162197372e-05,
+      "loss": 0.4726,
+      "step": 7829
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7755334978763347e-05,
+      "loss": 0.4087,
+      "step": 7830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7754705718287265e-05,
+      "loss": 0.4021,
+      "step": 7831
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7754076380775375e-05,
+      "loss": 0.3919,
+      "step": 7832
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7753446966233926e-05,
+      "loss": 0.3813,
+      "step": 7833
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775281747466917e-05,
+      "loss": 0.426,
+      "step": 7834
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7752187906087366e-05,
+      "loss": 0.4318,
+      "step": 7835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751558260494764e-05,
+      "loss": 0.3994,
+      "step": 7836
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775092853789762e-05,
+      "loss": 0.3598,
+      "step": 7837
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775029873830219e-05,
+      "loss": 0.4338,
+      "step": 7838
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7749668861714728e-05,
+      "loss": 0.4196,
+      "step": 7839
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7749038908141495e-05,
+      "loss": 0.4471,
+      "step": 7840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774840887758875e-05,
+      "loss": 0.4459,
+      "step": 7841
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774777877006274e-05,
+      "loss": 0.3984,
+      "step": 7842
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7747148585569743e-05,
+      "loss": 0.3949,
+      "step": 7843
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7746518324116008e-05,
+      "loss": 0.4117,
+      "step": 7844
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7745887985707796e-05,
+      "loss": 0.4166,
+      "step": 7845
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774525757035137e-05,
+      "loss": 0.4347,
+      "step": 7846
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7744627078052995e-05,
+      "loss": 0.4371,
+      "step": 7847
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7743996508818933e-05,
+      "loss": 0.4763,
+      "step": 7848
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7743365862655445e-05,
+      "loss": 0.4084,
+      "step": 7849
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.77427351395688e-05,
+      "loss": 0.426,
+      "step": 7850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7742104339565265e-05,
+      "loss": 0.3657,
+      "step": 7851
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7741473462651102e-05,
+      "loss": 0.4238,
+      "step": 7852
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774084250883258e-05,
+      "loss": 0.4096,
+      "step": 7853
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7740211478115967e-05,
+      "loss": 0.3389,
+      "step": 7854
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7739580370507533e-05,
+      "loss": 0.3478,
+      "step": 7855
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7738949186013542e-05,
+      "loss": 0.4501,
+      "step": 7856
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7738317924640273e-05,
+      "loss": 0.339,
+      "step": 7857
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773768658639399e-05,
+      "loss": 0.4107,
+      "step": 7858
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7737055171280965e-05,
+      "loss": 0.3693,
+      "step": 7859
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7736423679307476e-05,
+      "loss": 0.4087,
+      "step": 7860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7735792110479793e-05,
+      "loss": 0.4292,
+      "step": 7861
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773516046480419e-05,
+      "loss": 0.4141,
+      "step": 7862
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773452874228694e-05,
+      "loss": 0.3363,
+      "step": 7863
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773389694293432e-05,
+      "loss": 0.3901,
+      "step": 7864
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7733265066752606e-05,
+      "loss": 0.432,
+      "step": 7865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773263311374808e-05,
+      "loss": 0.4856,
+      "step": 7866
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7732001083927014e-05,
+      "loss": 0.3326,
+      "step": 7867
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7731368977295688e-05,
+      "loss": 0.4394,
+      "step": 7868
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773073679386038e-05,
+      "loss": 0.4192,
+      "step": 7869
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 0.4136,
+      "step": 7870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7729472196602944e-05,
+      "loss": 0.3449,
+      "step": 7871
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772883978279338e-05,
+      "loss": 0.4881,
+      "step": 7872
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772820729220496e-05,
+      "loss": 0.4219,
+      "step": 7873
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772757472484397e-05,
+      "loss": 0.4475,
+      "step": 7874
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7726942080716693e-05,
+      "loss": 0.4826,
+      "step": 7875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772630935982941e-05,
+      "loss": 0.4103,
+      "step": 7876
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7725676562188408e-05,
+      "loss": 0.4271,
+      "step": 7877
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7725043687799977e-05,
+      "loss": 0.3886,
+      "step": 7878
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7724410736670402e-05,
+      "loss": 0.3835,
+      "step": 7879
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7723777708805968e-05,
+      "loss": 0.4174,
+      "step": 7880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772314460421297e-05,
+      "loss": 0.3216,
+      "step": 7881
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7722511422897688e-05,
+      "loss": 0.4661,
+      "step": 7882
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7721878164866422e-05,
+      "loss": 0.4097,
+      "step": 7883
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7721244830125454e-05,
+      "loss": 0.3927,
+      "step": 7884
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7720611418681082e-05,
+      "loss": 0.3942,
+      "step": 7885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7719977930539595e-05,
+      "loss": 0.3773,
+      "step": 7886
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771934436570729e-05,
+      "loss": 0.4229,
+      "step": 7887
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718710724190457e-05,
+      "loss": 0.4664,
+      "step": 7888
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771807700599539e-05,
+      "loss": 0.4466,
+      "step": 7889
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7717443211128385e-05,
+      "loss": 0.4665,
+      "step": 7890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771680933959574e-05,
+      "loss": 0.4149,
+      "step": 7891
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771617539140375e-05,
+      "loss": 0.4529,
+      "step": 7892
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7715541366558718e-05,
+      "loss": 0.3952,
+      "step": 7893
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7714907265066938e-05,
+      "loss": 0.4223,
+      "step": 7894
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7714273086934705e-05,
+      "loss": 0.4096,
+      "step": 7895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771363883216833e-05,
+      "loss": 0.4782,
+      "step": 7896
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7713004500774097e-05,
+      "loss": 0.5082,
+      "step": 7897
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7712370092758325e-05,
+      "loss": 0.3624,
+      "step": 7898
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.771173560812731e-05,
+      "loss": 0.3636,
+      "step": 7899
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7711101046887353e-05,
+      "loss": 0.4196,
+      "step": 7900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7710466409044756e-05,
+      "loss": 0.3739,
+      "step": 7901
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7709831694605827e-05,
+      "loss": 0.4111,
+      "step": 7902
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7709196903576874e-05,
+      "loss": 0.4519,
+      "step": 7903
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7708562035964197e-05,
+      "loss": 0.4004,
+      "step": 7904
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7707927091774103e-05,
+      "loss": 0.4145,
+      "step": 7905
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770729207101291e-05,
+      "loss": 0.5576,
+      "step": 7906
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770665697368691e-05,
+      "loss": 0.3773,
+      "step": 7907
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770602179980242e-05,
+      "loss": 0.4249,
+      "step": 7908
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7705386549365756e-05,
+      "loss": 0.4137,
+      "step": 7909
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770475122238322e-05,
+      "loss": 0.37,
+      "step": 7910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770411581886113e-05,
+      "loss": 0.405,
+      "step": 7911
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7703480338805792e-05,
+      "loss": 0.4041,
+      "step": 7912
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7702844782223522e-05,
+      "loss": 0.4786,
+      "step": 7913
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770220914912063e-05,
+      "loss": 0.3794,
+      "step": 7914
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7701573439503437e-05,
+      "loss": 0.3914,
+      "step": 7915
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7700937653378254e-05,
+      "loss": 0.4159,
+      "step": 7916
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7700301790751396e-05,
+      "loss": 0.3837,
+      "step": 7917
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7699665851629185e-05,
+      "loss": 0.3735,
+      "step": 7918
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769902983601793e-05,
+      "loss": 0.4199,
+      "step": 7919
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769839374392396e-05,
+      "loss": 0.3991,
+      "step": 7920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769775757535358e-05,
+      "loss": 0.3556,
+      "step": 7921
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7697121330313126e-05,
+      "loss": 0.4272,
+      "step": 7922
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696485008808903e-05,
+      "loss": 0.3714,
+      "step": 7923
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7695848610847245e-05,
+      "loss": 0.3594,
+      "step": 7924
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769521213643447e-05,
+      "loss": 0.3283,
+      "step": 7925
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7694575585576892e-05,
+      "loss": 0.3783,
+      "step": 7926
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7693938958280846e-05,
+      "loss": 0.4564,
+      "step": 7927
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7693302254552653e-05,
+      "loss": 0.4608,
+      "step": 7928
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769266547439864e-05,
+      "loss": 0.4286,
+      "step": 7929
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7692028617825123e-05,
+      "loss": 0.474,
+      "step": 7930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769139168483844e-05,
+      "loss": 0.4526,
+      "step": 7931
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7690754675444913e-05,
+      "loss": 0.5086,
+      "step": 7932
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769011758965087e-05,
+      "loss": 0.368,
+      "step": 7933
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7689480427462644e-05,
+      "loss": 0.3944,
+      "step": 7934
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768884318888656e-05,
+      "loss": 0.3778,
+      "step": 7935
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768820587392895e-05,
+      "loss": 0.4217,
+      "step": 7936
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7687568482596143e-05,
+      "loss": 0.3122,
+      "step": 7937
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7686931014894474e-05,
+      "loss": 0.4395,
+      "step": 7938
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7686293470830275e-05,
+      "loss": 0.4249,
+      "step": 7939
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768565585040988e-05,
+      "loss": 0.3897,
+      "step": 7940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768501815363962e-05,
+      "loss": 0.3694,
+      "step": 7941
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7684380380525836e-05,
+      "loss": 0.4326,
+      "step": 7942
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7683742531074858e-05,
+      "loss": 0.317,
+      "step": 7943
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768310460529302e-05,
+      "loss": 0.4452,
+      "step": 7944
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768246660318667e-05,
+      "loss": 0.3937,
+      "step": 7945
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7681828524762136e-05,
+      "loss": 0.3797,
+      "step": 7946
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768119037002576e-05,
+      "loss": 0.3509,
+      "step": 7947
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7680552138983883e-05,
+      "loss": 0.4069,
+      "step": 7948
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679913831642842e-05,
+      "loss": 0.4163,
+      "step": 7949
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767927544800898e-05,
+      "loss": 0.4101,
+      "step": 7950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767863698808864e-05,
+      "loss": 0.4321,
+      "step": 7951
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7677998451888164e-05,
+      "loss": 0.3467,
+      "step": 7952
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767735983941389e-05,
+      "loss": 0.3494,
+      "step": 7953
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7676721150672173e-05,
+      "loss": 0.4465,
+      "step": 7954
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7676082385669343e-05,
+      "loss": 0.4167,
+      "step": 7955
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7675443544411763e-05,
+      "loss": 0.3689,
+      "step": 7956
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767480462690576e-05,
+      "loss": 0.3731,
+      "step": 7957
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.76741656331577e-05,
+      "loss": 0.3741,
+      "step": 7958
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7673526563173915e-05,
+      "loss": 0.4183,
+      "step": 7959
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7672887416960763e-05,
+      "loss": 0.4513,
+      "step": 7960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767224819452459e-05,
+      "loss": 0.362,
+      "step": 7961
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.767160889587175e-05,
+      "loss": 0.4212,
+      "step": 7962
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7670969521008586e-05,
+      "loss": 0.3912,
+      "step": 7963
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.767033006994146e-05,
+      "loss": 0.3511,
+      "step": 7964
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7669690542676713e-05,
+      "loss": 0.4304,
+      "step": 7965
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7669050939220706e-05,
+      "loss": 0.441,
+      "step": 7966
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766841125957979e-05,
+      "loss": 0.3368,
+      "step": 7967
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7667771503760324e-05,
+      "loss": 0.4232,
+      "step": 7968
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7667131671768654e-05,
+      "loss": 0.4113,
+      "step": 7969
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7666491763611145e-05,
+      "loss": 0.4582,
+      "step": 7970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766585177929415e-05,
+      "loss": 0.4224,
+      "step": 7971
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7665211718824026e-05,
+      "loss": 0.4738,
+      "step": 7972
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7664571582207135e-05,
+      "loss": 0.4009,
+      "step": 7973
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7663931369449832e-05,
+      "loss": 0.3711,
+      "step": 7974
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7663291080558477e-05,
+      "loss": 0.3934,
+      "step": 7975
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7662650715539437e-05,
+      "loss": 0.3996,
+      "step": 7976
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7662010274399066e-05,
+      "loss": 0.4195,
+      "step": 7977
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766136975714373e-05,
+      "loss": 0.4119,
+      "step": 7978
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766072916377979e-05,
+      "loss": 0.408,
+      "step": 7979
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7660088494313613e-05,
+      "loss": 0.403,
+      "step": 7980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765944774875156e-05,
+      "loss": 0.43,
+      "step": 7981
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7658806927099994e-05,
+      "loss": 0.3446,
+      "step": 7982
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765816602936529e-05,
+      "loss": 0.3815,
+      "step": 7983
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7657525055553804e-05,
+      "loss": 0.3926,
+      "step": 7984
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7656884005671912e-05,
+      "loss": 0.3935,
+      "step": 7985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7656242879725977e-05,
+      "loss": 0.4381,
+      "step": 7986
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765560167772237e-05,
+      "loss": 0.4342,
+      "step": 7987
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765496039966746e-05,
+      "loss": 0.3831,
+      "step": 7988
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765431904556762e-05,
+      "loss": 0.3641,
+      "step": 7989
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7653677615429217e-05,
+      "loss": 0.4214,
+      "step": 7990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7653036109258628e-05,
+      "loss": 0.4195,
+      "step": 7991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7652394527062224e-05,
+      "loss": 0.4119,
+      "step": 7992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765175286884637e-05,
+      "loss": 0.4106,
+      "step": 7993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7651111134617457e-05,
+      "loss": 0.4429,
+      "step": 7994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7650469324381843e-05,
+      "loss": 0.3749,
+      "step": 7995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7649827438145917e-05,
+      "loss": 0.3962,
+      "step": 7996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764918547591605e-05,
+      "loss": 0.4761,
+      "step": 7997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7648543437698617e-05,
+      "loss": 0.4551,
+      "step": 7998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76479013235e-05,
+      "loss": 0.43,
+      "step": 7999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7647259133326575e-05,
+      "loss": 0.3926,
+      "step": 8000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7646616867184727e-05,
+      "loss": 0.3756,
+      "step": 8001
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764597452508083e-05,
+      "loss": 0.3631,
+      "step": 8002
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7645332107021264e-05,
+      "loss": 0.3666,
+      "step": 8003
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764468961301242e-05,
+      "loss": 0.385,
+      "step": 8004
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764404704306067e-05,
+      "loss": 0.3782,
+      "step": 8005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7643404397172403e-05,
+      "loss": 0.3524,
+      "step": 8006
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7642761675354002e-05,
+      "loss": 0.2993,
+      "step": 8007
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7642118877611854e-05,
+      "loss": 0.3561,
+      "step": 8008
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7641476003952342e-05,
+      "loss": 0.368,
+      "step": 8009
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7640833054381856e-05,
+      "loss": 0.3715,
+      "step": 8010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7640190028906775e-05,
+      "loss": 0.4616,
+      "step": 8011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7639546927533493e-05,
+      "loss": 0.4566,
+      "step": 8012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7638903750268398e-05,
+      "loss": 0.3795,
+      "step": 8013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7638260497117878e-05,
+      "loss": 0.4626,
+      "step": 8014
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7637617168088327e-05,
+      "loss": 0.4372,
+      "step": 8015
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763697376318613e-05,
+      "loss": 0.3954,
+      "step": 8016
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7636330282417683e-05,
+      "loss": 0.3467,
+      "step": 8017
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763568672578938e-05,
+      "loss": 0.4076,
+      "step": 8018
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7635043093307605e-05,
+      "loss": 0.3447,
+      "step": 8019
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763439938497876e-05,
+      "loss": 0.4341,
+      "step": 8020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763375560080924e-05,
+      "loss": 0.3562,
+      "step": 8021
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7633111740805436e-05,
+      "loss": 0.4291,
+      "step": 8022
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7632467804973747e-05,
+      "loss": 0.3393,
+      "step": 8023
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763182379332057e-05,
+      "loss": 0.3797,
+      "step": 8024
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76311797058523e-05,
+      "loss": 0.3904,
+      "step": 8025
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763053554257534e-05,
+      "loss": 0.4812,
+      "step": 8026
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7629891303496083e-05,
+      "loss": 0.325,
+      "step": 8027
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762924698862093e-05,
+      "loss": 0.4552,
+      "step": 8028
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762860259795629e-05,
+      "loss": 0.3654,
+      "step": 8029
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7627958131508553e-05,
+      "loss": 0.4336,
+      "step": 8030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762731358928413e-05,
+      "loss": 0.4004,
+      "step": 8031
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762666897128942e-05,
+      "loss": 0.4512,
+      "step": 8032
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7626024277530826e-05,
+      "loss": 0.3715,
+      "step": 8033
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7625379508014754e-05,
+      "loss": 0.3537,
+      "step": 8034
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762473466274761e-05,
+      "loss": 0.3591,
+      "step": 8035
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76240897417358e-05,
+      "loss": 0.3442,
+      "step": 8036
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7623444744985726e-05,
+      "loss": 0.3456,
+      "step": 8037
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7622799672503798e-05,
+      "loss": 0.4019,
+      "step": 8038
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7622154524296426e-05,
+      "loss": 0.3764,
+      "step": 8039
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7621509300370018e-05,
+      "loss": 0.394,
+      "step": 8040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7620864000730987e-05,
+      "loss": 0.4082,
+      "step": 8041
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7620218625385737e-05,
+      "loss": 0.3777,
+      "step": 8042
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7619573174340684e-05,
+      "loss": 0.4267,
+      "step": 8043
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7618927647602237e-05,
+      "loss": 0.435,
+      "step": 8044
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7618282045176812e-05,
+      "loss": 0.4585,
+      "step": 8045
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7617636367070815e-05,
+      "loss": 0.3184,
+      "step": 8046
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761699061329067e-05,
+      "loss": 0.5023,
+      "step": 8047
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7616344783842788e-05,
+      "loss": 0.4391,
+      "step": 8048
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761569887873358e-05,
+      "loss": 0.3747,
+      "step": 8049
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7615052897969472e-05,
+      "loss": 0.4,
+      "step": 8050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7614406841556878e-05,
+      "loss": 0.3398,
+      "step": 8051
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7613760709502208e-05,
+      "loss": 0.3569,
+      "step": 8052
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761311450181189e-05,
+      "loss": 0.4551,
+      "step": 8053
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7612468218492335e-05,
+      "loss": 0.4319,
+      "step": 8054
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7611821859549977e-05,
+      "loss": 0.3963,
+      "step": 8055
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7611175424991222e-05,
+      "loss": 0.4145,
+      "step": 8056
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76105289148225e-05,
+      "loss": 0.4111,
+      "step": 8057
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609882329050233e-05,
+      "loss": 0.2963,
+      "step": 8058
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609235667680842e-05,
+      "loss": 0.4509,
+      "step": 8059
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7608588930720748e-05,
+      "loss": 0.5054,
+      "step": 8060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7607942118176383e-05,
+      "loss": 0.3952,
+      "step": 8061
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7607295230054172e-05,
+      "loss": 0.3837,
+      "step": 8062
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760664826636053e-05,
+      "loss": 0.4092,
+      "step": 8063
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7606001227101903e-05,
+      "loss": 0.4234,
+      "step": 8064
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76053541122847e-05,
+      "loss": 0.4186,
+      "step": 8065
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760470692191536e-05,
+      "loss": 0.3669,
+      "step": 8066
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7604059656000313e-05,
+      "loss": 0.3366,
+      "step": 8067
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760341231454598e-05,
+      "loss": 0.3603,
+      "step": 8068
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76027648975588e-05,
+      "loss": 0.4293,
+      "step": 8069
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76021174050452e-05,
+      "loss": 0.3903,
+      "step": 8070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760146983701162e-05,
+      "loss": 0.4019,
+      "step": 8071
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760082219346448e-05,
+      "loss": 0.4023,
+      "step": 8072
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7600174474410225e-05,
+      "loss": 0.4254,
+      "step": 8073
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7599526679855285e-05,
+      "loss": 0.4851,
+      "step": 8074
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598878809806095e-05,
+      "loss": 0.3628,
+      "step": 8075
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759823086426909e-05,
+      "loss": 0.4024,
+      "step": 8076
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7597582843250708e-05,
+      "loss": 0.3943,
+      "step": 8077
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7596934746757393e-05,
+      "loss": 0.394,
+      "step": 8078
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759628657479557e-05,
+      "loss": 0.3754,
+      "step": 8079
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759563832737169e-05,
+      "loss": 0.3335,
+      "step": 8080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7594990004492184e-05,
+      "loss": 0.356,
+      "step": 8081
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.75943416061635e-05,
+      "loss": 0.4436,
+      "step": 8082
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759369313239207e-05,
+      "loss": 0.3702,
+      "step": 8083
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7593044583184345e-05,
+      "loss": 0.4333,
+      "step": 8084
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7592395958546767e-05,
+      "loss": 0.378,
+      "step": 8085
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7591747258485773e-05,
+      "loss": 0.3983,
+      "step": 8086
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7591098483007812e-05,
+      "loss": 0.3641,
+      "step": 8087
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7590449632119327e-05,
+      "loss": 0.4299,
+      "step": 8088
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7589800705826763e-05,
+      "loss": 0.4591,
+      "step": 8089
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7589151704136575e-05,
+      "loss": 0.402,
+      "step": 8090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.75885026270552e-05,
+      "loss": 0.4025,
+      "step": 8091
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587853474589087e-05,
+      "loss": 0.4051,
+      "step": 8092
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758720424674469e-05,
+      "loss": 0.3824,
+      "step": 8093
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758655494352845e-05,
+      "loss": 0.3582,
+      "step": 8094
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758590556494683e-05,
+      "loss": 0.4281,
+      "step": 8095
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758525611100627e-05,
+      "loss": 0.3639,
+      "step": 8096
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7584606581713223e-05,
+      "loss": 0.4573,
+      "step": 8097
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583956977074147e-05,
+      "loss": 0.4346,
+      "step": 8098
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583307297095492e-05,
+      "loss": 0.4372,
+      "step": 8099
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7582657541783716e-05,
+      "loss": 0.3698,
+      "step": 8100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7582007711145265e-05,
+      "loss": 0.3882,
+      "step": 8101
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.75813578051866e-05,
+      "loss": 0.3828,
+      "step": 8102
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580707823914177e-05,
+      "loss": 0.3949,
+      "step": 8103
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580057767334452e-05,
+      "loss": 0.4422,
+      "step": 8104
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7579407635453884e-05,
+      "loss": 0.4553,
+      "step": 8105
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757875742827893e-05,
+      "loss": 0.4082,
+      "step": 8106
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578107145816054e-05,
+      "loss": 0.4594,
+      "step": 8107
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7577456788071704e-05,
+      "loss": 0.356,
+      "step": 8108
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7576806355052353e-05,
+      "loss": 0.3621,
+      "step": 8109
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757615584676446e-05,
+      "loss": 0.3961,
+      "step": 8110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757550526321448e-05,
+      "loss": 0.4398,
+      "step": 8111
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7574854604408885e-05,
+      "loss": 0.4049,
+      "step": 8112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7574203870354133e-05,
+      "loss": 0.366,
+      "step": 8113
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7573553061056688e-05,
+      "loss": 0.4058,
+      "step": 8114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572902176523023e-05,
+      "loss": 0.3853,
+      "step": 8115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572251216759597e-05,
+      "loss": 0.4145,
+      "step": 8116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7571600181772877e-05,
+      "loss": 0.3431,
+      "step": 8117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757094907156933e-05,
+      "loss": 0.4388,
+      "step": 8118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7570297886155426e-05,
+      "loss": 0.4299,
+      "step": 8119
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7569646625537634e-05,
+      "loss": 0.4099,
+      "step": 8120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7568995289722423e-05,
+      "loss": 0.3781,
+      "step": 8121
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7568343878716266e-05,
+      "loss": 0.3721,
+      "step": 8122
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756769239252563e-05,
+      "loss": 0.4104,
+      "step": 8123
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756704083115699e-05,
+      "loss": 0.3515,
+      "step": 8124
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7566389194616816e-05,
+      "loss": 0.4184,
+      "step": 8125
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756573748291158e-05,
+      "loss": 0.4048,
+      "step": 8126
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7565085696047767e-05,
+      "loss": 0.3667,
+      "step": 8127
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7564433834031836e-05,
+      "loss": 0.4799,
+      "step": 8128
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7563781896870273e-05,
+      "loss": 0.4812,
+      "step": 8129
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7563129884569555e-05,
+      "loss": 0.3711,
+      "step": 8130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756247779713615e-05,
+      "loss": 0.4089,
+      "step": 8131
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756182563457655e-05,
+      "loss": 0.4241,
+      "step": 8132
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7561173396897216e-05,
+      "loss": 0.4525,
+      "step": 8133
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7560521084104642e-05,
+      "loss": 0.3837,
+      "step": 8134
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7559868696205304e-05,
+      "loss": 0.3971,
+      "step": 8135
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755921623320568e-05,
+      "loss": 0.4268,
+      "step": 8136
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7558563695112256e-05,
+      "loss": 0.3819,
+      "step": 8137
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755791108193151e-05,
+      "loss": 0.4104,
+      "step": 8138
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7557258393669923e-05,
+      "loss": 0.3994,
+      "step": 8139
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755660563033399e-05,
+      "loss": 0.378,
+      "step": 8140
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7555952791930186e-05,
+      "loss": 0.369,
+      "step": 8141
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7555299878464997e-05,
+      "loss": 0.4193,
+      "step": 8142
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755464688994491e-05,
+      "loss": 0.3893,
+      "step": 8143
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7553993826376415e-05,
+      "loss": 0.3722,
+      "step": 8144
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7553340687765996e-05,
+      "loss": 0.3562,
+      "step": 8145
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7552687474120143e-05,
+      "loss": 0.3424,
+      "step": 8146
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7552034185445345e-05,
+      "loss": 0.4585,
+      "step": 8147
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7551380821748094e-05,
+      "loss": 0.4102,
+      "step": 8148
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7550727383034873e-05,
+      "loss": 0.3637,
+      "step": 8149
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7550073869312183e-05,
+      "loss": 0.4127,
+      "step": 8150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7549420280586508e-05,
+      "loss": 0.4227,
+      "step": 8151
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7548766616864345e-05,
+      "loss": 0.4133,
+      "step": 8152
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.754811287815219e-05,
+      "loss": 0.3071,
+      "step": 8153
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7547459064456533e-05,
+      "loss": 0.3503,
+      "step": 8154
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7546805175783866e-05,
+      "loss": 0.448,
+      "step": 8155
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7546151212140693e-05,
+      "loss": 0.5092,
+      "step": 8156
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7545497173533502e-05,
+      "loss": 0.3903,
+      "step": 8157
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75448430599688e-05,
+      "loss": 0.3845,
+      "step": 8158
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7544188871453078e-05,
+      "loss": 0.3737,
+      "step": 8159
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7543534607992835e-05,
+      "loss": 0.3974,
+      "step": 8160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754288026959457e-05,
+      "loss": 0.4529,
+      "step": 8161
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754222585626479e-05,
+      "loss": 0.4058,
+      "step": 8162
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754157136800999e-05,
+      "loss": 0.3734,
+      "step": 8163
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754091680483667e-05,
+      "loss": 0.4264,
+      "step": 8164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7540262166751336e-05,
+      "loss": 0.3827,
+      "step": 8165
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7539607453760493e-05,
+      "loss": 0.3972,
+      "step": 8166
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753895266587064e-05,
+      "loss": 0.3644,
+      "step": 8167
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7538297803088285e-05,
+      "loss": 0.5331,
+      "step": 8168
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7537642865419933e-05,
+      "loss": 0.4899,
+      "step": 8169
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753698785287209e-05,
+      "loss": 0.454,
+      "step": 8170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753633276545126e-05,
+      "loss": 0.3481,
+      "step": 8171
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535677603163962e-05,
+      "loss": 0.3374,
+      "step": 8172
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535022366016686e-05,
+      "loss": 0.3931,
+      "step": 8173
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7534367054015958e-05,
+      "loss": 0.3445,
+      "step": 8174
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753371166716828e-05,
+      "loss": 0.4388,
+      "step": 8175
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753305620548016e-05,
+      "loss": 0.4295,
+      "step": 8176
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532400668958116e-05,
+      "loss": 0.3902,
+      "step": 8177
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753174505760866e-05,
+      "loss": 0.413,
+      "step": 8178
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7531089371438298e-05,
+      "loss": 0.3796,
+      "step": 8179
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753043361045355e-05,
+      "loss": 0.4287,
+      "step": 8180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7529777774660924e-05,
+      "loss": 0.422,
+      "step": 8181
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752912186406694e-05,
+      "loss": 0.4582,
+      "step": 8182
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752846587867812e-05,
+      "loss": 0.4267,
+      "step": 8183
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752780981850097e-05,
+      "loss": 0.3715,
+      "step": 8184
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7527153683542007e-05,
+      "loss": 0.4563,
+      "step": 8185
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7526497473807758e-05,
+      "loss": 0.3905,
+      "step": 8186
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7525841189304735e-05,
+      "loss": 0.3324,
+      "step": 8187
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752518483003946e-05,
+      "loss": 0.3861,
+      "step": 8188
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7524528396018455e-05,
+      "loss": 0.3314,
+      "step": 8189
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7523871887248236e-05,
+      "loss": 0.4562,
+      "step": 8190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752321530373533e-05,
+      "loss": 0.3536,
+      "step": 8191
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7522558645486257e-05,
+      "loss": 0.3821,
+      "step": 8192
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752190191250754e-05,
+      "loss": 0.4591,
+      "step": 8193
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75212451048057e-05,
+      "loss": 0.4179,
+      "step": 8194
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752058822238727e-05,
+      "loss": 0.3368,
+      "step": 8195
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7519931265258772e-05,
+      "loss": 0.385,
+      "step": 8196
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7519274233426728e-05,
+      "loss": 0.405,
+      "step": 8197
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751861712689767e-05,
+      "loss": 0.3489,
+      "step": 8198
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7517959945678125e-05,
+      "loss": 0.4334,
+      "step": 8199
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7517302689774617e-05,
+      "loss": 0.4168,
+      "step": 8200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751664535919368e-05,
+      "loss": 0.4007,
+      "step": 8201
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7515987953941845e-05,
+      "loss": 0.3881,
+      "step": 8202
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751533047402564e-05,
+      "loss": 0.4628,
+      "step": 8203
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7514672919451595e-05,
+      "loss": 0.4672,
+      "step": 8204
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7514015290226244e-05,
+      "loss": 0.4304,
+      "step": 8205
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7513357586356118e-05,
+      "loss": 0.4261,
+      "step": 8206
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751269980784776e-05,
+      "loss": 0.3902,
+      "step": 8207
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751204195470769e-05,
+      "loss": 0.3548,
+      "step": 8208
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7511384026942454e-05,
+      "loss": 0.4782,
+      "step": 8209
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751072602455858e-05,
+      "loss": 0.4412,
+      "step": 8210
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7510067947562612e-05,
+      "loss": 0.4033,
+      "step": 8211
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7509409795961086e-05,
+      "loss": 0.4318,
+      "step": 8212
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7508751569760534e-05,
+      "loss": 0.4969,
+      "step": 8213
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75080932689675e-05,
+      "loss": 0.4397,
+      "step": 8214
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7507434893588525e-05,
+      "loss": 0.3844,
+      "step": 8215
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7506776443630146e-05,
+      "loss": 0.382,
+      "step": 8216
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7506117919098905e-05,
+      "loss": 0.494,
+      "step": 8217
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7505459320001344e-05,
+      "loss": 0.389,
+      "step": 8218
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7504800646344007e-05,
+      "loss": 0.3692,
+      "step": 8219
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7504141898133435e-05,
+      "loss": 0.4445,
+      "step": 8220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750348307537617e-05,
+      "loss": 0.4857,
+      "step": 8221
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7502824178078764e-05,
+      "loss": 0.4168,
+      "step": 8222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7502165206247756e-05,
+      "loss": 0.3593,
+      "step": 8223
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7501506159889695e-05,
+      "loss": 0.4386,
+      "step": 8224
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7500847039011128e-05,
+      "loss": 0.4581,
+      "step": 8225
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7500187843618605e-05,
+      "loss": 0.42,
+      "step": 8226
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749952857371867e-05,
+      "loss": 0.3238,
+      "step": 8227
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7498869229317868e-05,
+      "loss": 0.3896,
+      "step": 8228
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7498209810422763e-05,
+      "loss": 0.453,
+      "step": 8229
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7497550317039892e-05,
+      "loss": 0.4169,
+      "step": 8230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7496890749175818e-05,
+      "loss": 0.4391,
+      "step": 8231
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7496231106837084e-05,
+      "loss": 0.4381,
+      "step": 8232
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7495571390030245e-05,
+      "loss": 0.3527,
+      "step": 8233
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7494911598761855e-05,
+      "loss": 0.447,
+      "step": 8234
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7494251733038473e-05,
+      "loss": 0.4725,
+      "step": 8235
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749359179286665e-05,
+      "loss": 0.3238,
+      "step": 8236
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7492931778252945e-05,
+      "loss": 0.5439,
+      "step": 8237
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7492271689203905e-05,
+      "loss": 0.3679,
+      "step": 8238
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7491611525726103e-05,
+      "loss": 0.391,
+      "step": 8239
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7490951287826083e-05,
+      "loss": 0.423,
+      "step": 8240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749029097551041e-05,
+      "loss": 0.3563,
+      "step": 8241
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7489630588785644e-05,
+      "loss": 0.4147,
+      "step": 8242
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7488970127658344e-05,
+      "loss": 0.4445,
+      "step": 8243
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7488309592135073e-05,
+      "loss": 0.4413,
+      "step": 8244
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748764898222239e-05,
+      "loss": 0.4029,
+      "step": 8245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7486988297926855e-05,
+      "loss": 0.3897,
+      "step": 8246
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748632753925504e-05,
+      "loss": 0.3518,
+      "step": 8247
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7485666706213504e-05,
+      "loss": 0.4624,
+      "step": 8248
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748500579880881e-05,
+      "loss": 0.333,
+      "step": 8249
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748434481704753e-05,
+      "loss": 0.4007,
+      "step": 8250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748368376093622e-05,
+      "loss": 0.4175,
+      "step": 8251
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748302263048145e-05,
+      "loss": 0.3519,
+      "step": 8252
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7482361425689795e-05,
+      "loss": 0.3608,
+      "step": 8253
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7481700146567815e-05,
+      "loss": 0.3851,
+      "step": 8254
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748103879312209e-05,
+      "loss": 0.4758,
+      "step": 8255
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7480377365359178e-05,
+      "loss": 0.3469,
+      "step": 8256
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7479715863285655e-05,
+      "loss": 0.3801,
+      "step": 8257
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7479054286908095e-05,
+      "loss": 0.4356,
+      "step": 8258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7478392636233064e-05,
+      "loss": 0.3405,
+      "step": 8259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7477730911267138e-05,
+      "loss": 0.4487,
+      "step": 8260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747706911201689e-05,
+      "loss": 0.3613,
+      "step": 8261
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7476407238488896e-05,
+      "loss": 0.401,
+      "step": 8262
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475745290689733e-05,
+      "loss": 0.4047,
+      "step": 8263
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747508326862597e-05,
+      "loss": 0.3987,
+      "step": 8264
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7474421172304188e-05,
+      "loss": 0.3843,
+      "step": 8265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7473759001730966e-05,
+      "loss": 0.4036,
+      "step": 8266
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7473096756912876e-05,
+      "loss": 0.4937,
+      "step": 8267
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7472434437856503e-05,
+      "loss": 0.4466,
+      "step": 8268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7471772044568425e-05,
+      "loss": 0.335,
+      "step": 8269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747110957705522e-05,
+      "loss": 0.3949,
+      "step": 8270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747044703532347e-05,
+      "loss": 0.4403,
+      "step": 8271
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746978441937976e-05,
+      "loss": 0.4456,
+      "step": 8272
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7469121729230667e-05,
+      "loss": 0.4244,
+      "step": 8273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468458964882778e-05,
+      "loss": 0.3443,
+      "step": 8274
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7467796126342677e-05,
+      "loss": 0.3692,
+      "step": 8275
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7467133213616943e-05,
+      "loss": 0.4012,
+      "step": 8276
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7466470226712167e-05,
+      "loss": 0.421,
+      "step": 8277
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746580716563494e-05,
+      "loss": 0.3384,
+      "step": 8278
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7465144030391837e-05,
+      "loss": 0.4201,
+      "step": 8279
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7464480820989456e-05,
+      "loss": 0.4111,
+      "step": 8280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7463817537434375e-05,
+      "loss": 0.3698,
+      "step": 8281
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746315417973319e-05,
+      "loss": 0.3884,
+      "step": 8282
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7462490747892498e-05,
+      "loss": 0.3977,
+      "step": 8283
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7461827241918878e-05,
+      "loss": 0.4442,
+      "step": 8284
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7461163661818917e-05,
+      "loss": 0.4483,
+      "step": 8285
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7460500007599224e-05,
+      "loss": 0.3945,
+      "step": 8286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7459836279266377e-05,
+      "loss": 0.3715,
+      "step": 8287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745917247682698e-05,
+      "loss": 0.431,
+      "step": 8288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745850860028762e-05,
+      "loss": 0.4061,
+      "step": 8289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7457844649654895e-05,
+      "loss": 0.4206,
+      "step": 8290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74571806249354e-05,
+      "loss": 0.4202,
+      "step": 8291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745651652613573e-05,
+      "loss": 0.3807,
+      "step": 8292
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7455852353262486e-05,
+      "loss": 0.3607,
+      "step": 8293
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745518810632226e-05,
+      "loss": 0.4616,
+      "step": 8294
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745452378532166e-05,
+      "loss": 0.351,
+      "step": 8295
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745385939026728e-05,
+      "loss": 0.3835,
+      "step": 8296
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7453194921165716e-05,
+      "loss": 0.3478,
+      "step": 8297
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7452530378023572e-05,
+      "loss": 0.5188,
+      "step": 8298
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7451865760847452e-05,
+      "loss": 0.4013,
+      "step": 8299
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7451201069643956e-05,
+      "loss": 0.419,
+      "step": 8300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7450536304419694e-05,
+      "loss": 0.4609,
+      "step": 8301
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744987146518126e-05,
+      "loss": 0.4304,
+      "step": 8302
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744920655193526e-05,
+      "loss": 0.4619,
+      "step": 8303
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74485415646883e-05,
+      "loss": 0.3489,
+      "step": 8304
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7447876503446996e-05,
+      "loss": 0.3713,
+      "step": 8305
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744721136821794e-05,
+      "loss": 0.4383,
+      "step": 8306
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7446546159007748e-05,
+      "loss": 0.364,
+      "step": 8307
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744588087582303e-05,
+      "loss": 0.3594,
+      "step": 8308
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744521551867038e-05,
+      "loss": 0.3839,
+      "step": 8309
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744455008755643e-05,
+      "loss": 0.3525,
+      "step": 8310
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7443884582487773e-05,
+      "loss": 0.4469,
+      "step": 8311
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744321900347103e-05,
+      "loss": 0.3683,
+      "step": 8312
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744255335051281e-05,
+      "loss": 0.42,
+      "step": 8313
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744188762361972e-05,
+      "loss": 0.3534,
+      "step": 8314
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7441221822798385e-05,
+      "loss": 0.4161,
+      "step": 8315
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744055594805541e-05,
+      "loss": 0.4536,
+      "step": 8316
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743988999939741e-05,
+      "loss": 0.4259,
+      "step": 8317
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7439223976831002e-05,
+      "loss": 0.3727,
+      "step": 8318
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7438557880362808e-05,
+      "loss": 0.338,
+      "step": 8319
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743789170999944e-05,
+      "loss": 0.3465,
+      "step": 8320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7437225465747515e-05,
+      "loss": 0.4804,
+      "step": 8321
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743655914761365e-05,
+      "loss": 0.4522,
+      "step": 8322
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7435892755604467e-05,
+      "loss": 0.368,
+      "step": 8323
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743522628972659e-05,
+      "loss": 0.3566,
+      "step": 8324
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7434559749986633e-05,
+      "loss": 0.3554,
+      "step": 8325
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743389313639122e-05,
+      "loss": 0.44,
+      "step": 8326
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7433226448946973e-05,
+      "loss": 0.4237,
+      "step": 8327
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7432559687660514e-05,
+      "loss": 0.3328,
+      "step": 8328
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743189285253847e-05,
+      "loss": 0.3788,
+      "step": 8329
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7431225943587468e-05,
+      "loss": 0.3904,
+      "step": 8330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743055896081412e-05,
+      "loss": 0.464,
+      "step": 8331
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7429891904225065e-05,
+      "loss": 0.4183,
+      "step": 8332
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7429224773826923e-05,
+      "loss": 0.4792,
+      "step": 8333
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7428557569626323e-05,
+      "loss": 0.4088,
+      "step": 8334
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7427890291629895e-05,
+      "loss": 0.4227,
+      "step": 8335
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7427222939844263e-05,
+      "loss": 0.4079,
+      "step": 8336
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742655551427606e-05,
+      "loss": 0.4084,
+      "step": 8337
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742588801493192e-05,
+      "loss": 0.4175,
+      "step": 8338
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7425220441818465e-05,
+      "loss": 0.4676,
+      "step": 8339
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7424552794942333e-05,
+      "loss": 0.4853,
+      "step": 8340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7423885074310156e-05,
+      "loss": 0.3433,
+      "step": 8341
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7423217279928562e-05,
+      "loss": 0.3909,
+      "step": 8342
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7422549411804192e-05,
+      "loss": 0.404,
+      "step": 8343
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742188146994368e-05,
+      "loss": 0.3361,
+      "step": 8344
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7421213454353657e-05,
+      "loss": 0.3552,
+      "step": 8345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7420545365040758e-05,
+      "loss": 0.3728,
+      "step": 8346
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7419877202011628e-05,
+      "loss": 0.448,
+      "step": 8347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7419208965272895e-05,
+      "loss": 0.3975,
+      "step": 8348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7418540654831206e-05,
+      "loss": 0.3832,
+      "step": 8349
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7417872270693195e-05,
+      "loss": 0.4027,
+      "step": 8350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74172038128655e-05,
+      "loss": 0.373,
+      "step": 8351
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7416535281354764e-05,
+      "loss": 0.4478,
+      "step": 8352
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7415866676167632e-05,
+      "loss": 0.373,
+      "step": 8353
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7415197997310738e-05,
+      "loss": 0.4474,
+      "step": 8354
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7414529244790733e-05,
+      "loss": 0.3705,
+      "step": 8355
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7413860418614257e-05,
+      "loss": 0.4289,
+      "step": 8356
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741319151878795e-05,
+      "loss": 0.4754,
+      "step": 8357
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7412522545318463e-05,
+      "loss": 0.4337,
+      "step": 8358
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741185349821244e-05,
+      "loss": 0.4785,
+      "step": 8359
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7411184377476526e-05,
+      "loss": 0.316,
+      "step": 8360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741051518311737e-05,
+      "loss": 0.4134,
+      "step": 8361
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7409845915141613e-05,
+      "loss": 0.3473,
+      "step": 8362
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7409176573555916e-05,
+      "loss": 0.4273,
+      "step": 8363
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7408507158366918e-05,
+      "loss": 0.421,
+      "step": 8364
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7407837669581274e-05,
+      "loss": 0.4336,
+      "step": 8365
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740716810720563e-05,
+      "loss": 0.4397,
+      "step": 8366
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7406498471246643e-05,
+      "loss": 0.3904,
+      "step": 8367
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740582876171096e-05,
+      "loss": 0.3529,
+      "step": 8368
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740515897860524e-05,
+      "loss": 0.4178,
+      "step": 8369
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404489121936135e-05,
+      "loss": 0.3912,
+      "step": 8370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7403819191710295e-05,
+      "loss": 0.3265,
+      "step": 8371
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7403149187934375e-05,
+      "loss": 0.4358,
+      "step": 8372
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740247911061504e-05,
+      "loss": 0.4683,
+      "step": 8373
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7401808959758936e-05,
+      "loss": 0.3484,
+      "step": 8374
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740113873537273e-05,
+      "loss": 0.4171,
+      "step": 8375
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740046843746307e-05,
+      "loss": 0.4073,
+      "step": 8376
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739979806603662e-05,
+      "loss": 0.3838,
+      "step": 8377
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739912762110004e-05,
+      "loss": 0.4186,
+      "step": 8378
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7398457102659992e-05,
+      "loss": 0.4682,
+      "step": 8379
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739778651072313e-05,
+      "loss": 0.3874,
+      "step": 8380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739711584529612e-05,
+      "loss": 0.3211,
+      "step": 8381
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739644510638563e-05,
+      "loss": 0.394,
+      "step": 8382
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739577429399831e-05,
+      "loss": 0.3887,
+      "step": 8383
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7395103408140837e-05,
+      "loss": 0.4465,
+      "step": 8384
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7394432448819874e-05,
+      "loss": 0.363,
+      "step": 8385
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7393761416042078e-05,
+      "loss": 0.4517,
+      "step": 8386
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739309030981412e-05,
+      "loss": 0.4134,
+      "step": 8387
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7392419130142663e-05,
+      "loss": 0.3866,
+      "step": 8388
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739174787703438e-05,
+      "loss": 0.3728,
+      "step": 8389
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739107655049594e-05,
+      "loss": 0.4168,
+      "step": 8390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7390405150534007e-05,
+      "loss": 0.4245,
+      "step": 8391
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7389733677155252e-05,
+      "loss": 0.4731,
+      "step": 8392
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7389062130366348e-05,
+      "loss": 0.4047,
+      "step": 8393
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738839051017397e-05,
+      "loss": 0.4315,
+      "step": 8394
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7387718816584775e-05,
+      "loss": 0.3832,
+      "step": 8395
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738704704960545e-05,
+      "loss": 0.4052,
+      "step": 8396
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7386375209242666e-05,
+      "loss": 0.3894,
+      "step": 8397
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738570329550309e-05,
+      "loss": 0.3574,
+      "step": 8398
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7385031308393405e-05,
+      "loss": 0.5308,
+      "step": 8399
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7384359247920283e-05,
+      "loss": 0.3853,
+      "step": 8400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7383687114090397e-05,
+      "loss": 0.3865,
+      "step": 8401
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7383014906910432e-05,
+      "loss": 0.3985,
+      "step": 8402
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7382342626387057e-05,
+      "loss": 0.4203,
+      "step": 8403
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738167027252696e-05,
+      "loss": 0.3548,
+      "step": 8404
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738099784533681e-05,
+      "loss": 0.3959,
+      "step": 8405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7380325344823296e-05,
+      "loss": 0.467,
+      "step": 8406
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7379652770993093e-05,
+      "loss": 0.3973,
+      "step": 8407
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7378980123852886e-05,
+      "loss": 0.3645,
+      "step": 8408
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737830740340935e-05,
+      "loss": 0.3712,
+      "step": 8409
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7377634609669173e-05,
+      "loss": 0.3922,
+      "step": 8410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7376961742639043e-05,
+      "loss": 0.4673,
+      "step": 8411
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737628880232564e-05,
+      "loss": 0.4056,
+      "step": 8412
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737561578873565e-05,
+      "loss": 0.3438,
+      "step": 8413
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737494270187575e-05,
+      "loss": 0.4384,
+      "step": 8414
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737426954175264e-05,
+      "loss": 0.4445,
+      "step": 8415
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7373596308372998e-05,
+      "loss": 0.3575,
+      "step": 8416
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737292300174352e-05,
+      "loss": 0.4267,
+      "step": 8417
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737224962187089e-05,
+      "loss": 0.4215,
+      "step": 8418
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737157616876179e-05,
+      "loss": 0.4937,
+      "step": 8419
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7370902642422925e-05,
+      "loss": 0.4532,
+      "step": 8420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7370229042860973e-05,
+      "loss": 0.3963,
+      "step": 8421
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7369555370082634e-05,
+      "loss": 0.3629,
+      "step": 8422
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7368881624094598e-05,
+      "loss": 0.3661,
+      "step": 8423
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7368207804903558e-05,
+      "loss": 0.3865,
+      "step": 8424
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7367533912516203e-05,
+      "loss": 0.3911,
+      "step": 8425
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7366859946939232e-05,
+      "loss": 0.3534,
+      "step": 8426
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7366185908179346e-05,
+      "loss": 0.4225,
+      "step": 8427
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736551179624323e-05,
+      "loss": 0.358,
+      "step": 8428
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7364837611137586e-05,
+      "loss": 0.387,
+      "step": 8429
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736416335286911e-05,
+      "loss": 0.3627,
+      "step": 8430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363489021444503e-05,
+      "loss": 0.3859,
+      "step": 8431
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736281461687046e-05,
+      "loss": 0.4056,
+      "step": 8432
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7362140139153683e-05,
+      "loss": 0.3318,
+      "step": 8433
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7361465588300873e-05,
+      "loss": 0.4623,
+      "step": 8434
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7360790964318727e-05,
+      "loss": 0.4455,
+      "step": 8435
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7360116267213952e-05,
+      "loss": 0.3878,
+      "step": 8436
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735944149699325e-05,
+      "loss": 0.4523,
+      "step": 8437
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7358766653663316e-05,
+      "loss": 0.3544,
+      "step": 8438
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7358091737230865e-05,
+      "loss": 0.3578,
+      "step": 8439
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7357416747702594e-05,
+      "loss": 0.4708,
+      "step": 8440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7356741685085213e-05,
+      "loss": 0.4333,
+      "step": 8441
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735606654938543e-05,
+      "loss": 0.441,
+      "step": 8442
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735539134060994e-05,
+      "loss": 0.3964,
+      "step": 8443
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7354716058765464e-05,
+      "loss": 0.382,
+      "step": 8444
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7354040703858707e-05,
+      "loss": 0.421,
+      "step": 8445
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735336527589637e-05,
+      "loss": 0.5505,
+      "step": 8446
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7352689774885175e-05,
+      "loss": 0.3647,
+      "step": 8447
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735201420083182e-05,
+      "loss": 0.4529,
+      "step": 8448
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735133855374303e-05,
+      "loss": 0.4654,
+      "step": 8449
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7350662833625502e-05,
+      "loss": 0.39,
+      "step": 8450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7349987040485962e-05,
+      "loss": 0.4007,
+      "step": 8451
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7349311174331114e-05,
+      "loss": 0.324,
+      "step": 8452
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7348635235167677e-05,
+      "loss": 0.4047,
+      "step": 8453
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7347959223002364e-05,
+      "loss": 0.4262,
+      "step": 8454
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734728313784189e-05,
+      "loss": 0.3479,
+      "step": 8455
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7346606979692975e-05,
+      "loss": 0.3978,
+      "step": 8456
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7345930748562333e-05,
+      "loss": 0.4942,
+      "step": 8457
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7345254444456682e-05,
+      "loss": 0.416,
+      "step": 8458
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7344578067382734e-05,
+      "loss": 0.3805,
+      "step": 8459
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7343901617347222e-05,
+      "loss": 0.3681,
+      "step": 8460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7343225094356857e-05,
+      "loss": 0.5211,
+      "step": 8461
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734254849841836e-05,
+      "loss": 0.3899,
+      "step": 8462
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7341871829538456e-05,
+      "loss": 0.378,
+      "step": 8463
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734119508772386e-05,
+      "loss": 0.3143,
+      "step": 8464
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7340518272981304e-05,
+      "loss": 0.4083,
+      "step": 8465
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7339841385317505e-05,
+      "loss": 0.3421,
+      "step": 8466
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733916442473919e-05,
+      "loss": 0.4099,
+      "step": 8467
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733848739125308e-05,
+      "loss": 0.3935,
+      "step": 8468
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7337810284865907e-05,
+      "loss": 0.4444,
+      "step": 8469
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7337133105584392e-05,
+      "loss": 0.4313,
+      "step": 8470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7336455853415265e-05,
+      "loss": 0.3528,
+      "step": 8471
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7335778528365255e-05,
+      "loss": 0.4481,
+      "step": 8472
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7335101130441087e-05,
+      "loss": 0.5039,
+      "step": 8473
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7334423659649494e-05,
+      "loss": 0.3574,
+      "step": 8474
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7333746115997204e-05,
+      "loss": 0.5066,
+      "step": 8475
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733306849949095e-05,
+      "loss": 0.4562,
+      "step": 8476
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7332390810137455e-05,
+      "loss": 0.3758,
+      "step": 8477
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7331713047943465e-05,
+      "loss": 0.462,
+      "step": 8478
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73310352129157e-05,
+      "loss": 0.3683,
+      "step": 8479
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73303573050609e-05,
+      "loss": 0.3848,
+      "step": 8480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7329679324385803e-05,
+      "loss": 0.4453,
+      "step": 8481
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7329001270897138e-05,
+      "loss": 0.4046,
+      "step": 8482
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732832314460164e-05,
+      "loss": 0.3378,
+      "step": 8483
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732764494550605e-05,
+      "loss": 0.463,
+      "step": 8484
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7326966673617102e-05,
+      "loss": 0.4137,
+      "step": 8485
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732628832894154e-05,
+      "loss": 0.4279,
+      "step": 8486
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7325609911486093e-05,
+      "loss": 0.3607,
+      "step": 8487
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732493142125751e-05,
+      "loss": 0.3209,
+      "step": 8488
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7324252858262522e-05,
+      "loss": 0.3935,
+      "step": 8489
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7323574222507878e-05,
+      "loss": 0.4894,
+      "step": 8490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7322895514000317e-05,
+      "loss": 0.3694,
+      "step": 8491
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7322216732746578e-05,
+      "loss": 0.4324,
+      "step": 8492
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732153787875341e-05,
+      "loss": 0.3751,
+      "step": 8493
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7320858952027554e-05,
+      "loss": 0.3476,
+      "step": 8494
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732017995257575e-05,
+      "loss": 0.3851,
+      "step": 8495
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731950088040475e-05,
+      "loss": 0.4467,
+      "step": 8496
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7318821735521295e-05,
+      "loss": 0.4014,
+      "step": 8497
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7318142517932134e-05,
+      "loss": 0.4709,
+      "step": 8498
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7317463227644018e-05,
+      "loss": 0.3952,
+      "step": 8499
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7316783864663687e-05,
+      "loss": 0.4624,
+      "step": 8500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7316104428997898e-05,
+      "loss": 0.3899,
+      "step": 8501
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7315424920653394e-05,
+      "loss": 0.4199,
+      "step": 8502
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731474533963693e-05,
+      "loss": 0.4166,
+      "step": 8503
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731406568595525e-05,
+      "loss": 0.3528,
+      "step": 8504
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731338595961512e-05,
+      "loss": 0.3615,
+      "step": 8505
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731270616062328e-05,
+      "loss": 0.3978,
+      "step": 8506
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731202628898648e-05,
+      "loss": 0.3729,
+      "step": 8507
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7311346344711487e-05,
+      "loss": 0.4303,
+      "step": 8508
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7310666327805048e-05,
+      "loss": 0.3968,
+      "step": 8509
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7309986238273916e-05,
+      "loss": 0.3705,
+      "step": 8510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7309306076124854e-05,
+      "loss": 0.3754,
+      "step": 8511
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7308625841364614e-05,
+      "loss": 0.4394,
+      "step": 8512
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7307945533999955e-05,
+      "loss": 0.3681,
+      "step": 8513
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7307265154037634e-05,
+      "loss": 0.3654,
+      "step": 8514
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7306584701484414e-05,
+      "loss": 0.431,
+      "step": 8515
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7305904176347046e-05,
+      "loss": 0.4227,
+      "step": 8516
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.73052235786323e-05,
+      "loss": 0.5256,
+      "step": 8517
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730454290834693e-05,
+      "loss": 0.4134,
+      "step": 8518
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7303862165497703e-05,
+      "loss": 0.3754,
+      "step": 8519
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7303181350091377e-05,
+      "loss": 0.3738,
+      "step": 8520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730250046213472e-05,
+      "loss": 0.378,
+      "step": 8521
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7301819501634493e-05,
+      "loss": 0.385,
+      "step": 8522
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730113846859746e-05,
+      "loss": 0.3753,
+      "step": 8523
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7300457363030393e-05,
+      "loss": 0.3972,
+      "step": 8524
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729977618494005e-05,
+      "loss": 0.3295,
+      "step": 8525
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.72990949343332e-05,
+      "loss": 0.3984,
+      "step": 8526
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729841361121661e-05,
+      "loss": 0.4317,
+      "step": 8527
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729773221559705e-05,
+      "loss": 0.3891,
+      "step": 8528
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7297050747481288e-05,
+      "loss": 0.4094,
+      "step": 8529
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7296369206876096e-05,
+      "loss": 0.418,
+      "step": 8530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7295687593788245e-05,
+      "loss": 0.3704,
+      "step": 8531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.72950059082245e-05,
+      "loss": 0.4069,
+      "step": 8532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729432415019164e-05,
+      "loss": 0.3856,
+      "step": 8533
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7293642319696436e-05,
+      "loss": 0.4207,
+      "step": 8534
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7292960416745657e-05,
+      "loss": 0.4163,
+      "step": 8535
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729227844134608e-05,
+      "loss": 0.435,
+      "step": 8536
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729159639350448e-05,
+      "loss": 0.4909,
+      "step": 8537
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7290914273227634e-05,
+      "loss": 0.3692,
+      "step": 8538
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7290232080522316e-05,
+      "loss": 0.4,
+      "step": 8539
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7289549815395306e-05,
+      "loss": 0.4036,
+      "step": 8540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7288867477853376e-05,
+      "loss": 0.3188,
+      "step": 8541
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7288185067903308e-05,
+      "loss": 0.4258,
+      "step": 8542
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728750258555188e-05,
+      "loss": 0.3691,
+      "step": 8543
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7286820030805877e-05,
+      "loss": 0.3656,
+      "step": 8544
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728613740367207e-05,
+      "loss": 0.3435,
+      "step": 8545
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7285454704157246e-05,
+      "loss": 0.3268,
+      "step": 8546
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728477193226819e-05,
+      "loss": 0.411,
+      "step": 8547
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7284089088011676e-05,
+      "loss": 0.4313,
+      "step": 8548
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7283406171394495e-05,
+      "loss": 0.3363,
+      "step": 8549
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7282723182423432e-05,
+      "loss": 0.4479,
+      "step": 8550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7282040121105265e-05,
+      "loss": 0.424,
+      "step": 8551
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7281356987446785e-05,
+      "loss": 0.3783,
+      "step": 8552
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7280673781454774e-05,
+      "loss": 0.4198,
+      "step": 8553
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7279990503136022e-05,
+      "loss": 0.3867,
+      "step": 8554
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7279307152497316e-05,
+      "loss": 0.4399,
+      "step": 8555
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7278623729545446e-05,
+      "loss": 0.3891,
+      "step": 8556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72779402342872e-05,
+      "loss": 0.3888,
+      "step": 8557
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7277256666729368e-05,
+      "loss": 0.3767,
+      "step": 8558
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7276573026878743e-05,
+      "loss": 0.3876,
+      "step": 8559
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7275889314742107e-05,
+      "loss": 0.3934,
+      "step": 8560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727520553032626e-05,
+      "loss": 0.402,
+      "step": 8561
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7274521673637998e-05,
+      "loss": 0.3779,
+      "step": 8562
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7273837744684105e-05,
+      "loss": 0.3407,
+      "step": 8563
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7273153743471383e-05,
+      "loss": 0.4634,
+      "step": 8564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7272469670006625e-05,
+      "loss": 0.4072,
+      "step": 8565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727178552429662e-05,
+      "loss": 0.3762,
+      "step": 8566
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7271101306348175e-05,
+      "loss": 0.3925,
+      "step": 8567
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7270417016168083e-05,
+      "loss": 0.4566,
+      "step": 8568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726973265376314e-05,
+      "loss": 0.5082,
+      "step": 8569
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7269048219140144e-05,
+      "loss": 0.4393,
+      "step": 8570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7268363712305897e-05,
+      "loss": 0.3907,
+      "step": 8571
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7267679133267193e-05,
+      "loss": 0.378,
+      "step": 8572
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726699448203084e-05,
+      "loss": 0.4033,
+      "step": 8573
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7266309758603638e-05,
+      "loss": 0.483,
+      "step": 8574
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726562496299239e-05,
+      "loss": 0.3315,
+      "step": 8575
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264940095203892e-05,
+      "loss": 0.4565,
+      "step": 8576
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264255155244954e-05,
+      "loss": 0.4593,
+      "step": 8577
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7263570143122378e-05,
+      "loss": 0.3442,
+      "step": 8578
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7262885058842968e-05,
+      "loss": 0.3699,
+      "step": 8579
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7262199902413533e-05,
+      "loss": 0.5,
+      "step": 8580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7261514673840875e-05,
+      "loss": 0.4008,
+      "step": 8581
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7260829373131808e-05,
+      "loss": 0.4501,
+      "step": 8582
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7260144000293132e-05,
+      "loss": 0.4271,
+      "step": 8583
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7259458555331658e-05,
+      "loss": 0.3846,
+      "step": 8584
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7258773038254194e-05,
+      "loss": 0.502,
+      "step": 8585
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7258087449067556e-05,
+      "loss": 0.334,
+      "step": 8586
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725740178777855e-05,
+      "loss": 0.4039,
+      "step": 8587
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7256716054393987e-05,
+      "loss": 0.3945,
+      "step": 8588
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725603024892068e-05,
+      "loss": 0.4068,
+      "step": 8589
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7255344371365442e-05,
+      "loss": 0.3223,
+      "step": 8590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725465842173509e-05,
+      "loss": 0.3731,
+      "step": 8591
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725397240003643e-05,
+      "loss": 0.4732,
+      "step": 8592
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7253286306276283e-05,
+      "loss": 0.351,
+      "step": 8593
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7252600140461466e-05,
+      "loss": 0.39,
+      "step": 8594
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7251913902598795e-05,
+      "loss": 0.4224,
+      "step": 8595
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725122759269508e-05,
+      "loss": 0.3576,
+      "step": 8596
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7250541210757145e-05,
+      "loss": 0.5055,
+      "step": 8597
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724985475679181e-05,
+      "loss": 0.3766,
+      "step": 8598
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724916823080589e-05,
+      "loss": 0.3656,
+      "step": 8599
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724848163280621e-05,
+      "loss": 0.4114,
+      "step": 8600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7247794962799584e-05,
+      "loss": 0.4247,
+      "step": 8601
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7247108220792837e-05,
+      "loss": 0.3683,
+      "step": 8602
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7246421406792793e-05,
+      "loss": 0.3862,
+      "step": 8603
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724573452080627e-05,
+      "loss": 0.3642,
+      "step": 8604
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72450475628401e-05,
+      "loss": 0.3743,
+      "step": 8605
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7244360532901097e-05,
+      "loss": 0.4262,
+      "step": 8606
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724367343099609e-05,
+      "loss": 0.4216,
+      "step": 8607
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724298625713191e-05,
+      "loss": 0.4122,
+      "step": 8608
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7242299011315377e-05,
+      "loss": 0.462,
+      "step": 8609
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7241611693553323e-05,
+      "loss": 0.4104,
+      "step": 8610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724092430385257e-05,
+      "loss": 0.4152,
+      "step": 8611
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.724023684221995e-05,
+      "loss": 0.4295,
+      "step": 8612
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723954930866229e-05,
+      "loss": 0.3906,
+      "step": 8613
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238861703186426e-05,
+      "loss": 0.3943,
+      "step": 8614
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238174025799183e-05,
+      "loss": 0.423,
+      "step": 8615
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723748627650739e-05,
+      "loss": 0.4021,
+      "step": 8616
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723679845531789e-05,
+      "loss": 0.3782,
+      "step": 8617
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7236110562237506e-05,
+      "loss": 0.4027,
+      "step": 8618
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7235422597273077e-05,
+      "loss": 0.4262,
+      "step": 8619
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723473456043143e-05,
+      "loss": 0.4042,
+      "step": 8620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723404645171941e-05,
+      "loss": 0.3661,
+      "step": 8621
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7233358271143844e-05,
+      "loss": 0.4109,
+      "step": 8622
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723267001871157e-05,
+      "loss": 0.4109,
+      "step": 8623
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7231981694429433e-05,
+      "loss": 0.3465,
+      "step": 8624
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7231293298304265e-05,
+      "loss": 0.3633,
+      "step": 8625
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7230604830342903e-05,
+      "loss": 0.4629,
+      "step": 8626
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7229916290552185e-05,
+      "loss": 0.4015,
+      "step": 8627
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722922767893896e-05,
+      "loss": 0.2961,
+      "step": 8628
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7228538995510057e-05,
+      "loss": 0.4081,
+      "step": 8629
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722785024027233e-05,
+      "loss": 0.4212,
+      "step": 8630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7227161413232606e-05,
+      "loss": 0.4239,
+      "step": 8631
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7226472514397744e-05,
+      "loss": 0.4902,
+      "step": 8632
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7225783543774573e-05,
+      "loss": 0.4064,
+      "step": 8633
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7225094501369947e-05,
+      "loss": 0.3565,
+      "step": 8634
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722440538719071e-05,
+      "loss": 0.4395,
+      "step": 8635
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7223716201243705e-05,
+      "loss": 0.3863,
+      "step": 8636
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7223026943535776e-05,
+      "loss": 0.4091,
+      "step": 8637
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7222337614073775e-05,
+      "loss": 0.4157,
+      "step": 8638
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722164821286455e-05,
+      "loss": 0.4168,
+      "step": 8639
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7220958739914947e-05,
+      "loss": 0.372,
+      "step": 8640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7220269195231815e-05,
+      "loss": 0.3377,
+      "step": 8641
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7219579578822007e-05,
+      "loss": 0.4316,
+      "step": 8642
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7218889890692368e-05,
+      "loss": 0.4494,
+      "step": 8643
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7218200130849755e-05,
+      "loss": 0.3713,
+      "step": 8644
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721751029930102e-05,
+      "loss": 0.4225,
+      "step": 8645
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7216820396053008e-05,
+      "loss": 0.3999,
+      "step": 8646
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7216130421112585e-05,
+      "loss": 0.4278,
+      "step": 8647
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7215440374486596e-05,
+      "loss": 0.4342,
+      "step": 8648
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.72147502561819e-05,
+      "loss": 0.329,
+      "step": 8649
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721406006620535e-05,
+      "loss": 0.3821,
+      "step": 8650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7213369804563805e-05,
+      "loss": 0.4552,
+      "step": 8651
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7212679471264122e-05,
+      "loss": 0.4473,
+      "step": 8652
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7211989066313156e-05,
+      "loss": 0.4248,
+      "step": 8653
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7211298589717772e-05,
+      "loss": 0.385,
+      "step": 8654
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721060804148482e-05,
+      "loss": 0.4072,
+      "step": 8655
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7209917421621165e-05,
+      "loss": 0.3751,
+      "step": 8656
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7209226730133668e-05,
+      "loss": 0.396,
+      "step": 8657
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720853596702919e-05,
+      "loss": 0.4592,
+      "step": 8658
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720784513231459e-05,
+      "loss": 0.3959,
+      "step": 8659
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720715422599674e-05,
+      "loss": 0.4329,
+      "step": 8660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720646324808249e-05,
+      "loss": 0.4652,
+      "step": 8661
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7205772198578716e-05,
+      "loss": 0.3548,
+      "step": 8662
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7205081077492277e-05,
+      "loss": 0.4278,
+      "step": 8663
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7204389884830038e-05,
+      "loss": 0.4451,
+      "step": 8664
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720369862059887e-05,
+      "loss": 0.3693,
+      "step": 8665
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720300728480563e-05,
+      "loss": 0.4002,
+      "step": 8666
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.72023158774572e-05,
+      "loss": 0.4506,
+      "step": 8667
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720162439856044e-05,
+      "loss": 0.3896,
+      "step": 8668
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7200932848122222e-05,
+      "loss": 0.4436,
+      "step": 8669
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7200241226149415e-05,
+      "loss": 0.3612,
+      "step": 8670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7199549532648887e-05,
+      "loss": 0.4363,
+      "step": 8671
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7198857767627513e-05,
+      "loss": 0.4122,
+      "step": 8672
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7198165931092163e-05,
+      "loss": 0.4354,
+      "step": 8673
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719747402304971e-05,
+      "loss": 0.4264,
+      "step": 8674
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719678204350703e-05,
+      "loss": 0.3533,
+      "step": 8675
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719608999247099e-05,
+      "loss": 0.4644,
+      "step": 8676
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7195397869948475e-05,
+      "loss": 0.4387,
+      "step": 8677
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194705675946353e-05,
+      "loss": 0.4496,
+      "step": 8678
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194013410471506e-05,
+      "loss": 0.3514,
+      "step": 8679
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7193321073530806e-05,
+      "loss": 0.4035,
+      "step": 8680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719262866513113e-05,
+      "loss": 0.3415,
+      "step": 8681
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7191936185279362e-05,
+      "loss": 0.4413,
+      "step": 8682
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719124363398238e-05,
+      "loss": 0.3871,
+      "step": 8683
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7190551011247056e-05,
+      "loss": 0.353,
+      "step": 8684
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718985831708028e-05,
+      "loss": 0.4052,
+      "step": 8685
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718916555148893e-05,
+      "loss": 0.4821,
+      "step": 8686
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718847271447989e-05,
+      "loss": 0.4074,
+      "step": 8687
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7187779806060037e-05,
+      "loss": 0.4686,
+      "step": 8688
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7187086826236257e-05,
+      "loss": 0.3348,
+      "step": 8689
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718639377501544e-05,
+      "loss": 0.3519,
+      "step": 8690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718570065240446e-05,
+      "loss": 0.3688,
+      "step": 8691
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718500745841021e-05,
+      "loss": 0.3926,
+      "step": 8692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718431419303958e-05,
+      "loss": 0.4775,
+      "step": 8693
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7183620856299448e-05,
+      "loss": 0.5104,
+      "step": 8694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7182927448196705e-05,
+      "loss": 0.3529,
+      "step": 8695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718223396873824e-05,
+      "loss": 0.3705,
+      "step": 8696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718154041793094e-05,
+      "loss": 0.4782,
+      "step": 8697
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180846795781697e-05,
+      "loss": 0.4354,
+      "step": 8698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180153102297402e-05,
+      "loss": 0.4106,
+      "step": 8699
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7179459337484945e-05,
+      "loss": 0.4127,
+      "step": 8700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7178765501351215e-05,
+      "loss": 0.3245,
+      "step": 8701
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7178071593903116e-05,
+      "loss": 0.4419,
+      "step": 8702
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7177377615147523e-05,
+      "loss": 0.3648,
+      "step": 8703
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7176683565091344e-05,
+      "loss": 0.4104,
+      "step": 8704
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7175989443741472e-05,
+      "loss": 0.4287,
+      "step": 8705
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71752952511048e-05,
+      "loss": 0.3324,
+      "step": 8706
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7174600987188223e-05,
+      "loss": 0.3501,
+      "step": 8707
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717390665199864e-05,
+      "loss": 0.4034,
+      "step": 8708
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7173212245542945e-05,
+      "loss": 0.3933,
+      "step": 8709
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717251776782804e-05,
+      "loss": 0.4624,
+      "step": 8710
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7171823218860825e-05,
+      "loss": 0.4198,
+      "step": 8711
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71711285986482e-05,
+      "loss": 0.3595,
+      "step": 8712
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7170433907197058e-05,
+      "loss": 0.4361,
+      "step": 8713
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716973914451431e-05,
+      "loss": 0.4262,
+      "step": 8714
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716904431060685e-05,
+      "loss": 0.5079,
+      "step": 8715
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7168349405481587e-05,
+      "loss": 0.487,
+      "step": 8716
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7167654429145417e-05,
+      "loss": 0.3709,
+      "step": 8717
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716695938160525e-05,
+      "loss": 0.3642,
+      "step": 8718
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7166264262867988e-05,
+      "loss": 0.3883,
+      "step": 8719
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7165569072940537e-05,
+      "loss": 0.3615,
+      "step": 8720
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7164873811829802e-05,
+      "loss": 0.3964,
+      "step": 8721
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7164178479542693e-05,
+      "loss": 0.4219,
+      "step": 8722
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7163483076086113e-05,
+      "loss": 0.3714,
+      "step": 8723
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716278760146697e-05,
+      "loss": 0.3495,
+      "step": 8724
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716209205569218e-05,
+      "loss": 0.383,
+      "step": 8725
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716139643876865e-05,
+      "loss": 0.3549,
+      "step": 8726
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716070075070328e-05,
+      "loss": 0.3264,
+      "step": 8727
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7160004991502998e-05,
+      "loss": 0.3588,
+      "step": 8728
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71593091611747e-05,
+      "loss": 0.4633,
+      "step": 8729
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715861325972531e-05,
+      "loss": 0.4234,
+      "step": 8730
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7157917287161737e-05,
+      "loss": 0.4469,
+      "step": 8731
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715722124349089e-05,
+      "loss": 0.3526,
+      "step": 8732
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715652512871969e-05,
+      "loss": 0.3498,
+      "step": 8733
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715582894285505e-05,
+      "loss": 0.4144,
+      "step": 8734
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7155132685903888e-05,
+      "loss": 0.3867,
+      "step": 8735
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7154436357873116e-05,
+      "loss": 0.4539,
+      "step": 8736
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7153739958769656e-05,
+      "loss": 0.4552,
+      "step": 8737
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7153043488600423e-05,
+      "loss": 0.3423,
+      "step": 8738
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715234694737234e-05,
+      "loss": 0.4478,
+      "step": 8739
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715165033509232e-05,
+      "loss": 0.3826,
+      "step": 8740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715095365176729e-05,
+      "loss": 0.4695,
+      "step": 8741
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7150256897404164e-05,
+      "loss": 0.4121,
+      "step": 8742
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7149560072009865e-05,
+      "loss": 0.4418,
+      "step": 8743
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148863175591322e-05,
+      "loss": 0.4758,
+      "step": 8744
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148166208155453e-05,
+      "loss": 0.4506,
+      "step": 8745
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7147469169709183e-05,
+      "loss": 0.4215,
+      "step": 8746
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7146772060259435e-05,
+      "loss": 0.3891,
+      "step": 8747
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7146074879813136e-05,
+      "loss": 0.426,
+      "step": 8748
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7145377628377207e-05,
+      "loss": 0.3992,
+      "step": 8749
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714468030595858e-05,
+      "loss": 0.4085,
+      "step": 8750
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7143982912564184e-05,
+      "loss": 0.4807,
+      "step": 8751
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714328544820094e-05,
+      "loss": 0.4076,
+      "step": 8752
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7142587912875778e-05,
+      "loss": 0.4273,
+      "step": 8753
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7141890306595636e-05,
+      "loss": 0.3948,
+      "step": 8754
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714119262936743e-05,
+      "loss": 0.3986,
+      "step": 8755
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7140494881198106e-05,
+      "loss": 0.3859,
+      "step": 8756
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7139797062094583e-05,
+      "loss": 0.4524,
+      "step": 8757
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7139099172063797e-05,
+      "loss": 0.3786,
+      "step": 8758
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7138401211112684e-05,
+      "loss": 0.4261,
+      "step": 8759
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7137703179248176e-05,
+      "loss": 0.4065,
+      "step": 8760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7137005076477208e-05,
+      "loss": 0.4346,
+      "step": 8761
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713630690280671e-05,
+      "loss": 0.3951,
+      "step": 8762
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7135608658243623e-05,
+      "loss": 0.446,
+      "step": 8763
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7134910342794882e-05,
+      "loss": 0.455,
+      "step": 8764
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7134211956467425e-05,
+      "loss": 0.3532,
+      "step": 8765
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7133513499268185e-05,
+      "loss": 0.3695,
+      "step": 8766
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713281497120411e-05,
+      "loss": 0.3975,
+      "step": 8767
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713211637228213e-05,
+      "loss": 0.4265,
+      "step": 8768
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713141770250919e-05,
+      "loss": 0.5021,
+      "step": 8769
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713071896189223e-05,
+      "loss": 0.383,
+      "step": 8770
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7130020150438186e-05,
+      "loss": 0.4438,
+      "step": 8771
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712932126815401e-05,
+      "loss": 0.3973,
+      "step": 8772
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7128622315046635e-05,
+      "loss": 0.3341,
+      "step": 8773
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7127923291123014e-05,
+      "loss": 0.3216,
+      "step": 8774
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712722419639008e-05,
+      "loss": 0.359,
+      "step": 8775
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7126525030854786e-05,
+      "loss": 0.3663,
+      "step": 8776
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7125825794524077e-05,
+      "loss": 0.4315,
+      "step": 8777
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7125126487404893e-05,
+      "loss": 0.3776,
+      "step": 8778
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712442710950419e-05,
+      "loss": 0.3365,
+      "step": 8779
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7123727660828912e-05,
+      "loss": 0.365,
+      "step": 8780
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7123028141386003e-05,
+      "loss": 0.3394,
+      "step": 8781
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7122328551182418e-05,
+      "loss": 0.3326,
+      "step": 8782
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7121628890225104e-05,
+      "loss": 0.4775,
+      "step": 8783
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712092915852101e-05,
+      "loss": 0.3902,
+      "step": 8784
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7120229356077095e-05,
+      "loss": 0.3962,
+      "step": 8785
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.71195294829003e-05,
+      "loss": 0.4557,
+      "step": 8786
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711882953899758e-05,
+      "loss": 0.3541,
+      "step": 8787
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7118129524375896e-05,
+      "loss": 0.3722,
+      "step": 8788
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7117429439042194e-05,
+      "loss": 0.3542,
+      "step": 8789
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7116729283003433e-05,
+      "loss": 0.4444,
+      "step": 8790
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711602905626657e-05,
+      "loss": 0.3546,
+      "step": 8791
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711532875883855e-05,
+      "loss": 0.4128,
+      "step": 8792
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7114628390726345e-05,
+      "loss": 0.3945,
+      "step": 8793
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7113927951936903e-05,
+      "loss": 0.467,
+      "step": 8794
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7113227442477182e-05,
+      "loss": 0.3878,
+      "step": 8795
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7112526862354147e-05,
+      "loss": 0.4012,
+      "step": 8796
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7111826211574755e-05,
+      "loss": 0.3424,
+      "step": 8797
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7111125490145964e-05,
+      "loss": 0.4641,
+      "step": 8798
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7110424698074733e-05,
+      "loss": 0.3934,
+      "step": 8799
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7109723835368032e-05,
+      "loss": 0.4357,
+      "step": 8800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7109022902032818e-05,
+      "loss": 0.4607,
+      "step": 8801
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710832189807605e-05,
+      "loss": 0.4235,
+      "step": 8802
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7107620823504706e-05,
+      "loss": 0.451,
+      "step": 8803
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710691967832573e-05,
+      "loss": 0.4156,
+      "step": 8804
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7106218462546107e-05,
+      "loss": 0.4345,
+      "step": 8805
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710551717617279e-05,
+      "loss": 0.3903,
+      "step": 8806
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710481581921275e-05,
+      "loss": 0.4068,
+      "step": 8807
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7104114391672957e-05,
+      "loss": 0.4468,
+      "step": 8808
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7103412893560375e-05,
+      "loss": 0.3831,
+      "step": 8809
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7102711324881972e-05,
+      "loss": 0.4023,
+      "step": 8810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7102009685644722e-05,
+      "loss": 0.4269,
+      "step": 8811
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710130797585559e-05,
+      "loss": 0.4269,
+      "step": 8812
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710060619552155e-05,
+      "loss": 0.4242,
+      "step": 8813
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7099904344649574e-05,
+      "loss": 0.4857,
+      "step": 8814
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7099202423246632e-05,
+      "loss": 0.3701,
+      "step": 8815
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70985004313197e-05,
+      "loss": 0.3737,
+      "step": 8816
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7097798368875746e-05,
+      "loss": 0.441,
+      "step": 8817
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709709623592175e-05,
+      "loss": 0.4055,
+      "step": 8818
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7096394032464684e-05,
+      "loss": 0.4261,
+      "step": 8819
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709569175851153e-05,
+      "loss": 0.4192,
+      "step": 8820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7094989414069256e-05,
+      "loss": 0.4112,
+      "step": 8821
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7094286999144843e-05,
+      "loss": 0.482,
+      "step": 8822
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7093584513745264e-05,
+      "loss": 0.3763,
+      "step": 8823
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7092881957877507e-05,
+      "loss": 0.4286,
+      "step": 8824
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7092179331548545e-05,
+      "loss": 0.4163,
+      "step": 8825
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7091476634765357e-05,
+      "loss": 0.3519,
+      "step": 8826
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7090773867534928e-05,
+      "loss": 0.3894,
+      "step": 8827
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7090071029864232e-05,
+      "loss": 0.3776,
+      "step": 8828
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7089368121760263e-05,
+      "loss": 0.4222,
+      "step": 8829
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708866514322999e-05,
+      "loss": 0.4253,
+      "step": 8830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7087962094280407e-05,
+      "loss": 0.323,
+      "step": 8831
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7087258974918497e-05,
+      "loss": 0.3414,
+      "step": 8832
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7086555785151238e-05,
+      "loss": 0.437,
+      "step": 8833
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708585252498562e-05,
+      "loss": 0.3965,
+      "step": 8834
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7085149194428628e-05,
+      "loss": 0.336,
+      "step": 8835
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7084445793487253e-05,
+      "loss": 0.4419,
+      "step": 8836
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708374232216848e-05,
+      "loss": 0.4216,
+      "step": 8837
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7083038780479295e-05,
+      "loss": 0.4529,
+      "step": 8838
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7082335168426687e-05,
+      "loss": 0.4151,
+      "step": 8839
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708163148601765e-05,
+      "loss": 0.4018,
+      "step": 8840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7080927733259168e-05,
+      "loss": 0.3614,
+      "step": 8841
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7080223910158246e-05,
+      "loss": 0.3658,
+      "step": 8842
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7079520016721856e-05,
+      "loss": 0.3627,
+      "step": 8843
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7078816052957005e-05,
+      "loss": 0.3751,
+      "step": 8844
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707811201887068e-05,
+      "loss": 0.3547,
+      "step": 8845
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7077407914469878e-05,
+      "loss": 0.4247,
+      "step": 8846
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707670373976159e-05,
+      "loss": 0.3895,
+      "step": 8847
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7075999494752816e-05,
+      "loss": 0.4168,
+      "step": 8848
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7075295179450552e-05,
+      "loss": 0.3614,
+      "step": 8849
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707459079386179e-05,
+      "loss": 0.4169,
+      "step": 8850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707388633799353e-05,
+      "loss": 0.4237,
+      "step": 8851
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7073181811852768e-05,
+      "loss": 0.3382,
+      "step": 8852
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7072477215446506e-05,
+      "loss": 0.3683,
+      "step": 8853
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707177254878174e-05,
+      "loss": 0.4507,
+      "step": 8854
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.3905,
+      "step": 8855
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707036300470471e-05,
+      "loss": 0.3839,
+      "step": 8856
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7069658127306445e-05,
+      "loss": 0.4104,
+      "step": 8857
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7068953179677684e-05,
+      "loss": 0.4057,
+      "step": 8858
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706824816182543e-05,
+      "loss": 0.4275,
+      "step": 8859
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706754307375668e-05,
+      "loss": 0.3864,
+      "step": 8860
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7066837915478453e-05,
+      "loss": 0.3203,
+      "step": 8861
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706613268699774e-05,
+      "loss": 0.307,
+      "step": 8862
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7065427388321555e-05,
+      "loss": 0.3966,
+      "step": 8863
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70647220194569e-05,
+      "loss": 0.3901,
+      "step": 8864
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7064016580410782e-05,
+      "loss": 0.4135,
+      "step": 8865
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7063311071190215e-05,
+      "loss": 0.384,
+      "step": 8866
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7062605491802198e-05,
+      "loss": 0.3627,
+      "step": 8867
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7061899842253754e-05,
+      "loss": 0.3837,
+      "step": 8868
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7061194122551877e-05,
+      "loss": 0.4016,
+      "step": 8869
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7060488332703585e-05,
+      "loss": 0.3754,
+      "step": 8870
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7059782472715894e-05,
+      "loss": 0.3925,
+      "step": 8871
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7059076542595813e-05,
+      "loss": 0.4179,
+      "step": 8872
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7058370542350348e-05,
+      "loss": 0.4778,
+      "step": 8873
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7057664471986524e-05,
+      "loss": 0.4219,
+      "step": 8874
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7056958331511343e-05,
+      "loss": 0.3774,
+      "step": 8875
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7056252120931832e-05,
+      "loss": 0.3728,
+      "step": 8876
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7055545840254996e-05,
+      "loss": 0.4098,
+      "step": 8877
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7054839489487856e-05,
+      "loss": 0.4213,
+      "step": 8878
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705413306863743e-05,
+      "loss": 0.3986,
+      "step": 8879
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7053426577710734e-05,
+      "loss": 0.3385,
+      "step": 8880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7052720016714787e-05,
+      "loss": 0.437,
+      "step": 8881
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7052013385656606e-05,
+      "loss": 0.3784,
+      "step": 8882
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7051306684543216e-05,
+      "loss": 0.4289,
+      "step": 8883
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705059991338163e-05,
+      "loss": 0.4736,
+      "step": 8884
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7049893072178874e-05,
+      "loss": 0.3917,
+      "step": 8885
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704918616094197e-05,
+      "loss": 0.3847,
+      "step": 8886
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704847917967794e-05,
+      "loss": 0.4229,
+      "step": 8887
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7047772128393806e-05,
+      "loss": 0.3915,
+      "step": 8888
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704706500709659e-05,
+      "loss": 0.4137,
+      "step": 8889
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7046357815793317e-05,
+      "loss": 0.4377,
+      "step": 8890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704565055449102e-05,
+      "loss": 0.3647,
+      "step": 8891
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7044943223196716e-05,
+      "loss": 0.3609,
+      "step": 8892
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7044235821917435e-05,
+      "loss": 0.4052,
+      "step": 8893
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7043528350660206e-05,
+      "loss": 0.4481,
+      "step": 8894
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7042820809432057e-05,
+      "loss": 0.3443,
+      "step": 8895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704211319824001e-05,
+      "loss": 0.4496,
+      "step": 8896
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7041405517091103e-05,
+      "loss": 0.4635,
+      "step": 8897
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704069776599236e-05,
+      "loss": 0.3745,
+      "step": 8898
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703998994495082e-05,
+      "loss": 0.3646,
+      "step": 8899
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7039282053973504e-05,
+      "loss": 0.4079,
+      "step": 8900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703857409306745e-05,
+      "loss": 0.3748,
+      "step": 8901
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7037866062239688e-05,
+      "loss": 0.4632,
+      "step": 8902
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7037157961497262e-05,
+      "loss": 0.383,
+      "step": 8903
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703644979084719e-05,
+      "loss": 0.4537,
+      "step": 8904
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7035741550296522e-05,
+      "loss": 0.3636,
+      "step": 8905
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7035033239852282e-05,
+      "loss": 0.4182,
+      "step": 8906
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7034324859521512e-05,
+      "loss": 0.3908,
+      "step": 8907
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703361640931125e-05,
+      "loss": 0.4045,
+      "step": 8908
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7032907889228534e-05,
+      "loss": 0.4058,
+      "step": 8909
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70321992992804e-05,
+      "loss": 0.4132,
+      "step": 8910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7031490639473884e-05,
+      "loss": 0.4288,
+      "step": 8911
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030781909816033e-05,
+      "loss": 0.3775,
+      "step": 8912
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030073110313882e-05,
+      "loss": 0.3273,
+      "step": 8913
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7029364240974478e-05,
+      "loss": 0.4414,
+      "step": 8914
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7028655301804858e-05,
+      "loss": 0.3763,
+      "step": 8915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702794629281207e-05,
+      "loss": 0.3874,
+      "step": 8916
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702723721400315e-05,
+      "loss": 0.3637,
+      "step": 8917
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7026528065385143e-05,
+      "loss": 0.4509,
+      "step": 8918
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7025818846965098e-05,
+      "loss": 0.5256,
+      "step": 8919
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7025109558750063e-05,
+      "loss": 0.4454,
+      "step": 8920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7024400200747075e-05,
+      "loss": 0.3911,
+      "step": 8921
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702369077296319e-05,
+      "loss": 0.3648,
+      "step": 8922
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702298127540545e-05,
+      "loss": 0.3988,
+      "step": 8923
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70222717080809e-05,
+      "loss": 0.4414,
+      "step": 8924
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7021562070996597e-05,
+      "loss": 0.3982,
+      "step": 8925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7020852364159586e-05,
+      "loss": 0.399,
+      "step": 8926
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702014258757692e-05,
+      "loss": 0.4032,
+      "step": 8927
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701943274125565e-05,
+      "loss": 0.3193,
+      "step": 8928
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701872282520282e-05,
+      "loss": 0.4065,
+      "step": 8929
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7018012839425492e-05,
+      "loss": 0.3814,
+      "step": 8930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7017302783930717e-05,
+      "loss": 0.4654,
+      "step": 8931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7016592658725547e-05,
+      "loss": 0.3654,
+      "step": 8932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7015882463817032e-05,
+      "loss": 0.4096,
+      "step": 8933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7015172199212235e-05,
+      "loss": 0.4497,
+      "step": 8934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7014461864918205e-05,
+      "loss": 0.4514,
+      "step": 8935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7013751460942007e-05,
+      "loss": 0.4076,
+      "step": 8936
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701304098729069e-05,
+      "loss": 0.4014,
+      "step": 8937
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7012330443971318e-05,
+      "loss": 0.3854,
+      "step": 8938
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7011619830990943e-05,
+      "loss": 0.4754,
+      "step": 8939
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7010909148356626e-05,
+      "loss": 0.5041,
+      "step": 8940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7010198396075435e-05,
+      "loss": 0.4203,
+      "step": 8941
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700948757415442e-05,
+      "loss": 0.3568,
+      "step": 8942
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700877668260065e-05,
+      "loss": 0.4286,
+      "step": 8943
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700806572142118e-05,
+      "loss": 0.3513,
+      "step": 8944
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7007354690623082e-05,
+      "loss": 0.3941,
+      "step": 8945
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700664359021341e-05,
+      "loss": 0.3617,
+      "step": 8946
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7005932420199235e-05,
+      "loss": 0.4566,
+      "step": 8947
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7005221180587616e-05,
+      "loss": 0.3639,
+      "step": 8948
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7004509871385623e-05,
+      "loss": 0.4166,
+      "step": 8949
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700379849260032e-05,
+      "loss": 0.3803,
+      "step": 8950
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7003087044238777e-05,
+      "loss": 0.3334,
+      "step": 8951
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700237552630806e-05,
+      "loss": 0.4687,
+      "step": 8952
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700166393881523e-05,
+      "loss": 0.4134,
+      "step": 8953
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7000952281767367e-05,
+      "loss": 0.4389,
+      "step": 8954
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7000240555171536e-05,
+      "loss": 0.4361,
+      "step": 8955
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6999528759034808e-05,
+      "loss": 0.3609,
+      "step": 8956
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6998816893364255e-05,
+      "loss": 0.4529,
+      "step": 8957
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6998104958166943e-05,
+      "loss": 0.3315,
+      "step": 8958
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699739295344995e-05,
+      "loss": 0.3658,
+      "step": 8959
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699668087922035e-05,
+      "loss": 0.3875,
+      "step": 8960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6995968735485214e-05,
+      "loss": 0.433,
+      "step": 8961
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6995256522251614e-05,
+      "loss": 0.4584,
+      "step": 8962
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699454423952663e-05,
+      "loss": 0.491,
+      "step": 8963
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6993831887317337e-05,
+      "loss": 0.3913,
+      "step": 8964
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6993119465630806e-05,
+      "loss": 0.385,
+      "step": 8965
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6992406974474126e-05,
+      "loss": 0.4505,
+      "step": 8966
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6991694413854364e-05,
+      "loss": 0.3655,
+      "step": 8967
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6990981783778606e-05,
+      "loss": 0.3999,
+      "step": 8968
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699026908425392e-05,
+      "loss": 0.4223,
+      "step": 8969
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6989556315287398e-05,
+      "loss": 0.4056,
+      "step": 8970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698884347688612e-05,
+      "loss": 0.3813,
+      "step": 8971
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6988130569057157e-05,
+      "loss": 0.4485,
+      "step": 8972
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.69874175918076e-05,
+      "loss": 0.3911,
+      "step": 8973
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6986704545144535e-05,
+      "loss": 0.4471,
+      "step": 8974
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6985991429075038e-05,
+      "loss": 0.4099,
+      "step": 8975
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6985278243606192e-05,
+      "loss": 0.4589,
+      "step": 8976
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6984564988745088e-05,
+      "loss": 0.3402,
+      "step": 8977
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698385166449881e-05,
+      "loss": 0.4935,
+      "step": 8978
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698313827087444e-05,
+      "loss": 0.427,
+      "step": 8979
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6982424807879067e-05,
+      "loss": 0.4602,
+      "step": 8980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6981711275519777e-05,
+      "loss": 0.4022,
+      "step": 8981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6980997673803668e-05,
+      "loss": 0.4227,
+      "step": 8982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6980284002737817e-05,
+      "loss": 0.3779,
+      "step": 8983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697957026232932e-05,
+      "loss": 0.3871,
+      "step": 8984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6978856452585266e-05,
+      "loss": 0.4161,
+      "step": 8985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6978142573512745e-05,
+      "loss": 0.2892,
+      "step": 8986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6977428625118847e-05,
+      "loss": 0.418,
+      "step": 8987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697671460741067e-05,
+      "loss": 0.4241,
+      "step": 8988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6976000520395303e-05,
+      "loss": 0.3194,
+      "step": 8989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6975286364079838e-05,
+      "loss": 0.4278,
+      "step": 8990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6974572138471376e-05,
+      "loss": 0.3673,
+      "step": 8991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6973857843577008e-05,
+      "loss": 0.4719,
+      "step": 8992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697314347940383e-05,
+      "loss": 0.3742,
+      "step": 8993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6972429045958937e-05,
+      "loss": 0.4136,
+      "step": 8994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697171454324943e-05,
+      "loss": 0.3881,
+      "step": 8995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6970999971282402e-05,
+      "loss": 0.3674,
+      "step": 8996
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697028533006496e-05,
+      "loss": 0.3922,
+      "step": 8997
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6969570619604193e-05,
+      "loss": 0.3164,
+      "step": 8998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6968855839907208e-05,
+      "loss": 0.3882,
+      "step": 8999
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.69681409909811e-05,
+      "loss": 0.3536,
+      "step": 9000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.696742607283298e-05,
+      "loss": 0.3974,
+      "step": 9001
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696671108546994e-05,
+      "loss": 0.3892,
+      "step": 9002
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6965996028899086e-05,
+      "loss": 0.3633,
+      "step": 9003
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6965280903127525e-05,
+      "loss": 0.3572,
+      "step": 9004
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6964565708162354e-05,
+      "loss": 0.4146,
+      "step": 9005
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6963850444010685e-05,
+      "loss": 0.4216,
+      "step": 9006
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696313511067962e-05,
+      "loss": 0.3727,
+      "step": 9007
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6962419708176265e-05,
+      "loss": 0.3574,
+      "step": 9008
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6961704236507728e-05,
+      "loss": 0.361,
+      "step": 9009
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6960988695681115e-05,
+      "loss": 0.4502,
+      "step": 9010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696027308570354e-05,
+      "loss": 0.3859,
+      "step": 9011
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69595574065821e-05,
+      "loss": 0.3913,
+      "step": 9012
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6958841658323913e-05,
+      "loss": 0.402,
+      "step": 9013
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6958125840936093e-05,
+      "loss": 0.413,
+      "step": 9014
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6957409954425747e-05,
+      "loss": 0.4096,
+      "step": 9015
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695669399879998e-05,
+      "loss": 0.376,
+      "step": 9016
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6955977974065913e-05,
+      "loss": 0.3809,
+      "step": 9017
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6955261880230652e-05,
+      "loss": 0.326,
+      "step": 9018
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695454571730132e-05,
+      "loss": 0.403,
+      "step": 9019
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6953829485285026e-05,
+      "loss": 0.4986,
+      "step": 9020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6953113184188884e-05,
+      "loss": 0.3973,
+      "step": 9021
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695239681402001e-05,
+      "loss": 0.3895,
+      "step": 9022
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695168037478552e-05,
+      "loss": 0.394,
+      "step": 9023
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695096386649254e-05,
+      "loss": 0.4362,
+      "step": 9024
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6950247289148176e-05,
+      "loss": 0.3802,
+      "step": 9025
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694953064275955e-05,
+      "loss": 0.3757,
+      "step": 9026
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6948813927333787e-05,
+      "loss": 0.4216,
+      "step": 9027
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6948097142878e-05,
+      "loss": 0.452,
+      "step": 9028
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6947380289399308e-05,
+      "loss": 0.3897,
+      "step": 9029
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6946663366904837e-05,
+      "loss": 0.38,
+      "step": 9030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6945946375401715e-05,
+      "loss": 0.4272,
+      "step": 9031
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6945229314897052e-05,
+      "loss": 0.396,
+      "step": 9032
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6944512185397978e-05,
+      "loss": 0.3569,
+      "step": 9033
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6943794986911615e-05,
+      "loss": 0.4178,
+      "step": 9034
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694307771944509e-05,
+      "loss": 0.3532,
+      "step": 9035
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6942360383005527e-05,
+      "loss": 0.3932,
+      "step": 9036
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6941642977600056e-05,
+      "loss": 0.4361,
+      "step": 9037
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6940925503235797e-05,
+      "loss": 0.3831,
+      "step": 9038
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6940207959919878e-05,
+      "loss": 0.4006,
+      "step": 9039
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6939490347659434e-05,
+      "loss": 0.396,
+      "step": 9040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6938772666461587e-05,
+      "loss": 0.4293,
+      "step": 9041
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693805491633347e-05,
+      "loss": 0.3681,
+      "step": 9042
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693733709728221e-05,
+      "loss": 0.3796,
+      "step": 9043
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6936619209314938e-05,
+      "loss": 0.4352,
+      "step": 9044
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6935901252438794e-05,
+      "loss": 0.3888,
+      "step": 9045
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69351832266609e-05,
+      "loss": 0.3904,
+      "step": 9046
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6934465131988393e-05,
+      "loss": 0.3805,
+      "step": 9047
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6933746968428408e-05,
+      "loss": 0.442,
+      "step": 9048
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6933028735988075e-05,
+      "loss": 0.3395,
+      "step": 9049
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693231043467453e-05,
+      "loss": 0.4162,
+      "step": 9050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6931592064494915e-05,
+      "loss": 0.3584,
+      "step": 9051
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6930873625456362e-05,
+      "loss": 0.3187,
+      "step": 9052
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6930155117566004e-05,
+      "loss": 0.4486,
+      "step": 9053
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692943654083098e-05,
+      "loss": 0.5022,
+      "step": 9054
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692871789525844e-05,
+      "loss": 0.4303,
+      "step": 9055
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6927999180855502e-05,
+      "loss": 0.4504,
+      "step": 9056
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6927280397629326e-05,
+      "loss": 0.4087,
+      "step": 9057
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6926561545587042e-05,
+      "loss": 0.3561,
+      "step": 9058
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6925842624735792e-05,
+      "loss": 0.4443,
+      "step": 9059
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692512363508272e-05,
+      "loss": 0.4688,
+      "step": 9060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6924404576634966e-05,
+      "loss": 0.3782,
+      "step": 9061
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6923685449399674e-05,
+      "loss": 0.4409,
+      "step": 9062
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6922966253383988e-05,
+      "loss": 0.3877,
+      "step": 9063
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6922246988595058e-05,
+      "loss": 0.4098,
+      "step": 9064
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692152765504002e-05,
+      "loss": 0.4516,
+      "step": 9065
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6920808252726023e-05,
+      "loss": 0.3768,
+      "step": 9066
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6920088781660216e-05,
+      "loss": 0.3223,
+      "step": 9067
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6919369241849744e-05,
+      "loss": 0.4019,
+      "step": 9068
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691864963330176e-05,
+      "loss": 0.3256,
+      "step": 9069
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6917929956023406e-05,
+      "loss": 0.3592,
+      "step": 9070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6917210210021835e-05,
+      "loss": 0.4125,
+      "step": 9071
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6916490395304194e-05,
+      "loss": 0.357,
+      "step": 9072
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6915770511877636e-05,
+      "loss": 0.4821,
+      "step": 9073
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6915050559749312e-05,
+      "loss": 0.3654,
+      "step": 9074
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6914330538926377e-05,
+      "loss": 0.4188,
+      "step": 9075
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6913610449415977e-05,
+      "loss": 0.3653,
+      "step": 9076
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691289029122527e-05,
+      "loss": 0.5141,
+      "step": 9077
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691217006436141e-05,
+      "loss": 0.382,
+      "step": 9078
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691144976883155e-05,
+      "loss": 0.4372,
+      "step": 9079
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6910729404642848e-05,
+      "loss": 0.3469,
+      "step": 9080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691000897180246e-05,
+      "loss": 0.4459,
+      "step": 9081
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690928847031754e-05,
+      "loss": 0.4104,
+      "step": 9082
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6908567900195248e-05,
+      "loss": 0.4043,
+      "step": 9083
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690784726144274e-05,
+      "loss": 0.3712,
+      "step": 9084
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6907126554067175e-05,
+      "loss": 0.423,
+      "step": 9085
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6906405778075715e-05,
+      "loss": 0.3737,
+      "step": 9086
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690568493347552e-05,
+      "loss": 0.2995,
+      "step": 9087
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6904964020273747e-05,
+      "loss": 0.4549,
+      "step": 9088
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6904243038477563e-05,
+      "loss": 0.366,
+      "step": 9089
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6903521988094127e-05,
+      "loss": 0.3703,
+      "step": 9090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6902800869130604e-05,
+      "loss": 0.3538,
+      "step": 9091
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6902079681594156e-05,
+      "loss": 0.3808,
+      "step": 9092
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6901358425491947e-05,
+      "loss": 0.4315,
+      "step": 9093
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690063710083114e-05,
+      "loss": 0.4562,
+      "step": 9094
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6899915707618905e-05,
+      "loss": 0.4134,
+      "step": 9095
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689919424586241e-05,
+      "loss": 0.3613,
+      "step": 9096
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6898472715568813e-05,
+      "loss": 0.432,
+      "step": 9097
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6897751116745293e-05,
+      "loss": 0.3928,
+      "step": 9098
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6897029449399012e-05,
+      "loss": 0.4937,
+      "step": 9099
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6896307713537138e-05,
+      "loss": 0.4169,
+      "step": 9100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6895585909166846e-05,
+      "loss": 0.4517,
+      "step": 9101
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68948640362953e-05,
+      "loss": 0.3966,
+      "step": 9102
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689414209492968e-05,
+      "loss": 0.4132,
+      "step": 9103
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6893420085077142e-05,
+      "loss": 0.3413,
+      "step": 9104
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6892698006744876e-05,
+      "loss": 0.3403,
+      "step": 9105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6891975859940047e-05,
+      "loss": 0.4092,
+      "step": 9106
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689125364466983e-05,
+      "loss": 0.3505,
+      "step": 9107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68905313609414e-05,
+      "loss": 0.559,
+      "step": 9108
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688980900876193e-05,
+      "loss": 0.4197,
+      "step": 9109
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6889086588138596e-05,
+      "loss": 0.3853,
+      "step": 9110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688836409907858e-05,
+      "loss": 0.4296,
+      "step": 9111
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688764154158905e-05,
+      "loss": 0.4167,
+      "step": 9112
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6886918915677194e-05,
+      "loss": 0.3264,
+      "step": 9113
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6886196221350184e-05,
+      "loss": 0.3874,
+      "step": 9114
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6885473458615204e-05,
+      "loss": 0.4028,
+      "step": 9115
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6884750627479423e-05,
+      "loss": 0.2971,
+      "step": 9116
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688402772795004e-05,
+      "loss": 0.4458,
+      "step": 9117
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688330476003422e-05,
+      "loss": 0.4233,
+      "step": 9118
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688258172373915e-05,
+      "loss": 0.3612,
+      "step": 9119
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6881858619072017e-05,
+      "loss": 0.4109,
+      "step": 9120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6881135446040004e-05,
+      "loss": 0.4197,
+      "step": 9121
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6880412204650287e-05,
+      "loss": 0.4403,
+      "step": 9122
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687968889491006e-05,
+      "loss": 0.3427,
+      "step": 9123
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6878965516826497e-05,
+      "loss": 0.4638,
+      "step": 9124
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6878242070406797e-05,
+      "loss": 0.4387,
+      "step": 9125
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6877518555658145e-05,
+      "loss": 0.3726,
+      "step": 9126
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6876794972587717e-05,
+      "loss": 0.3593,
+      "step": 9127
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6876071321202713e-05,
+      "loss": 0.418,
+      "step": 9128
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687534760151032e-05,
+      "loss": 0.4237,
+      "step": 9129
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687462381351772e-05,
+      "loss": 0.4603,
+      "step": 9130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687389995723211e-05,
+      "loss": 0.402,
+      "step": 9131
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687317603266068e-05,
+      "loss": 0.2897,
+      "step": 9132
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687245203981062e-05,
+      "loss": 0.3623,
+      "step": 9133
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687172797868912e-05,
+      "loss": 0.4064,
+      "step": 9134
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687100384930338e-05,
+      "loss": 0.3992,
+      "step": 9135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687027965166059e-05,
+      "loss": 0.3938,
+      "step": 9136
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686955538576794e-05,
+      "loss": 0.3785,
+      "step": 9137
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686883105163263e-05,
+      "loss": 0.4018,
+      "step": 9138
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686810664926186e-05,
+      "loss": 0.3939,
+      "step": 9139
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686738217866281e-05,
+      "loss": 0.3734,
+      "step": 9140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686665763984269e-05,
+      "loss": 0.4427,
+      "step": 9141
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865933032808697e-05,
+      "loss": 0.3583,
+      "step": 9142
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865208357568028e-05,
+      "loss": 0.3624,
+      "step": 9143
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6864483614127877e-05,
+      "loss": 0.4988,
+      "step": 9144
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686375880249545e-05,
+      "loss": 0.4569,
+      "step": 9145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6863033922677948e-05,
+      "loss": 0.3629,
+      "step": 9146
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6862308974682565e-05,
+      "loss": 0.3965,
+      "step": 9147
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6861583958516508e-05,
+      "loss": 0.4693,
+      "step": 9148
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6860858874186982e-05,
+      "loss": 0.4009,
+      "step": 9149
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686013372170118e-05,
+      "loss": 0.405,
+      "step": 9150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6859408501066316e-05,
+      "loss": 0.3825,
+      "step": 9151
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6858683212289586e-05,
+      "loss": 0.4337,
+      "step": 9152
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6857957855378203e-05,
+      "loss": 0.3525,
+      "step": 9153
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6857232430339367e-05,
+      "loss": 0.364,
+      "step": 9154
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685650693718029e-05,
+      "loss": 0.41,
+      "step": 9155
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685578137590817e-05,
+      "loss": 0.3587,
+      "step": 9156
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6855055746530226e-05,
+      "loss": 0.3698,
+      "step": 9157
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685433004905366e-05,
+      "loss": 0.3743,
+      "step": 9158
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685360428348568e-05,
+      "loss": 0.3533,
+      "step": 9159
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68528784498335e-05,
+      "loss": 0.3928,
+      "step": 9160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6852152548104324e-05,
+      "loss": 0.4514,
+      "step": 9161
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685142657830537e-05,
+      "loss": 0.4256,
+      "step": 9162
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6850700540443847e-05,
+      "loss": 0.4049,
+      "step": 9163
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684997443452697e-05,
+      "loss": 0.415,
+      "step": 9164
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6849248260561947e-05,
+      "loss": 0.3584,
+      "step": 9165
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6848522018555998e-05,
+      "loss": 0.3833,
+      "step": 9166
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6847795708516333e-05,
+      "loss": 0.4032,
+      "step": 9167
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6847069330450167e-05,
+      "loss": 0.4165,
+      "step": 9168
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6846342884364718e-05,
+      "loss": 0.3786,
+      "step": 9169
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6845616370267202e-05,
+      "loss": 0.392,
+      "step": 9170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684488978816484e-05,
+      "loss": 0.4609,
+      "step": 9171
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684416313806484e-05,
+      "loss": 0.4535,
+      "step": 9172
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6843436419974436e-05,
+      "loss": 0.3463,
+      "step": 9173
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6842709633900828e-05,
+      "loss": 0.3808,
+      "step": 9174
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6841982779851254e-05,
+      "loss": 0.3869,
+      "step": 9175
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684125585783292e-05,
+      "loss": 0.394,
+      "step": 9176
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684052886785306e-05,
+      "loss": 0.3984,
+      "step": 9177
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6839801809918888e-05,
+      "loss": 0.3104,
+      "step": 9178
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6839074684037626e-05,
+      "loss": 0.4144,
+      "step": 9179
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6838347490216503e-05,
+      "loss": 0.4151,
+      "step": 9180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6837620228462737e-05,
+      "loss": 0.3893,
+      "step": 9181
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6836892898783553e-05,
+      "loss": 0.3891,
+      "step": 9182
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6836165501186187e-05,
+      "loss": 0.3538,
+      "step": 9183
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683543803567785e-05,
+      "loss": 0.3561,
+      "step": 9184
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683471050226578e-05,
+      "loss": 0.3672,
+      "step": 9185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68339829009572e-05,
+      "loss": 0.4197,
+      "step": 9186
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6833255231759338e-05,
+      "loss": 0.3823,
+      "step": 9187
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683252749467942e-05,
+      "loss": 0.3824,
+      "step": 9188
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683179968972468e-05,
+      "loss": 0.3511,
+      "step": 9189
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683107181690235e-05,
+      "loss": 0.4469,
+      "step": 9190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6830343876219647e-05,
+      "loss": 0.407,
+      "step": 9191
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682961586768382e-05,
+      "loss": 0.3132,
+      "step": 9192
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682888779130209e-05,
+      "loss": 0.3306,
+      "step": 9193
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6828159647081696e-05,
+      "loss": 0.4078,
+      "step": 9194
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682743143502987e-05,
+      "loss": 0.3932,
+      "step": 9195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682670315515384e-05,
+      "loss": 0.3626,
+      "step": 9196
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682597480746085e-05,
+      "loss": 0.4088,
+      "step": 9197
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682524639195813e-05,
+      "loss": 0.3707,
+      "step": 9198
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682451790865291e-05,
+      "loss": 0.3905,
+      "step": 9199
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6823789357552444e-05,
+      "loss": 0.4115,
+      "step": 9200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6823060738663954e-05,
+      "loss": 0.485,
+      "step": 9201
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6822332051994688e-05,
+      "loss": 0.3492,
+      "step": 9202
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6821603297551876e-05,
+      "loss": 0.4127,
+      "step": 9203
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6820874475342766e-05,
+      "loss": 0.3816,
+      "step": 9204
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6820145585374588e-05,
+      "loss": 0.3992,
+      "step": 9205
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6819416627654593e-05,
+      "loss": 0.4025,
+      "step": 9206
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6818687602190017e-05,
+      "loss": 0.3598,
+      "step": 9207
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6817958508988108e-05,
+      "loss": 0.4563,
+      "step": 9208
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6817229348056102e-05,
+      "loss": 0.4023,
+      "step": 9209
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681650011940124e-05,
+      "loss": 0.4418,
+      "step": 9210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681577082303078e-05,
+      "loss": 0.3874,
+      "step": 9211
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6815041458951954e-05,
+      "loss": 0.3766,
+      "step": 9212
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681431202717201e-05,
+      "loss": 0.3866,
+      "step": 9213
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6813582527698197e-05,
+      "loss": 0.407,
+      "step": 9214
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6812852960537763e-05,
+      "loss": 0.3393,
+      "step": 9215
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6812123325697952e-05,
+      "loss": 0.4116,
+      "step": 9216
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681139362318601e-05,
+      "loss": 0.3976,
+      "step": 9217
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6810663853009195e-05,
+      "loss": 0.4107,
+      "step": 9218
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680993401517475e-05,
+      "loss": 0.4064,
+      "step": 9219
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6809204109689926e-05,
+      "loss": 0.4078,
+      "step": 9220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6808474136561974e-05,
+      "loss": 0.3214,
+      "step": 9221
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6807744095798144e-05,
+      "loss": 0.4015,
+      "step": 9222
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6807013987405693e-05,
+      "loss": 0.3968,
+      "step": 9223
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680628381139187e-05,
+      "loss": 0.4163,
+      "step": 9224
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680555356776393e-05,
+      "loss": 0.4499,
+      "step": 9225
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6804823256529127e-05,
+      "loss": 0.4942,
+      "step": 9226
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6804092877694715e-05,
+      "loss": 0.372,
+      "step": 9227
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6803362431267952e-05,
+      "loss": 0.4166,
+      "step": 9228
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6802631917256097e-05,
+      "loss": 0.364,
+      "step": 9229
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6801901335666396e-05,
+      "loss": 0.375,
+      "step": 9230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6801170686506115e-05,
+      "loss": 0.3933,
+      "step": 9231
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6800439969782515e-05,
+      "loss": 0.3345,
+      "step": 9232
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679970918550285e-05,
+      "loss": 0.4036,
+      "step": 9233
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679897833367438e-05,
+      "loss": 0.3569,
+      "step": 9234
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6798247414304363e-05,
+      "loss": 0.355,
+      "step": 9235
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6797516427400066e-05,
+      "loss": 0.4695,
+      "step": 9236
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6796785372968746e-05,
+      "loss": 0.4629,
+      "step": 9237
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679605425101767e-05,
+      "loss": 0.4037,
+      "step": 9238
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6795323061554096e-05,
+      "loss": 0.2911,
+      "step": 9239
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6794591804585286e-05,
+      "loss": 0.3389,
+      "step": 9240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6793860480118514e-05,
+      "loss": 0.3781,
+      "step": 9241
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679312908816104e-05,
+      "loss": 0.3703,
+      "step": 9242
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6792397628720123e-05,
+      "loss": 0.3379,
+      "step": 9243
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6791666101803037e-05,
+      "loss": 0.5119,
+      "step": 9244
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679093450741705e-05,
+      "loss": 0.3919,
+      "step": 9245
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6790202845569426e-05,
+      "loss": 0.3806,
+      "step": 9246
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6789471116267432e-05,
+      "loss": 0.366,
+      "step": 9247
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678873931951834e-05,
+      "loss": 0.4474,
+      "step": 9248
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678800745532942e-05,
+      "loss": 0.4204,
+      "step": 9249
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678727552370794e-05,
+      "loss": 0.4144,
+      "step": 9250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6786543524661173e-05,
+      "loss": 0.3595,
+      "step": 9251
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678581145819639e-05,
+      "loss": 0.3345,
+      "step": 9252
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6785079324320864e-05,
+      "loss": 0.4648,
+      "step": 9253
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678434712304187e-05,
+      "loss": 0.4138,
+      "step": 9254
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678361485436668e-05,
+      "loss": 0.4263,
+      "step": 9255
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6782882518302563e-05,
+      "loss": 0.3974,
+      "step": 9256
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6782150114856804e-05,
+      "loss": 0.4005,
+      "step": 9257
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678141764403667e-05,
+      "loss": 0.3706,
+      "step": 9258
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678068510584944e-05,
+      "loss": 0.4165,
+      "step": 9259
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6779952500302396e-05,
+      "loss": 0.377,
+      "step": 9260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6779219827402812e-05,
+      "loss": 0.4495,
+      "step": 9261
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6778487087157966e-05,
+      "loss": 0.379,
+      "step": 9262
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6777754279575135e-05,
+      "loss": 0.438,
+      "step": 9263
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.67770214046616e-05,
+      "loss": 0.4164,
+      "step": 9264
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677628846242465e-05,
+      "loss": 0.4905,
+      "step": 9265
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6775555452871552e-05,
+      "loss": 0.4404,
+      "step": 9266
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.67748223760096e-05,
+      "loss": 0.5218,
+      "step": 9267
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6774089231846066e-05,
+      "loss": 0.4301,
+      "step": 9268
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677335602038824e-05,
+      "loss": 0.3783,
+      "step": 9269
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6772622741643407e-05,
+      "loss": 0.3741,
+      "step": 9270
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677188939561884e-05,
+      "loss": 0.3589,
+      "step": 9271
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6771155982321842e-05,
+      "loss": 0.4041,
+      "step": 9272
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6770422501759683e-05,
+      "loss": 0.4034,
+      "step": 9273
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676968895393966e-05,
+      "loss": 0.3793,
+      "step": 9274
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676895533886905e-05,
+      "loss": 0.3953,
+      "step": 9275
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6768221656555147e-05,
+      "loss": 0.4179,
+      "step": 9276
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6767487907005248e-05,
+      "loss": 0.4609,
+      "step": 9277
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6766754090226625e-05,
+      "loss": 0.35,
+      "step": 9278
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676602020622658e-05,
+      "loss": 0.4507,
+      "step": 9279
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6765286255012393e-05,
+      "loss": 0.3165,
+      "step": 9280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6764552236591367e-05,
+      "loss": 0.3655,
+      "step": 9281
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6763818150970788e-05,
+      "loss": 0.3909,
+      "step": 9282
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676308399815795e-05,
+      "loss": 0.4072,
+      "step": 9283
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676234977816014e-05,
+      "loss": 0.3635,
+      "step": 9284
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676161549098466e-05,
+      "loss": 0.4313,
+      "step": 9285
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6760881136638804e-05,
+      "loss": 0.4043,
+      "step": 9286
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6760146715129858e-05,
+      "loss": 0.4293,
+      "step": 9287
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6759412226465125e-05,
+      "loss": 0.396,
+      "step": 9288
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6758677670651905e-05,
+      "loss": 0.4345,
+      "step": 9289
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6757943047697488e-05,
+      "loss": 0.4041,
+      "step": 9290
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6757208357609177e-05,
+      "loss": 0.3687,
+      "step": 9291
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6756473600394264e-05,
+      "loss": 0.3776,
+      "step": 9292
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6755738776060053e-05,
+      "loss": 0.3894,
+      "step": 9293
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6755003884613844e-05,
+      "loss": 0.3741,
+      "step": 9294
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6754268926062936e-05,
+      "loss": 0.3464,
+      "step": 9295
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6753533900414632e-05,
+      "loss": 0.4385,
+      "step": 9296
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675279880767623e-05,
+      "loss": 0.4947,
+      "step": 9297
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6752063647855034e-05,
+      "loss": 0.4379,
+      "step": 9298
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675132842095835e-05,
+      "loss": 0.4384,
+      "step": 9299
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675059312699348e-05,
+      "loss": 0.3965,
+      "step": 9300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6749857765967726e-05,
+      "loss": 0.4085,
+      "step": 9301
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6749122337888394e-05,
+      "loss": 0.3715,
+      "step": 9302
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6748386842762795e-05,
+      "loss": 0.4355,
+      "step": 9303
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6747651280598232e-05,
+      "loss": 0.3712,
+      "step": 9304
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6746915651402006e-05,
+      "loss": 0.4299,
+      "step": 9305
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6746179955181438e-05,
+      "loss": 0.3561,
+      "step": 9306
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6745444191943826e-05,
+      "loss": 0.3941,
+      "step": 9307
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6744708361696476e-05,
+      "loss": 0.4575,
+      "step": 9308
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6743972464446713e-05,
+      "loss": 0.3739,
+      "step": 9309
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674323650020183e-05,
+      "loss": 0.4004,
+      "step": 9310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674250046896915e-05,
+      "loss": 0.4625,
+      "step": 9311
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6741764370755985e-05,
+      "loss": 0.4039,
+      "step": 9312
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674102820556964e-05,
+      "loss": 0.4058,
+      "step": 9313
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674029197341743e-05,
+      "loss": 0.4688,
+      "step": 9314
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6739555674306675e-05,
+      "loss": 0.4216,
+      "step": 9315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673881930824468e-05,
+      "loss": 0.305,
+      "step": 9316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6738082875238767e-05,
+      "loss": 0.3669,
+      "step": 9317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6737346375296252e-05,
+      "loss": 0.3634,
+      "step": 9318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6736609808424447e-05,
+      "loss": 0.5173,
+      "step": 9319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6735873174630675e-05,
+      "loss": 0.4026,
+      "step": 9320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6735136473922246e-05,
+      "loss": 0.37,
+      "step": 9321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673439970630648e-05,
+      "loss": 0.4135,
+      "step": 9322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6733662871790705e-05,
+      "loss": 0.4005,
+      "step": 9323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673292597038223e-05,
+      "loss": 0.3966,
+      "step": 9324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6732189002088386e-05,
+      "loss": 0.4414,
+      "step": 9325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6731451966916485e-05,
+      "loss": 0.4225,
+      "step": 9326
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.673071486487385e-05,
+      "loss": 0.4246,
+      "step": 9327
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6729977695967804e-05,
+      "loss": 0.4118,
+      "step": 9328
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6729240460205677e-05,
+      "loss": 0.3513,
+      "step": 9329
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6728503157594782e-05,
+      "loss": 0.3828,
+      "step": 9330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672776578814245e-05,
+      "loss": 0.3718,
+      "step": 9331
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6727028351856007e-05,
+      "loss": 0.3523,
+      "step": 9332
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672629084874277e-05,
+      "loss": 0.3763,
+      "step": 9333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6725553278810072e-05,
+      "loss": 0.4147,
+      "step": 9334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6724815642065245e-05,
+      "loss": 0.3664,
+      "step": 9335
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6724077938515614e-05,
+      "loss": 0.5301,
+      "step": 9336
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.67233401681685e-05,
+      "loss": 0.399,
+      "step": 9337
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6722602331031237e-05,
+      "loss": 0.4112,
+      "step": 9338
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6721864427111154e-05,
+      "loss": 0.4017,
+      "step": 9339
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6721126456415585e-05,
+      "loss": 0.4071,
+      "step": 9340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6720388418951858e-05,
+      "loss": 0.4582,
+      "step": 9341
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6719650314727304e-05,
+      "loss": 0.4237,
+      "step": 9342
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6718912143749256e-05,
+      "loss": 0.4228,
+      "step": 9343
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6718173906025046e-05,
+      "loss": 0.4306,
+      "step": 9344
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671743560156201e-05,
+      "loss": 0.4019,
+      "step": 9345
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671669723036748e-05,
+      "loss": 0.4218,
+      "step": 9346
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6715958792448798e-05,
+      "loss": 0.4371,
+      "step": 9347
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6715220287813288e-05,
+      "loss": 0.4945,
+      "step": 9348
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6714481716468292e-05,
+      "loss": 0.4171,
+      "step": 9349
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6713743078421153e-05,
+      "loss": 0.4668,
+      "step": 9350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6713004373679198e-05,
+      "loss": 0.4515,
+      "step": 9351
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6712265602249777e-05,
+      "loss": 0.4713,
+      "step": 9352
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6711526764140215e-05,
+      "loss": 0.4429,
+      "step": 9353
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6710787859357863e-05,
+      "loss": 0.4103,
+      "step": 9354
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6710048887910058e-05,
+      "loss": 0.3445,
+      "step": 9355
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670930984980414e-05,
+      "loss": 0.3376,
+      "step": 9356
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670857074504745e-05,
+      "loss": 0.3994,
+      "step": 9357
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6707831573647332e-05,
+      "loss": 0.4345,
+      "step": 9358
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670709233561113e-05,
+      "loss": 0.3873,
+      "step": 9359
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6706353030946184e-05,
+      "loss": 0.4866,
+      "step": 9360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670561365965984e-05,
+      "loss": 0.4049,
+      "step": 9361
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6704874221759445e-05,
+      "loss": 0.4008,
+      "step": 9362
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6704134717252346e-05,
+      "loss": 0.3206,
+      "step": 9363
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6703395146145882e-05,
+      "loss": 0.3519,
+      "step": 9364
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6702655508447402e-05,
+      "loss": 0.4708,
+      "step": 9365
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6701915804164262e-05,
+      "loss": 0.4261,
+      "step": 9366
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6701176033303804e-05,
+      "loss": 0.4474,
+      "step": 9367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670043619587337e-05,
+      "loss": 0.4222,
+      "step": 9368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6699696291880324e-05,
+      "loss": 0.3449,
+      "step": 9369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6698956321332005e-05,
+      "loss": 0.3407,
+      "step": 9370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6698216284235772e-05,
+      "loss": 0.3319,
+      "step": 9371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669747618059897e-05,
+      "loss": 0.3794,
+      "step": 9372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669673601042895e-05,
+      "loss": 0.3608,
+      "step": 9373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695995773733075e-05,
+      "loss": 0.34,
+      "step": 9374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695255470518687e-05,
+      "loss": 0.3789,
+      "step": 9375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6694515100793148e-05,
+      "loss": 0.3324,
+      "step": 9376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669377466456381e-05,
+      "loss": 0.3979,
+      "step": 9377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669303416183803e-05,
+      "loss": 0.4229,
+      "step": 9378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669229359262316e-05,
+      "loss": 0.4974,
+      "step": 9379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669155295692656e-05,
+      "loss": 0.4046,
+      "step": 9380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6690812254755593e-05,
+      "loss": 0.378,
+      "step": 9381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6690071486117604e-05,
+      "loss": 0.4286,
+      "step": 9382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6689330651019967e-05,
+      "loss": 0.3651,
+      "step": 9383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668858974947003e-05,
+      "loss": 0.4105,
+      "step": 9384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6687848781475156e-05,
+      "loss": 0.4502,
+      "step": 9385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668710774704271e-05,
+      "loss": 0.4116,
+      "step": 9386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6686366646180047e-05,
+      "loss": 0.5047,
+      "step": 9387
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6685625478894536e-05,
+      "loss": 0.3752,
+      "step": 9388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6684884245193534e-05,
+      "loss": 0.4051,
+      "step": 9389
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668414294508441e-05,
+      "loss": 0.3975,
+      "step": 9390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6683401578574524e-05,
+      "loss": 0.3727,
+      "step": 9391
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6682660145671238e-05,
+      "loss": 0.3937,
+      "step": 9392
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6681918646381924e-05,
+      "loss": 0.4201,
+      "step": 9393
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6681177080713945e-05,
+      "loss": 0.3987,
+      "step": 9394
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6680435448674673e-05,
+      "loss": 0.4116,
+      "step": 9395
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6679693750271464e-05,
+      "loss": 0.4062,
+      "step": 9396
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6678951985511694e-05,
+      "loss": 0.4122,
+      "step": 9397
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6678210154402732e-05,
+      "loss": 0.4655,
+      "step": 9398
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6677468256951944e-05,
+      "loss": 0.3871,
+      "step": 9399
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6676726293166706e-05,
+      "loss": 0.4342,
+      "step": 9400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667598426305438e-05,
+      "loss": 0.4019,
+      "step": 9401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6675242166622345e-05,
+      "loss": 0.3424,
+      "step": 9402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667450000387797e-05,
+      "loss": 0.4289,
+      "step": 9403
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6673757774828625e-05,
+      "loss": 0.4442,
+      "step": 9404
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6673015479481688e-05,
+      "loss": 0.3229,
+      "step": 9405
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6672273117844532e-05,
+      "loss": 0.4017,
+      "step": 9406
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6671530689924527e-05,
+      "loss": 0.4215,
+      "step": 9407
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6670788195729057e-05,
+      "loss": 0.402,
+      "step": 9408
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667004563526549e-05,
+      "loss": 0.4422,
+      "step": 9409
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6669303008541207e-05,
+      "loss": 0.4216,
+      "step": 9410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6668560315563584e-05,
+      "loss": 0.4287,
+      "step": 9411
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6667817556339998e-05,
+      "loss": 0.4093,
+      "step": 9412
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6667074730877826e-05,
+      "loss": 0.3541,
+      "step": 9413
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6666331839184457e-05,
+      "loss": 0.3241,
+      "step": 9414
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666558888126726e-05,
+      "loss": 0.3238,
+      "step": 9415
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6664845857133615e-05,
+      "loss": 0.3956,
+      "step": 9416
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666410276679091e-05,
+      "loss": 0.4141,
+      "step": 9417
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6663359610246532e-05,
+      "loss": 0.3929,
+      "step": 9418
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6662616387507852e-05,
+      "loss": 0.3906,
+      "step": 9419
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666187309858225e-05,
+      "loss": 0.3473,
+      "step": 9420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6661129743477126e-05,
+      "loss": 0.4761,
+      "step": 9421
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6660386322199856e-05,
+      "loss": 0.4052,
+      "step": 9422
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665964283475782e-05,
+      "loss": 0.3635,
+      "step": 9423
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6658899281158412e-05,
+      "loss": 0.4636,
+      "step": 9424
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6658155661409017e-05,
+      "loss": 0.4622,
+      "step": 9425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6657411975517018e-05,
+      "loss": 0.4215,
+      "step": 9426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6656668223489804e-05,
+      "loss": 0.4495,
+      "step": 9427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6655924405334766e-05,
+      "loss": 0.3323,
+      "step": 9428
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6655180521059295e-05,
+      "loss": 0.4221,
+      "step": 9429
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6654436570670775e-05,
+      "loss": 0.2897,
+      "step": 9430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6653692554176602e-05,
+      "loss": 0.401,
+      "step": 9431
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6652948471584163e-05,
+      "loss": 0.3653,
+      "step": 9432
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665220432290085e-05,
+      "loss": 0.3279,
+      "step": 9433
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6651460108134058e-05,
+      "loss": 0.4396,
+      "step": 9434
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6650715827291178e-05,
+      "loss": 0.3742,
+      "step": 9435
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6649971480379607e-05,
+      "loss": 0.3805,
+      "step": 9436
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6649227067406733e-05,
+      "loss": 0.3522,
+      "step": 9437
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6648482588379953e-05,
+      "loss": 0.4038,
+      "step": 9438
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664773804330667e-05,
+      "loss": 0.3877,
+      "step": 9439
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6646993432194275e-05,
+      "loss": 0.3682,
+      "step": 9440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664624875505016e-05,
+      "loss": 0.3776,
+      "step": 9441
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6645504011881734e-05,
+      "loss": 0.483,
+      "step": 9442
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664475920269638e-05,
+      "loss": 0.3604,
+      "step": 9443
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6644014327501515e-05,
+      "loss": 0.3719,
+      "step": 9444
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6643269386304526e-05,
+      "loss": 0.4068,
+      "step": 9445
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664252437911282e-05,
+      "loss": 0.3738,
+      "step": 9446
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664177930593379e-05,
+      "loss": 0.4777,
+      "step": 9447
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6641034166774846e-05,
+      "loss": 0.3611,
+      "step": 9448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6640288961643384e-05,
+      "loss": 0.4045,
+      "step": 9449
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6639543690546812e-05,
+      "loss": 0.44,
+      "step": 9450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6638798353492533e-05,
+      "loss": 0.401,
+      "step": 9451
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6638052950487943e-05,
+      "loss": 0.4426,
+      "step": 9452
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.663730748154046e-05,
+      "loss": 0.4328,
+      "step": 9453
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.663656194665748e-05,
+      "loss": 0.3817,
+      "step": 9454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6635816345846413e-05,
+      "loss": 0.4875,
+      "step": 9455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6635070679114663e-05,
+      "loss": 0.4948,
+      "step": 9456
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.663432494646964e-05,
+      "loss": 0.4335,
+      "step": 9457
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6633579147918755e-05,
+      "loss": 0.4296,
+      "step": 9458
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6632833283469413e-05,
+      "loss": 0.4129,
+      "step": 9459
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.663208735312902e-05,
+      "loss": 0.4078,
+      "step": 9460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6631341356904993e-05,
+      "loss": 0.4463,
+      "step": 9461
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6630595294804738e-05,
+      "loss": 0.4609,
+      "step": 9462
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6629849166835667e-05,
+      "loss": 0.3739,
+      "step": 9463
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6629102973005197e-05,
+      "loss": 0.3846,
+      "step": 9464
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6628356713320737e-05,
+      "loss": 0.3951,
+      "step": 9465
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66276103877897e-05,
+      "loss": 0.3698,
+      "step": 9466
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6626863996419497e-05,
+      "loss": 0.3578,
+      "step": 9467
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662611753921755e-05,
+      "loss": 0.3603,
+      "step": 9468
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662537101619127e-05,
+      "loss": 0.3786,
+      "step": 9469
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6624624427348074e-05,
+      "loss": 0.405,
+      "step": 9470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662387777269538e-05,
+      "loss": 0.392,
+      "step": 9471
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66231310522406e-05,
+      "loss": 0.4146,
+      "step": 9472
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6622384265991156e-05,
+      "loss": 0.3862,
+      "step": 9473
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6621637413954466e-05,
+      "loss": 0.4039,
+      "step": 9474
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6620890496137953e-05,
+      "loss": 0.4336,
+      "step": 9475
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6620143512549033e-05,
+      "loss": 0.3284,
+      "step": 9476
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6619396463195127e-05,
+      "loss": 0.3579,
+      "step": 9477
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661864934808365e-05,
+      "loss": 0.3275,
+      "step": 9478
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6617902167222035e-05,
+      "loss": 0.4253,
+      "step": 9479
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6617154920617703e-05,
+      "loss": 0.4572,
+      "step": 9480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6616407608278068e-05,
+      "loss": 0.4095,
+      "step": 9481
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6615660230210563e-05,
+      "loss": 0.4433,
+      "step": 9482
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661491278642261e-05,
+      "loss": 0.3551,
+      "step": 9483
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661416527692163e-05,
+      "loss": 0.4174,
+      "step": 9484
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6613417701715056e-05,
+      "loss": 0.4279,
+      "step": 9485
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6612670060810307e-05,
+      "loss": 0.3351,
+      "step": 9486
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661192235421482e-05,
+      "loss": 0.3984,
+      "step": 9487
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661117458193601e-05,
+      "loss": 0.3531,
+      "step": 9488
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6610426743981313e-05,
+      "loss": 0.3569,
+      "step": 9489
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6609678840358163e-05,
+      "loss": 0.383,
+      "step": 9490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6608930871073977e-05,
+      "loss": 0.385,
+      "step": 9491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6608182836136192e-05,
+      "loss": 0.4312,
+      "step": 9492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6607434735552247e-05,
+      "loss": 0.3623,
+      "step": 9493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660668656932956e-05,
+      "loss": 0.4037,
+      "step": 9494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660593833747557e-05,
+      "loss": 0.4476,
+      "step": 9495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660519003999771e-05,
+      "loss": 0.5548,
+      "step": 9496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6604441676903414e-05,
+      "loss": 0.3032,
+      "step": 9497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660369324820011e-05,
+      "loss": 0.4047,
+      "step": 9498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6602944753895244e-05,
+      "loss": 0.4449,
+      "step": 9499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660219619399624e-05,
+      "loss": 0.4246,
+      "step": 9500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6601447568510545e-05,
+      "loss": 0.3837,
+      "step": 9501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6600698877445594e-05,
+      "loss": 0.3404,
+      "step": 9502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6599950120808813e-05,
+      "loss": 0.4403,
+      "step": 9503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6599201298607657e-05,
+      "loss": 0.3843,
+      "step": 9504
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659845241084955e-05,
+      "loss": 0.4057,
+      "step": 9505
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6597703457541943e-05,
+      "loss": 0.4423,
+      "step": 9506
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6596954438692266e-05,
+      "loss": 0.4462,
+      "step": 9507
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659620535430797e-05,
+      "loss": 0.4156,
+      "step": 9508
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6595456204396487e-05,
+      "loss": 0.4455,
+      "step": 9509
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659470698896527e-05,
+      "loss": 0.3898,
+      "step": 9510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659395770802175e-05,
+      "loss": 0.396,
+      "step": 9511
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6593208361573375e-05,
+      "loss": 0.4044,
+      "step": 9512
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6592458949627593e-05,
+      "loss": 0.4446,
+      "step": 9513
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6591709472191842e-05,
+      "loss": 0.4131,
+      "step": 9514
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6590959929273574e-05,
+      "loss": 0.4348,
+      "step": 9515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659021032088023e-05,
+      "loss": 0.3739,
+      "step": 9516
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.658946064701926e-05,
+      "loss": 0.4474,
+      "step": 9517
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6588710907698108e-05,
+      "loss": 0.2933,
+      "step": 9518
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6587961102924225e-05,
+      "loss": 0.364,
+      "step": 9519
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.658721123270506e-05,
+      "loss": 0.3322,
+      "step": 9520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.658646129704806e-05,
+      "loss": 0.3426,
+      "step": 9521
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6585711295960674e-05,
+      "loss": 0.4257,
+      "step": 9522
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6584961229450358e-05,
+      "loss": 0.3983,
+      "step": 9523
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6584211097524554e-05,
+      "loss": 0.4531,
+      "step": 9524
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.658346090019072e-05,
+      "loss": 0.3803,
+      "step": 9525
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6582710637456312e-05,
+      "loss": 0.3564,
+      "step": 9526
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6581960309328774e-05,
+      "loss": 0.3448,
+      "step": 9527
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6581209915815568e-05,
+      "loss": 0.3984,
+      "step": 9528
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6580459456924144e-05,
+      "loss": 0.4235,
+      "step": 9529
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657970893266196e-05,
+      "loss": 0.4149,
+      "step": 9530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657895834303647e-05,
+      "loss": 0.353,
+      "step": 9531
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6578207688055128e-05,
+      "loss": 0.3483,
+      "step": 9532
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6577456967725396e-05,
+      "loss": 0.3463,
+      "step": 9533
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6576706182054726e-05,
+      "loss": 0.4322,
+      "step": 9534
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6575955331050585e-05,
+      "loss": 0.4698,
+      "step": 9535
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6575204414720424e-05,
+      "loss": 0.4336,
+      "step": 9536
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6574453433071703e-05,
+      "loss": 0.3619,
+      "step": 9537
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6573702386111884e-05,
+      "loss": 0.4018,
+      "step": 9538
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6572951273848433e-05,
+      "loss": 0.3348,
+      "step": 9539
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6572200096288805e-05,
+      "loss": 0.4825,
+      "step": 9540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6571448853440467e-05,
+      "loss": 0.3931,
+      "step": 9541
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570697545310876e-05,
+      "loss": 0.3761,
+      "step": 9542
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65699461719075e-05,
+      "loss": 0.3912,
+      "step": 9543
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6569194733237802e-05,
+      "loss": 0.3909,
+      "step": 9544
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6568443229309247e-05,
+      "loss": 0.4064,
+      "step": 9545
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65676916601293e-05,
+      "loss": 0.373,
+      "step": 9546
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656694002570543e-05,
+      "loss": 0.4041,
+      "step": 9547
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6566188326045097e-05,
+      "loss": 0.4542,
+      "step": 9548
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6565436561155778e-05,
+      "loss": 0.306,
+      "step": 9549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656468473104493e-05,
+      "loss": 0.4168,
+      "step": 9550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656393283572003e-05,
+      "loss": 0.4021,
+      "step": 9551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6563180875188546e-05,
+      "loss": 0.4248,
+      "step": 9552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6562428849457947e-05,
+      "loss": 0.3448,
+      "step": 9553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6561676758535706e-05,
+      "loss": 0.3668,
+      "step": 9554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656092460242929e-05,
+      "loss": 0.397,
+      "step": 9555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656017238114617e-05,
+      "loss": 0.4145,
+      "step": 9556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6559420094693822e-05,
+      "loss": 0.4373,
+      "step": 9557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6558667743079722e-05,
+      "loss": 0.3711,
+      "step": 9558
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655791532631134e-05,
+      "loss": 0.3903,
+      "step": 9559
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655716284439615e-05,
+      "loss": 0.371,
+      "step": 9560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655641029734163e-05,
+      "loss": 0.455,
+      "step": 9561
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6555657685155255e-05,
+      "loss": 0.4075,
+      "step": 9562
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65549050078445e-05,
+      "loss": 0.382,
+      "step": 9563
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6554152265416842e-05,
+      "loss": 0.3528,
+      "step": 9564
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655339945787976e-05,
+      "loss": 0.4916,
+      "step": 9565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6552646585240736e-05,
+      "loss": 0.3905,
+      "step": 9566
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655189364750724e-05,
+      "loss": 0.3932,
+      "step": 9567
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6551140644686762e-05,
+      "loss": 0.3926,
+      "step": 9568
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6550387576786775e-05,
+      "loss": 0.3705,
+      "step": 9569
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6549634443814762e-05,
+      "loss": 0.4377,
+      "step": 9570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654888124577821e-05,
+      "loss": 0.4497,
+      "step": 9571
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6548127982684588e-05,
+      "loss": 0.422,
+      "step": 9572
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6547374654541392e-05,
+      "loss": 0.4201,
+      "step": 9573
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6546621261356105e-05,
+      "loss": 0.4766,
+      "step": 9574
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6545867803136203e-05,
+      "loss": 0.4439,
+      "step": 9575
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6545114279889176e-05,
+      "loss": 0.4303,
+      "step": 9576
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6544360691622508e-05,
+      "loss": 0.4176,
+      "step": 9577
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654360703834369e-05,
+      "loss": 0.3937,
+      "step": 9578
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65428533200602e-05,
+      "loss": 0.426,
+      "step": 9579
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6542099536779533e-05,
+      "loss": 0.3518,
+      "step": 9580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6541345688509176e-05,
+      "loss": 0.4455,
+      "step": 9581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6540591775256613e-05,
+      "loss": 0.3942,
+      "step": 9582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.653983779702934e-05,
+      "loss": 0.4164,
+      "step": 9583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6539083753834843e-05,
+      "loss": 0.4164,
+      "step": 9584
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6538329645680612e-05,
+      "loss": 0.3294,
+      "step": 9585
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.653757547257414e-05,
+      "loss": 0.3539,
+      "step": 9586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653682123452292e-05,
+      "loss": 0.4303,
+      "step": 9587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6536066931534444e-05,
+      "loss": 0.3519,
+      "step": 9588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.65353125636162e-05,
+      "loss": 0.598,
+      "step": 9589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6534558130775695e-05,
+      "loss": 0.3923,
+      "step": 9590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6533803633020413e-05,
+      "loss": 0.3917,
+      "step": 9591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6533049070357846e-05,
+      "loss": 0.4124,
+      "step": 9592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6532294442795504e-05,
+      "loss": 0.3408,
+      "step": 9593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653153975034087e-05,
+      "loss": 0.4528,
+      "step": 9594
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6530784993001448e-05,
+      "loss": 0.4505,
+      "step": 9595
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6530030170784734e-05,
+      "loss": 0.3191,
+      "step": 9596
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652927528369822e-05,
+      "loss": 0.3894,
+      "step": 9597
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652852033174942e-05,
+      "loss": 0.4432,
+      "step": 9598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6527765314945822e-05,
+      "loss": 0.3488,
+      "step": 9599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652701023329493e-05,
+      "loss": 0.5349,
+      "step": 9600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652625508680425e-05,
+      "loss": 0.3207,
+      "step": 9601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652549987548127e-05,
+      "loss": 0.3923,
+      "step": 9602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6524744599333505e-05,
+      "loss": 0.3398,
+      "step": 9603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6523989258368454e-05,
+      "loss": 0.3396,
+      "step": 9604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6523233852593622e-05,
+      "loss": 0.4029,
+      "step": 9605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6522478382016507e-05,
+      "loss": 0.3589,
+      "step": 9606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652172284664462e-05,
+      "loss": 0.4041,
+      "step": 9607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652096724648547e-05,
+      "loss": 0.5112,
+      "step": 9608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652021158154655e-05,
+      "loss": 0.4333,
+      "step": 9609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651945585183538e-05,
+      "loss": 0.4668,
+      "step": 9610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6518700057359463e-05,
+      "loss": 0.4017,
+      "step": 9611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651794419812631e-05,
+      "loss": 0.3897,
+      "step": 9612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651718827414342e-05,
+      "loss": 0.3773,
+      "step": 9613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651643228541831e-05,
+      "loss": 0.3565,
+      "step": 9614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6515676231958488e-05,
+      "loss": 0.4137,
+      "step": 9615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6514920113771468e-05,
+      "loss": 0.3655,
+      "step": 9616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651416393086476e-05,
+      "loss": 0.4056,
+      "step": 9617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6513407683245872e-05,
+      "loss": 0.3905,
+      "step": 9618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6512651370922322e-05,
+      "loss": 0.4184,
+      "step": 9619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6511894993901617e-05,
+      "loss": 0.4107,
+      "step": 9620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651113855219128e-05,
+      "loss": 0.3394,
+      "step": 9621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651038204579882e-05,
+      "loss": 0.398,
+      "step": 9622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650962547473175e-05,
+      "loss": 0.3898,
+      "step": 9623
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6508868838997586e-05,
+      "loss": 0.4026,
+      "step": 9624
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650811213860385e-05,
+      "loss": 0.4823,
+      "step": 9625
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6507355373558054e-05,
+      "loss": 0.3647,
+      "step": 9626
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650659854386772e-05,
+      "loss": 0.4544,
+      "step": 9627
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650584164954036e-05,
+      "loss": 0.4434,
+      "step": 9628
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.65050846905835e-05,
+      "loss": 0.4354,
+      "step": 9629
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6504327667004657e-05,
+      "loss": 0.4487,
+      "step": 9630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650357057881135e-05,
+      "loss": 0.4611,
+      "step": 9631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6502813426011104e-05,
+      "loss": 0.443,
+      "step": 9632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6502056208611434e-05,
+      "loss": 0.4025,
+      "step": 9633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6501298926619867e-05,
+      "loss": 0.3733,
+      "step": 9634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650054158004392e-05,
+      "loss": 0.4228,
+      "step": 9635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649978416889113e-05,
+      "loss": 0.4867,
+      "step": 9636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649902669316901e-05,
+      "loss": 0.3889,
+      "step": 9637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6498269152885086e-05,
+      "loss": 0.3557,
+      "step": 9638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6497511548046884e-05,
+      "loss": 0.4079,
+      "step": 9639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6496753878661932e-05,
+      "loss": 0.4164,
+      "step": 9640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6495996144737753e-05,
+      "loss": 0.3921,
+      "step": 9641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649523834628188e-05,
+      "loss": 0.3564,
+      "step": 9642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6494480483301836e-05,
+      "loss": 0.464,
+      "step": 9643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6493722555805156e-05,
+      "loss": 0.4552,
+      "step": 9644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649296456379936e-05,
+      "loss": 0.3755,
+      "step": 9645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6492206507291987e-05,
+      "loss": 0.3812,
+      "step": 9646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6491448386290562e-05,
+      "loss": 0.4005,
+      "step": 9647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6490690200802615e-05,
+      "loss": 0.3402,
+      "step": 9648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6489931950835684e-05,
+      "loss": 0.3819,
+      "step": 9649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6489173636397294e-05,
+      "loss": 0.4868,
+      "step": 9650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6488415257494988e-05,
+      "loss": 0.4053,
+      "step": 9651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6487656814136293e-05,
+      "loss": 0.4024,
+      "step": 9652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648689830632874e-05,
+      "loss": 0.4154,
+      "step": 9653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6486139734079872e-05,
+      "loss": 0.4099,
+      "step": 9654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6485381097397223e-05,
+      "loss": 0.368,
+      "step": 9655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6484622396288327e-05,
+      "loss": 0.4253,
+      "step": 9656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6483863630760718e-05,
+      "loss": 0.3814,
+      "step": 9657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6483104800821942e-05,
+      "loss": 0.3768,
+      "step": 9658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6482345906479528e-05,
+      "loss": 0.3982,
+      "step": 9659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648158694774102e-05,
+      "loss": 0.3853,
+      "step": 9660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648082792461396e-05,
+      "loss": 0.526,
+      "step": 9661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6480068837105885e-05,
+      "loss": 0.3806,
+      "step": 9662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6479309685224333e-05,
+      "loss": 0.4155,
+      "step": 9663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6478550468976852e-05,
+      "loss": 0.3249,
+      "step": 9664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6477791188370978e-05,
+      "loss": 0.4221,
+      "step": 9665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6477031843414256e-05,
+      "loss": 0.3812,
+      "step": 9666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647627243411423e-05,
+      "loss": 0.3701,
+      "step": 9667
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647551296047845e-05,
+      "loss": 0.4676,
+      "step": 9668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6474753422514445e-05,
+      "loss": 0.4135,
+      "step": 9669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6473993820229778e-05,
+      "loss": 0.4636,
+      "step": 9670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647323415363198e-05,
+      "loss": 0.3852,
+      "step": 9671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647247442272861e-05,
+      "loss": 0.4526,
+      "step": 9672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6471714627527203e-05,
+      "loss": 0.3767,
+      "step": 9673
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6470954768035317e-05,
+      "loss": 0.4195,
+      "step": 9674
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6470194844260497e-05,
+      "loss": 0.4328,
+      "step": 9675
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6469434856210293e-05,
+      "loss": 0.4187,
+      "step": 9676
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6468674803892254e-05,
+      "loss": 0.3974,
+      "step": 9677
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6467914687313928e-05,
+      "loss": 0.4182,
+      "step": 9678
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6467154506482868e-05,
+      "loss": 0.4754,
+      "step": 9679
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6466394261406628e-05,
+      "loss": 0.3812,
+      "step": 9680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6465633952092755e-05,
+      "loss": 0.419,
+      "step": 9681
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646487357854881e-05,
+      "loss": 0.3237,
+      "step": 9682
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6464113140782337e-05,
+      "loss": 0.4174,
+      "step": 9683
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.64633526388009e-05,
+      "loss": 0.4324,
+      "step": 9684
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6462592072612044e-05,
+      "loss": 0.3821,
+      "step": 9685
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6461831442223332e-05,
+      "loss": 0.3462,
+      "step": 9686
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646107074764232e-05,
+      "loss": 0.4654,
+      "step": 9687
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646030998887656e-05,
+      "loss": 0.3833,
+      "step": 9688
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645954916593361e-05,
+      "loss": 0.3832,
+      "step": 9689
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6458788278821033e-05,
+      "loss": 0.436,
+      "step": 9690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645802732754638e-05,
+      "loss": 0.3444,
+      "step": 9691
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645726631211722e-05,
+      "loss": 0.3928,
+      "step": 9692
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6456505232541106e-05,
+      "loss": 0.3372,
+      "step": 9693
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6455744088825604e-05,
+      "loss": 0.4366,
+      "step": 9694
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645498288097827e-05,
+      "loss": 0.3817,
+      "step": 9695
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6454221609006663e-05,
+      "loss": 0.3775,
+      "step": 9696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6453460272918357e-05,
+      "loss": 0.4454,
+      "step": 9697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6452698872720902e-05,
+      "loss": 0.4475,
+      "step": 9698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6451937408421874e-05,
+      "loss": 0.37,
+      "step": 9699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645117588002883e-05,
+      "loss": 0.4211,
+      "step": 9700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6450414287549335e-05,
+      "loss": 0.3343,
+      "step": 9701
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6449652630990958e-05,
+      "loss": 0.4378,
+      "step": 9702
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644889091036126e-05,
+      "loss": 0.3865,
+      "step": 9703
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644812912566782e-05,
+      "loss": 0.402,
+      "step": 9704
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644736727691819e-05,
+      "loss": 0.3417,
+      "step": 9705
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6446605364119948e-05,
+      "loss": 0.4006,
+      "step": 9706
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6445843387280664e-05,
+      "loss": 0.411,
+      "step": 9707
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6445081346407897e-05,
+      "loss": 0.4011,
+      "step": 9708
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644431924150923e-05,
+      "loss": 0.4644,
+      "step": 9709
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6443557072592227e-05,
+      "loss": 0.3919,
+      "step": 9710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6442794839664463e-05,
+      "loss": 0.349,
+      "step": 9711
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6442032542733502e-05,
+      "loss": 0.3461,
+      "step": 9712
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6441270181806924e-05,
+      "loss": 0.3743,
+      "step": 9713
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6440507756892303e-05,
+      "loss": 0.3846,
+      "step": 9714
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.643974526799721e-05,
+      "loss": 0.4066,
+      "step": 9715
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6438982715129223e-05,
+      "loss": 0.4188,
+      "step": 9716
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6438220098295906e-05,
+      "loss": 0.4007,
+      "step": 9717
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6437457417504853e-05,
+      "loss": 0.4556,
+      "step": 9718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6436694672763625e-05,
+      "loss": 0.3405,
+      "step": 9719
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6435931864079805e-05,
+      "loss": 0.3973,
+      "step": 9720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6435168991460976e-05,
+      "loss": 0.4234,
+      "step": 9721
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6434406054914707e-05,
+      "loss": 0.3613,
+      "step": 9722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643364305444858e-05,
+      "loss": 0.4547,
+      "step": 9723
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643287999007018e-05,
+      "loss": 0.4063,
+      "step": 9724
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643211686178708e-05,
+      "loss": 0.3847,
+      "step": 9725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6431353669606868e-05,
+      "loss": 0.3532,
+      "step": 9726
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6430590413537122e-05,
+      "loss": 0.4701,
+      "step": 9727
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6429827093585425e-05,
+      "loss": 0.4364,
+      "step": 9728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6429063709759353e-05,
+      "loss": 0.3949,
+      "step": 9729
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.64283002620665e-05,
+      "loss": 0.4944,
+      "step": 9730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642753675051445e-05,
+      "loss": 0.3336,
+      "step": 9731
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6426773175110776e-05,
+      "loss": 0.3616,
+      "step": 9732
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6426009535863078e-05,
+      "loss": 0.4155,
+      "step": 9733
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642524583277893e-05,
+      "loss": 0.4017,
+      "step": 9734
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6424482065865925e-05,
+      "loss": 0.3598,
+      "step": 9735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642371823513165e-05,
+      "loss": 0.4771,
+      "step": 9736
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642295434058369e-05,
+      "loss": 0.3812,
+      "step": 9737
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6422190382229638e-05,
+      "loss": 0.3558,
+      "step": 9738
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6421426360077077e-05,
+      "loss": 0.3757,
+      "step": 9739
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6420662274133604e-05,
+      "loss": 0.353,
+      "step": 9740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641989812440681e-05,
+      "loss": 0.3375,
+      "step": 9741
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6419133910904273e-05,
+      "loss": 0.4603,
+      "step": 9742
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6418369633633598e-05,
+      "loss": 0.4028,
+      "step": 9743
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6417605292602375e-05,
+      "loss": 0.3878,
+      "step": 9744
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6416840887818195e-05,
+      "loss": 0.366,
+      "step": 9745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6416076419288652e-05,
+      "loss": 0.391,
+      "step": 9746
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641531188702134e-05,
+      "loss": 0.417,
+      "step": 9747
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6414547291023857e-05,
+      "loss": 0.4252,
+      "step": 9748
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641378263130379e-05,
+      "loss": 0.4059,
+      "step": 9749
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6413017907868748e-05,
+      "loss": 0.3832,
+      "step": 9750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6412253120726315e-05,
+      "loss": 0.3342,
+      "step": 9751
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6411488269884098e-05,
+      "loss": 0.3368,
+      "step": 9752
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6410723355349686e-05,
+      "loss": 0.3447,
+      "step": 9753
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6409958377130685e-05,
+      "loss": 0.3531,
+      "step": 9754
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6409193335234698e-05,
+      "loss": 0.3943,
+      "step": 9755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6408428229669315e-05,
+      "loss": 0.4324,
+      "step": 9756
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6407663060442144e-05,
+      "loss": 0.3484,
+      "step": 9757
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640689782756078e-05,
+      "loss": 0.4085,
+      "step": 9758
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6406132531032825e-05,
+      "loss": 0.331,
+      "step": 9759
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6405367170865892e-05,
+      "loss": 0.2318,
+      "step": 9760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6404601747067574e-05,
+      "loss": 0.4202,
+      "step": 9761
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6403836259645477e-05,
+      "loss": 0.3896,
+      "step": 9762
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6403070708607205e-05,
+      "loss": 0.4071,
+      "step": 9763
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6402305093960366e-05,
+      "loss": 0.3458,
+      "step": 9764
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640153941571256e-05,
+      "loss": 0.4048,
+      "step": 9765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6400773673871403e-05,
+      "loss": 0.4615,
+      "step": 9766
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640000786844449e-05,
+      "loss": 0.4504,
+      "step": 9767
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639924199943944e-05,
+      "loss": 0.4173,
+      "step": 9768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6398476066863852e-05,
+      "loss": 0.4259,
+      "step": 9769
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6397710070725338e-05,
+      "loss": 0.341,
+      "step": 9770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639694401103151e-05,
+      "loss": 0.4279,
+      "step": 9771
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6396177887789978e-05,
+      "loss": 0.3687,
+      "step": 9772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639541170100835e-05,
+      "loss": 0.31,
+      "step": 9773
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6394645450694234e-05,
+      "loss": 0.3855,
+      "step": 9774
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6393879136855247e-05,
+      "loss": 0.4735,
+      "step": 9775
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6393112759499007e-05,
+      "loss": 0.4259,
+      "step": 9776
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6392346318633115e-05,
+      "loss": 0.3986,
+      "step": 9777
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6391579814265193e-05,
+      "loss": 0.3952,
+      "step": 9778
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639081324640285e-05,
+      "loss": 0.3635,
+      "step": 9779
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639004661505371e-05,
+      "loss": 0.3378,
+      "step": 9780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638927992022538e-05,
+      "loss": 0.4097,
+      "step": 9781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6388513161925482e-05,
+      "loss": 0.4633,
+      "step": 9782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6387746340161626e-05,
+      "loss": 0.4341,
+      "step": 9783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638697945494144e-05,
+      "loss": 0.3235,
+      "step": 9784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6386212506272533e-05,
+      "loss": 0.3398,
+      "step": 9785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6385445494162532e-05,
+      "loss": 0.3956,
+      "step": 9786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638467841861905e-05,
+      "loss": 0.3622,
+      "step": 9787
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6383911279649707e-05,
+      "loss": 0.3499,
+      "step": 9788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6383144077262127e-05,
+      "loss": 0.524,
+      "step": 9789
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6382376811463935e-05,
+      "loss": 0.3641,
+      "step": 9790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6381609482262742e-05,
+      "loss": 0.4275,
+      "step": 9791
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6380842089666182e-05,
+      "loss": 0.4582,
+      "step": 9792
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638007463368187e-05,
+      "loss": 0.3839,
+      "step": 9793
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6379307114317437e-05,
+      "loss": 0.3888,
+      "step": 9794
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6378539531580502e-05,
+      "loss": 0.4184,
+      "step": 9795
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6377771885478697e-05,
+      "loss": 0.4019,
+      "step": 9796
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637700417601964e-05,
+      "loss": 0.424,
+      "step": 9797
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6376236403210962e-05,
+      "loss": 0.4093,
+      "step": 9798
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637546856706029e-05,
+      "loss": 0.4096,
+      "step": 9799
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6374700667575246e-05,
+      "loss": 0.4082,
+      "step": 9800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6373932704763467e-05,
+      "loss": 0.4696,
+      "step": 9801
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6373164678632575e-05,
+      "loss": 0.4467,
+      "step": 9802
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6372396589190208e-05,
+      "loss": 0.3859,
+      "step": 9803
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6371628436443986e-05,
+      "loss": 0.3368,
+      "step": 9804
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6370860220401547e-05,
+      "loss": 0.4073,
+      "step": 9805
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6370091941070518e-05,
+      "loss": 0.3781,
+      "step": 9806
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6369323598458534e-05,
+      "loss": 0.4442,
+      "step": 9807
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6368555192573233e-05,
+      "loss": 0.3408,
+      "step": 9808
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6367786723422235e-05,
+      "loss": 0.3874,
+      "step": 9809
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6367018191013187e-05,
+      "loss": 0.4293,
+      "step": 9810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6366249595353716e-05,
+      "loss": 0.4274,
+      "step": 9811
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6365480936451457e-05,
+      "loss": 0.4824,
+      "step": 9812
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6364712214314056e-05,
+      "loss": 0.3791,
+      "step": 9813
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6363943428949135e-05,
+      "loss": 0.3457,
+      "step": 9814
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636317458036434e-05,
+      "loss": 0.4332,
+      "step": 9815
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636240566856731e-05,
+      "loss": 0.3896,
+      "step": 9816
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636163669356568e-05,
+      "loss": 0.4492,
+      "step": 9817
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636086765536709e-05,
+      "loss": 0.4061,
+      "step": 9818
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6360098553979176e-05,
+      "loss": 0.3396,
+      "step": 9819
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6359329389409584e-05,
+      "loss": 0.3313,
+      "step": 9820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635856016166595e-05,
+      "loss": 0.4765,
+      "step": 9821
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635779087075592e-05,
+      "loss": 0.3926,
+      "step": 9822
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6357021516687133e-05,
+      "loss": 0.3811,
+      "step": 9823
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6356252099467235e-05,
+      "loss": 0.3397,
+      "step": 9824
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6355482619103866e-05,
+      "loss": 0.4166,
+      "step": 9825
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635471307560467e-05,
+      "loss": 0.3488,
+      "step": 9826
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6353943468977295e-05,
+      "loss": 0.4157,
+      "step": 9827
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6353173799229387e-05,
+      "loss": 0.4165,
+      "step": 9828
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6352404066368586e-05,
+      "loss": 0.3337,
+      "step": 9829
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6351634270402543e-05,
+      "loss": 0.3933,
+      "step": 9830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6350864411338906e-05,
+      "loss": 0.3709,
+      "step": 9831
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635009448918532e-05,
+      "loss": 0.4131,
+      "step": 9832
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6349324503949433e-05,
+      "loss": 0.4109,
+      "step": 9833
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63485544556389e-05,
+      "loss": 0.3957,
+      "step": 9834
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6347784344261365e-05,
+      "loss": 0.4007,
+      "step": 9835
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6347014169824474e-05,
+      "loss": 0.3384,
+      "step": 9836
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634624393233589e-05,
+      "loss": 0.342,
+      "step": 9837
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6345473631803255e-05,
+      "loss": 0.3877,
+      "step": 9838
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6344703268234227e-05,
+      "loss": 0.4323,
+      "step": 9839
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 0.3118,
+      "step": 9840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6343162352017595e-05,
+      "loss": 0.3877,
+      "step": 9841
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63423917993853e-05,
+      "loss": 0.3765,
+      "step": 9842
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6341621183747223e-05,
+      "loss": 0.448,
+      "step": 9843
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6340850505111023e-05,
+      "loss": 0.3932,
+      "step": 9844
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6340079763484357e-05,
+      "loss": 0.4205,
+      "step": 9845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6339308958874872e-05,
+      "loss": 0.4343,
+      "step": 9846
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6338538091290236e-05,
+      "loss": 0.3958,
+      "step": 9847
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.63377671607381e-05,
+      "loss": 0.4028,
+      "step": 9848
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633699616722613e-05,
+      "loss": 0.415,
+      "step": 9849
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6336225110761975e-05,
+      "loss": 0.3277,
+      "step": 9850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6335453991353305e-05,
+      "loss": 0.401,
+      "step": 9851
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633468280900777e-05,
+      "loss": 0.4085,
+      "step": 9852
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6333911563733037e-05,
+      "loss": 0.3617,
+      "step": 9853
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6333140255536773e-05,
+      "loss": 0.4594,
+      "step": 9854
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633236888442663e-05,
+      "loss": 0.3949,
+      "step": 9855
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6331597450410274e-05,
+      "loss": 0.3137,
+      "step": 9856
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633082595349537e-05,
+      "loss": 0.3727,
+      "step": 9857
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6330054393689583e-05,
+      "loss": 0.4106,
+      "step": 9858
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632928277100058e-05,
+      "loss": 0.373,
+      "step": 9859
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6328511085436013e-05,
+      "loss": 0.4192,
+      "step": 9860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6327739337003564e-05,
+      "loss": 0.3686,
+      "step": 9861
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6326967525710895e-05,
+      "loss": 0.3762,
+      "step": 9862
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632619565156567e-05,
+      "loss": 0.4626,
+      "step": 9863
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632542371457556e-05,
+      "loss": 0.4119,
+      "step": 9864
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6324651714748226e-05,
+      "loss": 0.4495,
+      "step": 9865
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632387965209135e-05,
+      "loss": 0.3711,
+      "step": 9866
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6323107526612593e-05,
+      "loss": 0.3553,
+      "step": 9867
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6322335338319623e-05,
+      "loss": 0.4014,
+      "step": 9868
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632156308722012e-05,
+      "loss": 0.4313,
+      "step": 9869
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6320790773321745e-05,
+      "loss": 0.4364,
+      "step": 9870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632001839663218e-05,
+      "loss": 0.4058,
+      "step": 9871
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6319245957159093e-05,
+      "loss": 0.3642,
+      "step": 9872
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631847345491016e-05,
+      "loss": 0.3465,
+      "step": 9873
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631770088989305e-05,
+      "loss": 0.4217,
+      "step": 9874
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6316928262115443e-05,
+      "loss": 0.4138,
+      "step": 9875
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6316155571585014e-05,
+      "loss": 0.4889,
+      "step": 9876
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631538281830943e-05,
+      "loss": 0.4159,
+      "step": 9877
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6314610002296383e-05,
+      "loss": 0.3999,
+      "step": 9878
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631383712355354e-05,
+      "loss": 0.3839,
+      "step": 9879
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631306418208858e-05,
+      "loss": 0.4405,
+      "step": 9880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6312291177909182e-05,
+      "loss": 0.3483,
+      "step": 9881
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6311518111023025e-05,
+      "loss": 0.33,
+      "step": 9882
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631074498143779e-05,
+      "loss": 0.4507,
+      "step": 9883
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6309971789161158e-05,
+      "loss": 0.4221,
+      "step": 9884
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6309198534200807e-05,
+      "loss": 0.4937,
+      "step": 9885
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630842521656442e-05,
+      "loss": 0.3598,
+      "step": 9886
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6307651836259677e-05,
+      "loss": 0.4093,
+      "step": 9887
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6306878393294268e-05,
+      "loss": 0.3735,
+      "step": 9888
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630610488767587e-05,
+      "loss": 0.4864,
+      "step": 9889
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6305331319412166e-05,
+      "loss": 0.4221,
+      "step": 9890
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6304557688510845e-05,
+      "loss": 0.4034,
+      "step": 9891
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6303783994979593e-05,
+      "loss": 0.4214,
+      "step": 9892
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6303010238826086e-05,
+      "loss": 0.2989,
+      "step": 9893
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6302236420058024e-05,
+      "loss": 0.3253,
+      "step": 9894
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6301462538683087e-05,
+      "loss": 0.4,
+      "step": 9895
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6300688594708963e-05,
+      "loss": 0.5056,
+      "step": 9896
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6299914588143345e-05,
+      "loss": 0.3631,
+      "step": 9897
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6299140518993913e-05,
+      "loss": 0.3181,
+      "step": 9898
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6298366387268366e-05,
+      "loss": 0.4175,
+      "step": 9899
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629759219297439e-05,
+      "loss": 0.3992,
+      "step": 9900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6296817936119674e-05,
+      "loss": 0.3958,
+      "step": 9901
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6296043616711914e-05,
+      "loss": 0.4754,
+      "step": 9902
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6295269234758798e-05,
+      "loss": 0.418,
+      "step": 9903
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629449479026802e-05,
+      "loss": 0.4592,
+      "step": 9904
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6293720283247276e-05,
+      "loss": 0.3632,
+      "step": 9905
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6292945713704257e-05,
+      "loss": 0.325,
+      "step": 9906
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629217108164666e-05,
+      "loss": 0.3824,
+      "step": 9907
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629139638708218e-05,
+      "loss": 0.3874,
+      "step": 9908
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6290621630018512e-05,
+      "loss": 0.4669,
+      "step": 9909
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.628984681046335e-05,
+      "loss": 0.3838,
+      "step": 9910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6289071928424394e-05,
+      "loss": 0.4034,
+      "step": 9911
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628829698390934e-05,
+      "loss": 0.3835,
+      "step": 9912
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628752197692589e-05,
+      "loss": 0.4103,
+      "step": 9913
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6286746907481742e-05,
+      "loss": 0.3796,
+      "step": 9914
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6285971775584587e-05,
+      "loss": 0.4,
+      "step": 9915
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6285196581242137e-05,
+      "loss": 0.3804,
+      "step": 9916
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628442132446209e-05,
+      "loss": 0.3582,
+      "step": 9917
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6283646005252147e-05,
+      "loss": 0.3973,
+      "step": 9918
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6282870623620006e-05,
+      "loss": 0.4115,
+      "step": 9919
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6282095179573368e-05,
+      "loss": 0.3377,
+      "step": 9920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6281319673119947e-05,
+      "loss": 0.3672,
+      "step": 9921
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6280544104267436e-05,
+      "loss": 0.3312,
+      "step": 9922
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6279768473023546e-05,
+      "loss": 0.4278,
+      "step": 9923
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6278992779395983e-05,
+      "loss": 0.4486,
+      "step": 9924
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6278217023392447e-05,
+      "loss": 0.3687,
+      "step": 9925
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627744120502065e-05,
+      "loss": 0.3666,
+      "step": 9926
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6276665324288292e-05,
+      "loss": 0.3898,
+      "step": 9927
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627588938120309e-05,
+      "loss": 0.4038,
+      "step": 9928
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6275113375772745e-05,
+      "loss": 0.4353,
+      "step": 9929
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627433730800497e-05,
+      "loss": 0.4069,
+      "step": 9930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6273561177907473e-05,
+      "loss": 0.3816,
+      "step": 9931
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6272784985487964e-05,
+      "loss": 0.4147,
+      "step": 9932
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6272008730754153e-05,
+      "loss": 0.3365,
+      "step": 9933
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6271232413713756e-05,
+      "loss": 0.3871,
+      "step": 9934
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6270456034374477e-05,
+      "loss": 0.3853,
+      "step": 9935
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626967959274403e-05,
+      "loss": 0.416,
+      "step": 9936
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626890308883014e-05,
+      "loss": 0.4101,
+      "step": 9937
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6268126522640506e-05,
+      "loss": 0.3946,
+      "step": 9938
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6267349894182846e-05,
+      "loss": 0.4383,
+      "step": 9939
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626657320346488e-05,
+      "loss": 0.3187,
+      "step": 9940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6265796450494326e-05,
+      "loss": 0.4128,
+      "step": 9941
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6265019635278888e-05,
+      "loss": 0.3571,
+      "step": 9942
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6264242757826298e-05,
+      "loss": 0.3675,
+      "step": 9943
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626346581814426e-05,
+      "loss": 0.4974,
+      "step": 9944
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.62626888162405e-05,
+      "loss": 0.3569,
+      "step": 9945
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6261911752122734e-05,
+      "loss": 0.441,
+      "step": 9946
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6261134625798684e-05,
+      "loss": 0.4145,
+      "step": 9947
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6260357437276067e-05,
+      "loss": 0.3445,
+      "step": 9948
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6259580186562606e-05,
+      "loss": 0.4473,
+      "step": 9949
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625880287366602e-05,
+      "loss": 0.3984,
+      "step": 9950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6258025498594033e-05,
+      "loss": 0.4612,
+      "step": 9951
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6257248061354366e-05,
+      "loss": 0.3829,
+      "step": 9952
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6256470561954746e-05,
+      "loss": 0.3412,
+      "step": 9953
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6255693000402888e-05,
+      "loss": 0.3864,
+      "step": 9954
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6254915376706526e-05,
+      "loss": 0.4086,
+      "step": 9955
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625413769087338e-05,
+      "loss": 0.3795,
+      "step": 9956
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6253359942911172e-05,
+      "loss": 0.465,
+      "step": 9957
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6252582132827643e-05,
+      "loss": 0.3787,
+      "step": 9958
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6251804260630503e-05,
+      "loss": 0.438,
+      "step": 9959
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6251026326327485e-05,
+      "loss": 0.3998,
+      "step": 9960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6250248329926316e-05,
+      "loss": 0.4239,
+      "step": 9961
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6249470271434732e-05,
+      "loss": 0.3755,
+      "step": 9962
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6248692150860455e-05,
+      "loss": 0.3111,
+      "step": 9963
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6247913968211215e-05,
+      "loss": 0.3621,
+      "step": 9964
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624713572349475e-05,
+      "loss": 0.3736,
+      "step": 9965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624635741671878e-05,
+      "loss": 0.3601,
+      "step": 9966
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624557904789104e-05,
+      "loss": 0.3244,
+      "step": 9967
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624480061701927e-05,
+      "loss": 0.4352,
+      "step": 9968
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6244022124111193e-05,
+      "loss": 0.4129,
+      "step": 9969
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6243243569174555e-05,
+      "loss": 0.2835,
+      "step": 9970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6242464952217076e-05,
+      "loss": 0.3768,
+      "step": 9971
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.62416862732465e-05,
+      "loss": 0.3757,
+      "step": 9972
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6240907532270555e-05,
+      "loss": 0.4228,
+      "step": 9973
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6240128729296988e-05,
+      "loss": 0.4103,
+      "step": 9974
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6239349864333527e-05,
+      "loss": 0.3456,
+      "step": 9975
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623857093738791e-05,
+      "loss": 0.3857,
+      "step": 9976
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623779194846788e-05,
+      "loss": 0.4156,
+      "step": 9977
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6237012897581166e-05,
+      "loss": 0.3725,
+      "step": 9978
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623623378473552e-05,
+      "loss": 0.4495,
+      "step": 9979
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6235454609938676e-05,
+      "loss": 0.3061,
+      "step": 9980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6234675373198367e-05,
+      "loss": 0.3997,
+      "step": 9981
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6233896074522345e-05,
+      "loss": 0.5067,
+      "step": 9982
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6233116713918348e-05,
+      "loss": 0.2904,
+      "step": 9983
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6232337291394115e-05,
+      "loss": 0.3757,
+      "step": 9984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623155780695739e-05,
+      "loss": 0.386,
+      "step": 9985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623077826061592e-05,
+      "loss": 0.4064,
+      "step": 9986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6229998652377447e-05,
+      "loss": 0.4155,
+      "step": 9987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6229218982249712e-05,
+      "loss": 0.3436,
+      "step": 9988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6228439250240467e-05,
+      "loss": 0.4217,
+      "step": 9989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6227659456357454e-05,
+      "loss": 0.4491,
+      "step": 9990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622687960060842e-05,
+      "loss": 0.3232,
+      "step": 9991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622609968300111e-05,
+      "loss": 0.413,
+      "step": 9992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622531970354328e-05,
+      "loss": 0.3704,
+      "step": 9993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6224539662242668e-05,
+      "loss": 0.3604,
+      "step": 9994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6223759559107028e-05,
+      "loss": 0.3287,
+      "step": 9995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6222979394144108e-05,
+      "loss": 0.4642,
+      "step": 9996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622219916736166e-05,
+      "loss": 0.3697,
+      "step": 9997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6221418878767436e-05,
+      "loss": 0.3687,
+      "step": 9998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6220638528369184e-05,
+      "loss": 0.3251,
+      "step": 9999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6219858116174657e-05,
+      "loss": 0.3855,
+      "step": 10000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621907764219161e-05,
+      "loss": 0.4728,
+      "step": 10001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621829710642779e-05,
+      "loss": 0.341,
+      "step": 10002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6217516508890963e-05,
+      "loss": 0.4188,
+      "step": 10003
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621673584958887e-05,
+      "loss": 0.3718,
+      "step": 10004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6215955128529273e-05,
+      "loss": 0.3603,
+      "step": 10005
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6215174345719925e-05,
+      "loss": 0.3604,
+      "step": 10006
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6214393501168588e-05,
+      "loss": 0.3617,
+      "step": 10007
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6213612594883016e-05,
+      "loss": 0.4562,
+      "step": 10008
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621283162687096e-05,
+      "loss": 0.467,
+      "step": 10009
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6212050597140186e-05,
+      "loss": 0.3625,
+      "step": 10010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6211269505698453e-05,
+      "loss": 0.4134,
+      "step": 10011
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6210488352553518e-05,
+      "loss": 0.4326,
+      "step": 10012
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620970713771314e-05,
+      "loss": 0.4342,
+      "step": 10013
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620892586118508e-05,
+      "loss": 0.4247,
+      "step": 10014
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.62081445229771e-05,
+      "loss": 0.3421,
+      "step": 10015
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6207363123096963e-05,
+      "loss": 0.4054,
+      "step": 10016
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6206581661552433e-05,
+      "loss": 0.3501,
+      "step": 10017
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6205800138351265e-05,
+      "loss": 0.357,
+      "step": 10018
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6205018553501232e-05,
+      "loss": 0.3658,
+      "step": 10019
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6204236907010095e-05,
+      "loss": 0.3249,
+      "step": 10020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6203455198885615e-05,
+      "loss": 0.4019,
+      "step": 10021
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6202673429135564e-05,
+      "loss": 0.3318,
+      "step": 10022
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6201891597767702e-05,
+      "loss": 0.4104,
+      "step": 10023
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6201109704789797e-05,
+      "loss": 0.4526,
+      "step": 10024
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620032775020963e-05,
+      "loss": 0.3021,
+      "step": 10025
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6199545734034943e-05,
+      "loss": 0.4751,
+      "step": 10026
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6198763656273528e-05,
+      "loss": 0.4231,
+      "step": 10027
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6197981516933143e-05,
+      "loss": 0.3199,
+      "step": 10028
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619719931602156e-05,
+      "loss": 0.3506,
+      "step": 10029
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619641705354655e-05,
+      "loss": 0.3188,
+      "step": 10030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619563472951588e-05,
+      "loss": 0.4676,
+      "step": 10031
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6194852343937327e-05,
+      "loss": 0.389,
+      "step": 10032
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619406989681866e-05,
+      "loss": 0.3556,
+      "step": 10033
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6193287388167655e-05,
+      "loss": 0.327,
+      "step": 10034
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6192504817992086e-05,
+      "loss": 0.4077,
+      "step": 10035
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6191722186299723e-05,
+      "loss": 0.4202,
+      "step": 10036
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6190939493098344e-05,
+      "loss": 0.4071,
+      "step": 10037
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619015673839572e-05,
+      "loss": 0.3859,
+      "step": 10038
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.618937392219963e-05,
+      "loss": 0.3925,
+      "step": 10039
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.618859104451785e-05,
+      "loss": 0.3904,
+      "step": 10040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618780810535816e-05,
+      "loss": 0.3792,
+      "step": 10041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6187025104728334e-05,
+      "loss": 0.4385,
+      "step": 10042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618624204263615e-05,
+      "loss": 0.4166,
+      "step": 10043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618545891908939e-05,
+      "loss": 0.4469,
+      "step": 10044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6184675734095838e-05,
+      "loss": 0.3947,
+      "step": 10045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6183892487663262e-05,
+      "loss": 0.3751,
+      "step": 10046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6183109179799452e-05,
+      "loss": 0.3287,
+      "step": 10047
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6182325810512184e-05,
+      "loss": 0.4747,
+      "step": 10048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6181542379809244e-05,
+      "loss": 0.3851,
+      "step": 10049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6180758887698414e-05,
+      "loss": 0.3572,
+      "step": 10050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617997533418748e-05,
+      "loss": 0.4316,
+      "step": 10051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617919171928422e-05,
+      "loss": 0.3936,
+      "step": 10052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617840804299642e-05,
+      "loss": 0.4634,
+      "step": 10053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617762430533187e-05,
+      "loss": 0.3328,
+      "step": 10054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6176840506298345e-05,
+      "loss": 0.3788,
+      "step": 10055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6176056645903645e-05,
+      "loss": 0.3947,
+      "step": 10056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6175272724155545e-05,
+      "loss": 0.3837,
+      "step": 10057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617448874106184e-05,
+      "loss": 0.4373,
+      "step": 10058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617370469663032e-05,
+      "loss": 0.4088,
+      "step": 10059
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6172920590868765e-05,
+      "loss": 0.4031,
+      "step": 10060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617213642378497e-05,
+      "loss": 0.5458,
+      "step": 10061
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6171352195386724e-05,
+      "loss": 0.4094,
+      "step": 10062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6170567905681814e-05,
+      "loss": 0.4004,
+      "step": 10063
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616978355467804e-05,
+      "loss": 0.4317,
+      "step": 10064
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6168999142383187e-05,
+      "loss": 0.3406,
+      "step": 10065
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616821466880505e-05,
+      "loss": 0.4062,
+      "step": 10066
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616743013395142e-05,
+      "loss": 0.4275,
+      "step": 10067
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616664553783009e-05,
+      "loss": 0.3849,
+      "step": 10068
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6165860880448856e-05,
+      "loss": 0.4324,
+      "step": 10069
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6165076161815514e-05,
+      "loss": 0.3521,
+      "step": 10070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616429138193786e-05,
+      "loss": 0.3704,
+      "step": 10071
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6163506540823684e-05,
+      "loss": 0.3866,
+      "step": 10072
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6162721638480792e-05,
+      "loss": 0.4235,
+      "step": 10073
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616193667491697e-05,
+      "loss": 0.3631,
+      "step": 10074
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6161151650140027e-05,
+      "loss": 0.3887,
+      "step": 10075
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160366564157755e-05,
+      "loss": 0.2902,
+      "step": 10076
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6159581416977954e-05,
+      "loss": 0.3891,
+      "step": 10077
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6158796208608424e-05,
+      "loss": 0.4102,
+      "step": 10078
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6158010939056966e-05,
+      "loss": 0.3713,
+      "step": 10079
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615722560833138e-05,
+      "loss": 0.3603,
+      "step": 10080
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615644021643947e-05,
+      "loss": 0.3745,
+      "step": 10081
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6155654763389036e-05,
+      "loss": 0.3695,
+      "step": 10082
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615486924918788e-05,
+      "loss": 0.3939,
+      "step": 10083
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6154083673843802e-05,
+      "loss": 0.4089,
+      "step": 10084
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6153298037364617e-05,
+      "loss": 0.3255,
+      "step": 10085
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6152512339758123e-05,
+      "loss": 0.3203,
+      "step": 10086
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6151726581032117e-05,
+      "loss": 0.4742,
+      "step": 10087
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6150940761194423e-05,
+      "loss": 0.4517,
+      "step": 10088
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615015488025283e-05,
+      "loss": 0.4311,
+      "step": 10089
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6149368938215157e-05,
+      "loss": 0.4749,
+      "step": 10090
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6148582935089206e-05,
+      "loss": 0.4308,
+      "step": 10091
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6147796870882783e-05,
+      "loss": 0.4295,
+      "step": 10092
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6147010745603703e-05,
+      "loss": 0.3404,
+      "step": 10093
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6146224559259772e-05,
+      "loss": 0.4411,
+      "step": 10094
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61454383118588e-05,
+      "loss": 0.3588,
+      "step": 10095
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6144652003408596e-05,
+      "loss": 0.3892,
+      "step": 10096
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614386563391697e-05,
+      "loss": 0.4,
+      "step": 10097
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6143079203391744e-05,
+      "loss": 0.4296,
+      "step": 10098
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614229271184072e-05,
+      "loss": 0.357,
+      "step": 10099
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614150615927172e-05,
+      "loss": 0.4112,
+      "step": 10100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6140719545692545e-05,
+      "loss": 0.3691,
+      "step": 10101
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6139932871111023e-05,
+      "loss": 0.3827,
+      "step": 10102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6139146135534957e-05,
+      "loss": 0.38,
+      "step": 10103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.613835933897217e-05,
+      "loss": 0.4373,
+      "step": 10104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6137572481430474e-05,
+      "loss": 0.4082,
+      "step": 10105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613678556291769e-05,
+      "loss": 0.4067,
+      "step": 10106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613599858344163e-05,
+      "loss": 0.3327,
+      "step": 10107
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613521154301012e-05,
+      "loss": 0.3882,
+      "step": 10108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613442444163097e-05,
+      "loss": 0.3978,
+      "step": 10109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6133637279312003e-05,
+      "loss": 0.3963,
+      "step": 10110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613285005606104e-05,
+      "loss": 0.3849,
+      "step": 10111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61320627718859e-05,
+      "loss": 0.4145,
+      "step": 10112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6131275426794397e-05,
+      "loss": 0.3959,
+      "step": 10113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6130488020794362e-05,
+      "loss": 0.4141,
+      "step": 10114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612970055389362e-05,
+      "loss": 0.4233,
+      "step": 10115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6128913026099985e-05,
+      "loss": 0.4212,
+      "step": 10116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6128125437421282e-05,
+      "loss": 0.4531,
+      "step": 10117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6127337787865336e-05,
+      "loss": 0.312,
+      "step": 10118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6126550077439973e-05,
+      "loss": 0.424,
+      "step": 10119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6125762306153015e-05,
+      "loss": 0.4242,
+      "step": 10120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6124974474012293e-05,
+      "loss": 0.3804,
+      "step": 10121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6124186581025627e-05,
+      "loss": 0.3837,
+      "step": 10122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6123398627200847e-05,
+      "loss": 0.4204,
+      "step": 10123
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6122610612545785e-05,
+      "loss": 0.4082,
+      "step": 10124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6121822537068263e-05,
+      "loss": 0.3894,
+      "step": 10125
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612103440077611e-05,
+      "loss": 0.4046,
+      "step": 10126
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6120246203677157e-05,
+      "loss": 0.3181,
+      "step": 10127
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6119457945779238e-05,
+      "loss": 0.4673,
+      "step": 10128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6118669627090177e-05,
+      "loss": 0.3818,
+      "step": 10129
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611788124761781e-05,
+      "loss": 0.4398,
+      "step": 10130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6117092807369962e-05,
+      "loss": 0.3752,
+      "step": 10131
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6116304306354474e-05,
+      "loss": 0.3594,
+      "step": 10132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6115515744579175e-05,
+      "loss": 0.3507,
+      "step": 10133
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6114727122051902e-05,
+      "loss": 0.394,
+      "step": 10134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611393843878048e-05,
+      "loss": 0.3832,
+      "step": 10135
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6113149694772755e-05,
+      "loss": 0.4274,
+      "step": 10136
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6112360890036555e-05,
+      "loss": 0.4325,
+      "step": 10137
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6111572024579717e-05,
+      "loss": 0.4091,
+      "step": 10138
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6110783098410082e-05,
+      "loss": 0.3387,
+      "step": 10139
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6109994111535483e-05,
+      "loss": 0.4192,
+      "step": 10140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610920506396376e-05,
+      "loss": 0.3832,
+      "step": 10141
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6108415955702752e-05,
+      "loss": 0.3265,
+      "step": 10142
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6107626786760294e-05,
+      "loss": 0.4281,
+      "step": 10143
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610683755714423e-05,
+      "loss": 0.4617,
+      "step": 10144
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6106048266862395e-05,
+      "loss": 0.415,
+      "step": 10145
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6105258915922636e-05,
+      "loss": 0.4687,
+      "step": 10146
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6104469504332795e-05,
+      "loss": 0.3786,
+      "step": 10147
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6103680032100707e-05,
+      "loss": 0.3808,
+      "step": 10148
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610289049923422e-05,
+      "loss": 0.3699,
+      "step": 10149
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6102100905741176e-05,
+      "loss": 0.3763,
+      "step": 10150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610131125162942e-05,
+      "loss": 0.3584,
+      "step": 10151
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6100521536906796e-05,
+      "loss": 0.432,
+      "step": 10152
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6099731761581146e-05,
+      "loss": 0.3735,
+      "step": 10153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6098941925660317e-05,
+      "loss": 0.3664,
+      "step": 10154
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609815202915216e-05,
+      "loss": 0.4106,
+      "step": 10155
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609736207206452e-05,
+      "loss": 0.4253,
+      "step": 10156
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6096572054405236e-05,
+      "loss": 0.3812,
+      "step": 10157
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609578197618217e-05,
+      "loss": 0.3979,
+      "step": 10158
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6094991837403155e-05,
+      "loss": 0.3569,
+      "step": 10159
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609420163807606e-05,
+      "loss": 0.3417,
+      "step": 10160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6093411378208713e-05,
+      "loss": 0.4792,
+      "step": 10161
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6092621057808982e-05,
+      "loss": 0.387,
+      "step": 10162
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609183067688471e-05,
+      "loss": 0.4378,
+      "step": 10163
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609104023544375e-05,
+      "loss": 0.4219,
+      "step": 10164
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6090249733493952e-05,
+      "loss": 0.4282,
+      "step": 10165
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6089459171043173e-05,
+      "loss": 0.499,
+      "step": 10166
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6088668548099265e-05,
+      "loss": 0.4367,
+      "step": 10167
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6087877864670083e-05,
+      "loss": 0.3581,
+      "step": 10168
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.608708712076348e-05,
+      "loss": 0.3751,
+      "step": 10169
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6086296316387312e-05,
+      "loss": 0.3954,
+      "step": 10170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6085505451549436e-05,
+      "loss": 0.3742,
+      "step": 10171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6084714526257705e-05,
+      "loss": 0.4177,
+      "step": 10172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608392354051998e-05,
+      "loss": 0.4281,
+      "step": 10173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6083132494344114e-05,
+      "loss": 0.3871,
+      "step": 10174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608234138773797e-05,
+      "loss": 0.363,
+      "step": 10175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608155022070941e-05,
+      "loss": 0.4119,
+      "step": 10176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6080758993266283e-05,
+      "loss": 0.3409,
+      "step": 10177
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6079967705416457e-05,
+      "loss": 0.3461,
+      "step": 10178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607917635716779e-05,
+      "loss": 0.4215,
+      "step": 10179
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6078384948528146e-05,
+      "loss": 0.375,
+      "step": 10180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6077593479505383e-05,
+      "loss": 0.3809,
+      "step": 10181
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6076801950107368e-05,
+      "loss": 0.3979,
+      "step": 10182
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607601036034196e-05,
+      "loss": 0.4296,
+      "step": 10183
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6075218710217022e-05,
+      "loss": 0.3434,
+      "step": 10184
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6074426999740422e-05,
+      "loss": 0.3741,
+      "step": 10185
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6073635228920026e-05,
+      "loss": 0.4714,
+      "step": 10186
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6072843397763695e-05,
+      "loss": 0.4213,
+      "step": 10187
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6072051506279298e-05,
+      "loss": 0.5389,
+      "step": 10188
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.60712595544747e-05,
+      "loss": 0.3809,
+      "step": 10189
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6070467542357765e-05,
+      "loss": 0.3792,
+      "step": 10190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606967546993637e-05,
+      "loss": 0.4289,
+      "step": 10191
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606888333721838e-05,
+      "loss": 0.3773,
+      "step": 10192
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606809114421166e-05,
+      "loss": 0.4585,
+      "step": 10193
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6067298890924083e-05,
+      "loss": 0.3698,
+      "step": 10194
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6066506577363518e-05,
+      "loss": 0.3724,
+      "step": 10195
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6065714203537836e-05,
+      "loss": 0.4532,
+      "step": 10196
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606492176945491e-05,
+      "loss": 0.4064,
+      "step": 10197
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606412927512261e-05,
+      "loss": 0.4598,
+      "step": 10198
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606333672054881e-05,
+      "loss": 0.4255,
+      "step": 10199
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6062544105741385e-05,
+      "loss": 0.394,
+      "step": 10200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606175143070821e-05,
+      "loss": 0.3563,
+      "step": 10201
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606095869545715e-05,
+      "loss": 0.4574,
+      "step": 10202
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6060165899996088e-05,
+      "loss": 0.4033,
+      "step": 10203
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.60593730443329e-05,
+      "loss": 0.405,
+      "step": 10204
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6058580128475458e-05,
+      "loss": 0.2947,
+      "step": 10205
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6057787152431648e-05,
+      "loss": 0.4383,
+      "step": 10206
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6056994116209335e-05,
+      "loss": 0.3947,
+      "step": 10207
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6056201019816403e-05,
+      "loss": 0.3277,
+      "step": 10208
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605540786326073e-05,
+      "loss": 0.373,
+      "step": 10209
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6054614646550198e-05,
+      "loss": 0.4251,
+      "step": 10210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6053821369692686e-05,
+      "loss": 0.4314,
+      "step": 10211
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605302803269607e-05,
+      "loss": 0.413,
+      "step": 10212
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605223463556824e-05,
+      "loss": 0.4623,
+      "step": 10213
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605144117831707e-05,
+      "loss": 0.3658,
+      "step": 10214
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605064766095044e-05,
+      "loss": 0.4482,
+      "step": 10215
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604985408347624e-05,
+      "loss": 0.4155,
+      "step": 10216
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604906044590235e-05,
+      "loss": 0.3573,
+      "step": 10217
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6048266748236654e-05,
+      "loss": 0.3595,
+      "step": 10218
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6047472990487042e-05,
+      "loss": 0.3786,
+      "step": 10219
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6046679172661387e-05,
+      "loss": 0.3594,
+      "step": 10220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6045885294767588e-05,
+      "loss": 0.4397,
+      "step": 10221
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6045091356813525e-05,
+      "loss": 0.3689,
+      "step": 10222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6044297358807087e-05,
+      "loss": 0.3977,
+      "step": 10223
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604350330075616e-05,
+      "loss": 0.4105,
+      "step": 10224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6042709182668628e-05,
+      "loss": 0.4025,
+      "step": 10225
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604191500455239e-05,
+      "loss": 0.4213,
+      "step": 10226
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6041120766415325e-05,
+      "loss": 0.4545,
+      "step": 10227
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6040326468265333e-05,
+      "loss": 0.3254,
+      "step": 10228
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6039532110110297e-05,
+      "loss": 0.3582,
+      "step": 10229
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6038737691958112e-05,
+      "loss": 0.3527,
+      "step": 10230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6037943213816668e-05,
+      "loss": 0.3704,
+      "step": 10231
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603714867569386e-05,
+      "loss": 0.433,
+      "step": 10232
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6036354077597578e-05,
+      "loss": 0.4508,
+      "step": 10233
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6035559419535714e-05,
+      "loss": 0.3751,
+      "step": 10234
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603476470151617e-05,
+      "loss": 0.3357,
+      "step": 10235
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603396992354683e-05,
+      "loss": 0.3747,
+      "step": 10236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60331750856356e-05,
+      "loss": 0.4101,
+      "step": 10237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603238018779037e-05,
+      "loss": 0.3519,
+      "step": 10238
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6031585230019034e-05,
+      "loss": 0.3472,
+      "step": 10239
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6030790212329495e-05,
+      "loss": 0.4467,
+      "step": 10240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6029995134729647e-05,
+      "loss": 0.4007,
+      "step": 10241
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6029199997227392e-05,
+      "loss": 0.3407,
+      "step": 10242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602840479983063e-05,
+      "loss": 0.3549,
+      "step": 10243
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602760954254725e-05,
+      "loss": 0.4758,
+      "step": 10244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6026814225385163e-05,
+      "loss": 0.4265,
+      "step": 10245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6026018848352263e-05,
+      "loss": 0.3424,
+      "step": 10246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6025223411456462e-05,
+      "loss": 0.3718,
+      "step": 10247
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6024427914705648e-05,
+      "loss": 0.3379,
+      "step": 10248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6023632358107733e-05,
+      "loss": 0.4016,
+      "step": 10249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6022836741670617e-05,
+      "loss": 0.4351,
+      "step": 10250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6022041065402205e-05,
+      "loss": 0.3596,
+      "step": 10251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60212453293104e-05,
+      "loss": 0.3531,
+      "step": 10252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6020449533403102e-05,
+      "loss": 0.3229,
+      "step": 10253
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6019653677688225e-05,
+      "loss": 0.2728,
+      "step": 10254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6018857762173672e-05,
+      "loss": 0.3994,
+      "step": 10255
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6018061786867348e-05,
+      "loss": 0.3569,
+      "step": 10256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6017265751777164e-05,
+      "loss": 0.5378,
+      "step": 10257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6016469656911022e-05,
+      "loss": 0.3559,
+      "step": 10258
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6015673502276835e-05,
+      "loss": 0.3655,
+      "step": 10259
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6014877287882513e-05,
+      "loss": 0.4094,
+      "step": 10260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601408101373596e-05,
+      "loss": 0.4033,
+      "step": 10261
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6013284679845097e-05,
+      "loss": 0.4171,
+      "step": 10262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601248828621782e-05,
+      "loss": 0.3353,
+      "step": 10263
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6011691832862052e-05,
+      "loss": 0.4105,
+      "step": 10264
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60108953197857e-05,
+      "loss": 0.4159,
+      "step": 10265
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6010098746996678e-05,
+      "loss": 0.3021,
+      "step": 10266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60093021145029e-05,
+      "loss": 0.397,
+      "step": 10267
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600850542231228e-05,
+      "loss": 0.3799,
+      "step": 10268
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600770867043273e-05,
+      "loss": 0.3747,
+      "step": 10269
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600691185887216e-05,
+      "loss": 0.398,
+      "step": 10270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6006114987638503e-05,
+      "loss": 0.384,
+      "step": 10271
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6005318056739657e-05,
+      "loss": 0.3661,
+      "step": 10272
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6004521066183546e-05,
+      "loss": 0.3777,
+      "step": 10273
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600372401597809e-05,
+      "loss": 0.3653,
+      "step": 10274
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6002926906131202e-05,
+      "loss": 0.3508,
+      "step": 10275
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60021297366508e-05,
+      "loss": 0.4005,
+      "step": 10276
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600133250754481e-05,
+      "loss": 0.4153,
+      "step": 10277
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6000535218821146e-05,
+      "loss": 0.3828,
+      "step": 10278
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599973787048773e-05,
+      "loss": 0.4671,
+      "step": 10279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5998940462552485e-05,
+      "loss": 0.3653,
+      "step": 10280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599814299502333e-05,
+      "loss": 0.3827,
+      "step": 10281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599734546790818e-05,
+      "loss": 0.3549,
+      "step": 10282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5996547881214972e-05,
+      "loss": 0.4309,
+      "step": 10283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5995750234951622e-05,
+      "loss": 0.3715,
+      "step": 10284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599495252912605e-05,
+      "loss": 0.4331,
+      "step": 10285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5994154763746187e-05,
+      "loss": 0.4262,
+      "step": 10286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5993356938819956e-05,
+      "loss": 0.4705,
+      "step": 10287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5992559054355284e-05,
+      "loss": 0.4483,
+      "step": 10288
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5991761110360094e-05,
+      "loss": 0.334,
+      "step": 10289
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5990963106842312e-05,
+      "loss": 0.3371,
+      "step": 10290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599016504380987e-05,
+      "loss": 0.3571,
+      "step": 10291
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5989366921270696e-05,
+      "loss": 0.3683,
+      "step": 10292
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598856873923271e-05,
+      "loss": 0.4058,
+      "step": 10293
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5987770497703855e-05,
+      "loss": 0.3778,
+      "step": 10294
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5986972196692053e-05,
+      "loss": 0.4072,
+      "step": 10295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598617383620523e-05,
+      "loss": 0.3668,
+      "step": 10296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5985375416251326e-05,
+      "loss": 0.3316,
+      "step": 10297
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5984576936838267e-05,
+      "loss": 0.3735,
+      "step": 10298
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5983778397973986e-05,
+      "loss": 0.4059,
+      "step": 10299
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5982979799666416e-05,
+      "loss": 0.3931,
+      "step": 10300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5982181141923493e-05,
+      "loss": 0.4099,
+      "step": 10301
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.598138242475314e-05,
+      "loss": 0.4659,
+      "step": 10302
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.598058364816331e-05,
+      "loss": 0.346,
+      "step": 10303
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5979784812161924e-05,
+      "loss": 0.4463,
+      "step": 10304
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5978985916756925e-05,
+      "loss": 0.4297,
+      "step": 10305
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5978186961956244e-05,
+      "loss": 0.3802,
+      "step": 10306
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5977387947767816e-05,
+      "loss": 0.4784,
+      "step": 10307
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597658887419959e-05,
+      "loss": 0.4174,
+      "step": 10308
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597578974125949e-05,
+      "loss": 0.4391,
+      "step": 10309
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5974990548955465e-05,
+      "loss": 0.3073,
+      "step": 10310
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5974191297295448e-05,
+      "loss": 0.4452,
+      "step": 10311
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5973391986287387e-05,
+      "loss": 0.3684,
+      "step": 10312
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597259261593921e-05,
+      "loss": 0.408,
+      "step": 10313
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5971793186258865e-05,
+      "loss": 0.3518,
+      "step": 10314
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970993697254294e-05,
+      "loss": 0.3675,
+      "step": 10315
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970194148933443e-05,
+      "loss": 0.3983,
+      "step": 10316
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5969394541304245e-05,
+      "loss": 0.3841,
+      "step": 10317
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596859487437465e-05,
+      "loss": 0.3782,
+      "step": 10318
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59677951481526e-05,
+      "loss": 0.4149,
+      "step": 10319
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5966995362646044e-05,
+      "loss": 0.3871,
+      "step": 10320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596619551786292e-05,
+      "loss": 0.4305,
+      "step": 10321
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5965395613811178e-05,
+      "loss": 0.3785,
+      "step": 10322
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5964595650498758e-05,
+      "loss": 0.325,
+      "step": 10323
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5963795627933616e-05,
+      "loss": 0.3743,
+      "step": 10324
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5962995546123696e-05,
+      "loss": 0.3831,
+      "step": 10325
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5962195405076943e-05,
+      "loss": 0.3299,
+      "step": 10326
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5961395204801314e-05,
+      "loss": 0.3693,
+      "step": 10327
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596059494530475e-05,
+      "loss": 0.3612,
+      "step": 10328
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59597946265952e-05,
+      "loss": 0.3886,
+      "step": 10329
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595899424868062e-05,
+      "loss": 0.378,
+      "step": 10330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5958193811568962e-05,
+      "loss": 0.343,
+      "step": 10331
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595739331526817e-05,
+      "loss": 0.3339,
+      "step": 10332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59565927597862e-05,
+      "loss": 0.3668,
+      "step": 10333
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5955792145131007e-05,
+      "loss": 0.6024,
+      "step": 10334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5954991471310542e-05,
+      "loss": 0.4683,
+      "step": 10335
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595419073833276e-05,
+      "loss": 0.3639,
+      "step": 10336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595338994620562e-05,
+      "loss": 0.3836,
+      "step": 10337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595258909493707e-05,
+      "loss": 0.4118,
+      "step": 10338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5951788184535065e-05,
+      "loss": 0.3826,
+      "step": 10339
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5950987215007565e-05,
+      "loss": 0.4074,
+      "step": 10340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5950186186362526e-05,
+      "loss": 0.432,
+      "step": 10341
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594938509860791e-05,
+      "loss": 0.4688,
+      "step": 10342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5948583951751663e-05,
+      "loss": 0.4292,
+      "step": 10343
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594778274580176e-05,
+      "loss": 0.3244,
+      "step": 10344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5946981480766146e-05,
+      "loss": 0.4136,
+      "step": 10345
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594618015665279e-05,
+      "loss": 0.3399,
+      "step": 10346
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5945378773469645e-05,
+      "loss": 0.4174,
+      "step": 10347
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594457733122468e-05,
+      "loss": 0.3963,
+      "step": 10348
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5943775829925853e-05,
+      "loss": 0.4903,
+      "step": 10349
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594297426958112e-05,
+      "loss": 0.3999,
+      "step": 10350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5942172650198453e-05,
+      "loss": 0.5182,
+      "step": 10351
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5941370971785814e-05,
+      "loss": 0.3919,
+      "step": 10352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5940569234351156e-05,
+      "loss": 0.3651,
+      "step": 10353
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593976743790246e-05,
+      "loss": 0.342,
+      "step": 10354
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5938965582447683e-05,
+      "loss": 0.4155,
+      "step": 10355
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593816366799479e-05,
+      "loss": 0.3984,
+      "step": 10356
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593736169455175e-05,
+      "loss": 0.4367,
+      "step": 10357
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5936559662126526e-05,
+      "loss": 0.4539,
+      "step": 10358
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5935757570727086e-05,
+      "loss": 0.4239,
+      "step": 10359
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5934955420361404e-05,
+      "loss": 0.4285,
+      "step": 10360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593415321103744e-05,
+      "loss": 0.3815,
+      "step": 10361
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593335094276317e-05,
+      "loss": 0.3742,
+      "step": 10362
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5932548615546563e-05,
+      "loss": 0.4398,
+      "step": 10363
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593174622939558e-05,
+      "loss": 0.3631,
+      "step": 10364
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593094378431821e-05,
+      "loss": 0.3533,
+      "step": 10365
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5930141280322407e-05,
+      "loss": 0.3445,
+      "step": 10366
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592933871741615e-05,
+      "loss": 0.3866,
+      "step": 10367
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5928536095607418e-05,
+      "loss": 0.3998,
+      "step": 10368
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5927733414904175e-05,
+      "loss": 0.4407,
+      "step": 10369
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.59269306753144e-05,
+      "loss": 0.4388,
+      "step": 10370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592612787684606e-05,
+      "loss": 0.3024,
+      "step": 10371
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5925325019507138e-05,
+      "loss": 0.4629,
+      "step": 10372
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5924522103305612e-05,
+      "loss": 0.4,
+      "step": 10373
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5923719128249448e-05,
+      "loss": 0.4302,
+      "step": 10374
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5922916094346633e-05,
+      "loss": 0.4226,
+      "step": 10375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5922113001605134e-05,
+      "loss": 0.4246,
+      "step": 10376
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592130985003294e-05,
+      "loss": 0.4178,
+      "step": 10377
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5920506639638022e-05,
+      "loss": 0.4221,
+      "step": 10378
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5919703370428363e-05,
+      "loss": 0.4371,
+      "step": 10379
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591890004241194e-05,
+      "loss": 0.4044,
+      "step": 10380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5918096655596733e-05,
+      "loss": 0.4173,
+      "step": 10381
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5917293209990728e-05,
+      "loss": 0.3619,
+      "step": 10382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5916489705601904e-05,
+      "loss": 0.4065,
+      "step": 10383
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5915686142438235e-05,
+      "loss": 0.3686,
+      "step": 10384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5914882520507716e-05,
+      "loss": 0.3956,
+      "step": 10385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5914078839818324e-05,
+      "loss": 0.3691,
+      "step": 10386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5913275100378047e-05,
+      "loss": 0.3568,
+      "step": 10387
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591247130219486e-05,
+      "loss": 0.3429,
+      "step": 10388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591166744527676e-05,
+      "loss": 0.3824,
+      "step": 10389
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591086352963172e-05,
+      "loss": 0.4087,
+      "step": 10390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591005955526774e-05,
+      "loss": 0.2999,
+      "step": 10391
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5909255522192794e-05,
+      "loss": 0.4444,
+      "step": 10392
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5908451430414874e-05,
+      "loss": 0.2766,
+      "step": 10393
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5907647279941975e-05,
+      "loss": 0.4109,
+      "step": 10394
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5906843070782077e-05,
+      "loss": 0.4073,
+      "step": 10395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590603880294317e-05,
+      "loss": 0.2979,
+      "step": 10396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5905234476433245e-05,
+      "loss": 0.4232,
+      "step": 10397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5904430091260294e-05,
+      "loss": 0.3993,
+      "step": 10398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5903625647432303e-05,
+      "loss": 0.322,
+      "step": 10399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590282114495727e-05,
+      "loss": 0.3645,
+      "step": 10400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5902016583843183e-05,
+      "loss": 0.3697,
+      "step": 10401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5901211964098033e-05,
+      "loss": 0.3724,
+      "step": 10402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5900407285729817e-05,
+      "loss": 0.3793,
+      "step": 10403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5899602548746525e-05,
+      "loss": 0.5209,
+      "step": 10404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5898797753156156e-05,
+      "loss": 0.4576,
+      "step": 10405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.58979928989667e-05,
+      "loss": 0.4046,
+      "step": 10406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5897187986186154e-05,
+      "loss": 0.3856,
+      "step": 10407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5896383014822517e-05,
+      "loss": 0.3703,
+      "step": 10408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5895577984883783e-05,
+      "loss": 0.3783,
+      "step": 10409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589477289637795e-05,
+      "loss": 0.4113,
+      "step": 10410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5893967749313013e-05,
+      "loss": 0.4,
+      "step": 10411
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5893162543696976e-05,
+      "loss": 0.4119,
+      "step": 10412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589235727953783e-05,
+      "loss": 0.3861,
+      "step": 10413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5891551956843584e-05,
+      "loss": 0.398,
+      "step": 10414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589074657562223e-05,
+      "loss": 0.3939,
+      "step": 10415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5889941135881778e-05,
+      "loss": 0.4838,
+      "step": 10416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5889135637630217e-05,
+      "loss": 0.4275,
+      "step": 10417
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5888330080875557e-05,
+      "loss": 0.3789,
+      "step": 10418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5887524465625797e-05,
+      "loss": 0.3983,
+      "step": 10419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5886718791888945e-05,
+      "loss": 0.3968,
+      "step": 10420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5885913059673002e-05,
+      "loss": 0.3626,
+      "step": 10421
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588510726898597e-05,
+      "loss": 0.3491,
+      "step": 10422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5884301419835854e-05,
+      "loss": 0.3947,
+      "step": 10423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5883495512230663e-05,
+      "loss": 0.3404,
+      "step": 10424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.58826895461784e-05,
+      "loss": 0.3497,
+      "step": 10425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5881883521687075e-05,
+      "loss": 0.4169,
+      "step": 10426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5881077438764686e-05,
+      "loss": 0.3932,
+      "step": 10427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5880271297419253e-05,
+      "loss": 0.3508,
+      "step": 10428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5879465097658776e-05,
+      "loss": 0.4179,
+      "step": 10429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5878658839491267e-05,
+      "loss": 0.3768,
+      "step": 10430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.3858,
+      "step": 10431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5877046147967186e-05,
+      "loss": 0.3881,
+      "step": 10432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5876239714626636e-05,
+      "loss": 0.4679,
+      "step": 10433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5875433222911097e-05,
+      "loss": 0.3628,
+      "step": 10434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5874626672828573e-05,
+      "loss": 0.4094,
+      "step": 10435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5873820064387086e-05,
+      "loss": 0.395,
+      "step": 10436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5873013397594644e-05,
+      "loss": 0.3849,
+      "step": 10437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587220667245926e-05,
+      "loss": 0.4131,
+      "step": 10438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5871399888988948e-05,
+      "loss": 0.3243,
+      "step": 10439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5870593047191723e-05,
+      "loss": 0.4429,
+      "step": 10440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5869786147075603e-05,
+      "loss": 0.4015,
+      "step": 10441
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58689791886486e-05,
+      "loss": 0.3838,
+      "step": 10442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586817217191873e-05,
+      "loss": 0.3692,
+      "step": 10443
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5867365096894017e-05,
+      "loss": 0.3781,
+      "step": 10444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586655796358247e-05,
+      "loss": 0.4192,
+      "step": 10445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5865750771992113e-05,
+      "loss": 0.424,
+      "step": 10446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586494352213096e-05,
+      "loss": 0.3924,
+      "step": 10447
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5864136214007033e-05,
+      "loss": 0.3697,
+      "step": 10448
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586332884762835e-05,
+      "loss": 0.5352,
+      "step": 10449
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5862521423002935e-05,
+      "loss": 0.3797,
+      "step": 10450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5861713940138805e-05,
+      "loss": 0.3102,
+      "step": 10451
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5860906399043986e-05,
+      "loss": 0.3919,
+      "step": 10452
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5860098799726498e-05,
+      "loss": 0.3759,
+      "step": 10453
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585929114219436e-05,
+      "loss": 0.3207,
+      "step": 10454
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58584834264556e-05,
+      "loss": 0.3926,
+      "step": 10455
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5857675652518242e-05,
+      "loss": 0.3608,
+      "step": 10456
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585686782039031e-05,
+      "loss": 0.4703,
+      "step": 10457
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5856059930079825e-05,
+      "loss": 0.4059,
+      "step": 10458
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5855251981594813e-05,
+      "loss": 0.3836,
+      "step": 10459
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585444397494331e-05,
+      "loss": 0.3424,
+      "step": 10460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585363591013333e-05,
+      "loss": 0.3429,
+      "step": 10461
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585282778717291e-05,
+      "loss": 0.4204,
+      "step": 10462
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5852019606070074e-05,
+      "loss": 0.4572,
+      "step": 10463
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585121136683285e-05,
+      "loss": 0.3619,
+      "step": 10464
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585040306946927e-05,
+      "loss": 0.3709,
+      "step": 10465
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5849594713987356e-05,
+      "loss": 0.4461,
+      "step": 10466
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584878630039515e-05,
+      "loss": 0.304,
+      "step": 10467
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5847977828700673e-05,
+      "loss": 0.3617,
+      "step": 10468
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584716929891196e-05,
+      "loss": 0.3162,
+      "step": 10469
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5846360711037042e-05,
+      "loss": 0.3365,
+      "step": 10470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5845552065083953e-05,
+      "loss": 0.3474,
+      "step": 10471
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5844743361060728e-05,
+      "loss": 0.4214,
+      "step": 10472
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5843934598975397e-05,
+      "loss": 0.3501,
+      "step": 10473
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5843125778835996e-05,
+      "loss": 0.4416,
+      "step": 10474
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584231690065056e-05,
+      "loss": 0.4526,
+      "step": 10475
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584150796442712e-05,
+      "loss": 0.3897,
+      "step": 10476
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584069897017372e-05,
+      "loss": 0.3848,
+      "step": 10477
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5839889917898394e-05,
+      "loss": 0.3737,
+      "step": 10478
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583908080760918e-05,
+      "loss": 0.345,
+      "step": 10479
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5838271639314107e-05,
+      "loss": 0.4061,
+      "step": 10480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5837462413021222e-05,
+      "loss": 0.4024,
+      "step": 10481
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583665312873856e-05,
+      "loss": 0.3958,
+      "step": 10482
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5835843786474167e-05,
+      "loss": 0.3948,
+      "step": 10483
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5835034386236077e-05,
+      "loss": 0.3292,
+      "step": 10484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583422492803233e-05,
+      "loss": 0.3143,
+      "step": 10485
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583341541187097e-05,
+      "loss": 0.3567,
+      "step": 10486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583260583776004e-05,
+      "loss": 0.4564,
+      "step": 10487
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5831796205707578e-05,
+      "loss": 0.3484,
+      "step": 10488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583098651572163e-05,
+      "loss": 0.3424,
+      "step": 10489
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5830176767810238e-05,
+      "loss": 0.3455,
+      "step": 10490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5829366961981452e-05,
+      "loss": 0.3673,
+      "step": 10491
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5828557098243308e-05,
+      "loss": 0.4409,
+      "step": 10492
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582774717660385e-05,
+      "loss": 0.3868,
+      "step": 10493
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5826937197071137e-05,
+      "loss": 0.3296,
+      "step": 10494
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5826127159653203e-05,
+      "loss": 0.3997,
+      "step": 10495
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.58253170643581e-05,
+      "loss": 0.4247,
+      "step": 10496
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5824506911193874e-05,
+      "loss": 0.355,
+      "step": 10497
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5823696700168576e-05,
+      "loss": 0.3708,
+      "step": 10498
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582288643129025e-05,
+      "loss": 0.3577,
+      "step": 10499
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582207610456695e-05,
+      "loss": 0.3886,
+      "step": 10500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582126572000672e-05,
+      "loss": 0.3922,
+      "step": 10501
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582045527761762e-05,
+      "loss": 0.4044,
+      "step": 10502
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581964477740769e-05,
+      "loss": 0.4274,
+      "step": 10503
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5818834219384987e-05,
+      "loss": 0.4619,
+      "step": 10504
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581802360355756e-05,
+      "loss": 0.3908,
+      "step": 10505
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5817212929933472e-05,
+      "loss": 0.3692,
+      "step": 10506
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5816402198520764e-05,
+      "loss": 0.4299,
+      "step": 10507
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5815591409327496e-05,
+      "loss": 0.437,
+      "step": 10508
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581478056236172e-05,
+      "loss": 0.4742,
+      "step": 10509
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5813969657631494e-05,
+      "loss": 0.3757,
+      "step": 10510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5813158695144867e-05,
+      "loss": 0.3566,
+      "step": 10511
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5812347674909902e-05,
+      "loss": 0.4197,
+      "step": 10512
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5811536596934656e-05,
+      "loss": 0.3554,
+      "step": 10513
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581072546122718e-05,
+      "loss": 0.3657,
+      "step": 10514
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5809914267795542e-05,
+      "loss": 0.4565,
+      "step": 10515
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5809103016647788e-05,
+      "loss": 0.4841,
+      "step": 10516
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5808291707791984e-05,
+      "loss": 0.3991,
+      "step": 10517
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580748034123619e-05,
+      "loss": 0.4244,
+      "step": 10518
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5806668916988465e-05,
+      "loss": 0.4027,
+      "step": 10519
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580585743505687e-05,
+      "loss": 0.414,
+      "step": 10520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5805045895449467e-05,
+      "loss": 0.4468,
+      "step": 10521
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5804234298174313e-05,
+      "loss": 0.3957,
+      "step": 10522
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5803422643239475e-05,
+      "loss": 0.4319,
+      "step": 10523
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580261093065302e-05,
+      "loss": 0.3961,
+      "step": 10524
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5801799160423004e-05,
+      "loss": 0.3838,
+      "step": 10525
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5800987332557496e-05,
+      "loss": 0.3985,
+      "step": 10526
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580017544706456e-05,
+      "loss": 0.3599,
+      "step": 10527
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5799363503952256e-05,
+      "loss": 0.3811,
+      "step": 10528
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5798551503228655e-05,
+      "loss": 0.2884,
+      "step": 10529
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5797739444901822e-05,
+      "loss": 0.3939,
+      "step": 10530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5796927328979826e-05,
+      "loss": 0.3595,
+      "step": 10531
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5796115155470736e-05,
+      "loss": 0.4121,
+      "step": 10532
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5795302924382613e-05,
+      "loss": 0.4222,
+      "step": 10533
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5794490635723535e-05,
+      "loss": 0.426,
+      "step": 10534
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5793678289501564e-05,
+      "loss": 0.4667,
+      "step": 10535
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5792865885724772e-05,
+      "loss": 0.4512,
+      "step": 10536
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579205342440123e-05,
+      "loss": 0.3945,
+      "step": 10537
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5791240905539012e-05,
+      "loss": 0.4365,
+      "step": 10538
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5790428329146182e-05,
+      "loss": 0.3042,
+      "step": 10539
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578961569523082e-05,
+      "loss": 0.3381,
+      "step": 10540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5788803003800995e-05,
+      "loss": 0.4178,
+      "step": 10541
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578799025486478e-05,
+      "loss": 0.356,
+      "step": 10542
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5787177448430254e-05,
+      "loss": 0.3516,
+      "step": 10543
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5786364584505483e-05,
+      "loss": 0.3672,
+      "step": 10544
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5785551663098546e-05,
+      "loss": 0.318,
+      "step": 10545
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5784738684217523e-05,
+      "loss": 0.3822,
+      "step": 10546
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5783925647870485e-05,
+      "loss": 0.3353,
+      "step": 10547
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5783112554065505e-05,
+      "loss": 0.4266,
+      "step": 10548
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5782299402810672e-05,
+      "loss": 0.4084,
+      "step": 10549
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5781486194114056e-05,
+      "loss": 0.4352,
+      "step": 10550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5780672927983737e-05,
+      "loss": 0.4057,
+      "step": 10551
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577985960442779e-05,
+      "loss": 0.3922,
+      "step": 10552
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5779046223454296e-05,
+      "loss": 0.4253,
+      "step": 10553
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5778232785071344e-05,
+      "loss": 0.3639,
+      "step": 10554
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5777419289287008e-05,
+      "loss": 0.4171,
+      "step": 10555
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5776605736109366e-05,
+      "loss": 0.3835,
+      "step": 10556
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5775792125546506e-05,
+      "loss": 0.3348,
+      "step": 10557
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5774978457606508e-05,
+      "loss": 0.4025,
+      "step": 10558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5774164732297453e-05,
+      "loss": 0.462,
+      "step": 10559
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5773350949627428e-05,
+      "loss": 0.3414,
+      "step": 10560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5772537109604518e-05,
+      "loss": 0.3241,
+      "step": 10561
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5771723212236803e-05,
+      "loss": 0.3569,
+      "step": 10562
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.577090925753237e-05,
+      "loss": 0.4061,
+      "step": 10563
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5770095245499306e-05,
+      "loss": 0.3897,
+      "step": 10564
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57692811761457e-05,
+      "loss": 0.3758,
+      "step": 10565
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5768467049479636e-05,
+      "loss": 0.3659,
+      "step": 10566
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57676528655092e-05,
+      "loss": 0.3518,
+      "step": 10567
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5766838624242483e-05,
+      "loss": 0.4354,
+      "step": 10568
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5766024325687567e-05,
+      "loss": 0.3939,
+      "step": 10569
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5765209969852553e-05,
+      "loss": 0.4042,
+      "step": 10570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5764395556745524e-05,
+      "loss": 0.3587,
+      "step": 10571
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5763581086374573e-05,
+      "loss": 0.3703,
+      "step": 10572
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5762766558747785e-05,
+      "loss": 0.4886,
+      "step": 10573
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576195197387326e-05,
+      "loss": 0.412,
+      "step": 10574
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5761137331759084e-05,
+      "loss": 0.496,
+      "step": 10575
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576032263241335e-05,
+      "loss": 0.3756,
+      "step": 10576
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575950787584416e-05,
+      "loss": 0.3789,
+      "step": 10577
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5758693062059593e-05,
+      "loss": 0.3561,
+      "step": 10578
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5757878191067756e-05,
+      "loss": 0.4075,
+      "step": 10579
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5757063262876737e-05,
+      "loss": 0.3882,
+      "step": 10580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5756248277494636e-05,
+      "loss": 0.3428,
+      "step": 10581
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5755433234929543e-05,
+      "loss": 0.4923,
+      "step": 10582
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5754618135189564e-05,
+      "loss": 0.3416,
+      "step": 10583
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5753802978282786e-05,
+      "loss": 0.3656,
+      "step": 10584
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5752987764217317e-05,
+      "loss": 0.3515,
+      "step": 10585
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5752172493001244e-05,
+      "loss": 0.3939,
+      "step": 10586
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575135716464268e-05,
+      "loss": 0.3558,
+      "step": 10587
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575054177914971e-05,
+      "loss": 0.4169,
+      "step": 10588
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5749726336530443e-05,
+      "loss": 0.387,
+      "step": 10589
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5748910836792978e-05,
+      "loss": 0.449,
+      "step": 10590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574809527994541e-05,
+      "loss": 0.375,
+      "step": 10591
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5747279665995857e-05,
+      "loss": 0.3895,
+      "step": 10592
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5746463994952406e-05,
+      "loss": 0.3631,
+      "step": 10593
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5745648266823163e-05,
+      "loss": 0.3986,
+      "step": 10594
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5744832481616235e-05,
+      "loss": 0.4064,
+      "step": 10595
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5744016639339724e-05,
+      "loss": 0.4339,
+      "step": 10596
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5743200740001734e-05,
+      "loss": 0.2858,
+      "step": 10597
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574238478361037e-05,
+      "loss": 0.3216,
+      "step": 10598
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574156877017374e-05,
+      "loss": 0.4504,
+      "step": 10599
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574075269969995e-05,
+      "loss": 0.4146,
+      "step": 10600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573993657219711e-05,
+      "loss": 0.3851,
+      "step": 10601
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5739120387673317e-05,
+      "loss": 0.3748,
+      "step": 10602
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573830414613669e-05,
+      "loss": 0.3638,
+      "step": 10603
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573748784759533e-05,
+      "loss": 0.4258,
+      "step": 10604
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5736671492057355e-05,
+      "loss": 0.3882,
+      "step": 10605
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5735855079530864e-05,
+      "loss": 0.3971,
+      "step": 10606
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5735038610023975e-05,
+      "loss": 0.3178,
+      "step": 10607
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5734222083544794e-05,
+      "loss": 0.3557,
+      "step": 10608
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5733405500101436e-05,
+      "loss": 0.4523,
+      "step": 10609
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5732588859702012e-05,
+      "loss": 0.371,
+      "step": 10610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573177216235463e-05,
+      "loss": 0.3822,
+      "step": 10611
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5730955408067415e-05,
+      "loss": 0.4398,
+      "step": 10612
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573013859684847e-05,
+      "loss": 0.3622,
+      "step": 10613
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5729321728705917e-05,
+      "loss": 0.4089,
+      "step": 10614
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5728504803647857e-05,
+      "loss": 0.4178,
+      "step": 10615
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572768782168242e-05,
+      "loss": 0.4595,
+      "step": 10616
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5726870782817717e-05,
+      "loss": 0.4264,
+      "step": 10617
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5726053687061864e-05,
+      "loss": 0.3268,
+      "step": 10618
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5725236534422975e-05,
+      "loss": 0.4175,
+      "step": 10619
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5724419324909174e-05,
+      "loss": 0.3297,
+      "step": 10620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572360205852858e-05,
+      "loss": 0.4099,
+      "step": 10621
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57227847352893e-05,
+      "loss": 0.3448,
+      "step": 10622
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5721967355199464e-05,
+      "loss": 0.4107,
+      "step": 10623
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572114991826719e-05,
+      "loss": 0.3551,
+      "step": 10624
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57203324245006e-05,
+      "loss": 0.387,
+      "step": 10625
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571951487390781e-05,
+      "loss": 0.3447,
+      "step": 10626
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5718697266496942e-05,
+      "loss": 0.3681,
+      "step": 10627
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5717879602276123e-05,
+      "loss": 0.3502,
+      "step": 10628
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5717061881253474e-05,
+      "loss": 0.4158,
+      "step": 10629
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5716244103437116e-05,
+      "loss": 0.4814,
+      "step": 10630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5715426268835176e-05,
+      "loss": 0.4373,
+      "step": 10631
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5714608377455776e-05,
+      "loss": 0.3844,
+      "step": 10632
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571379042930704e-05,
+      "loss": 0.3335,
+      "step": 10633
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5712972424397097e-05,
+      "loss": 0.4218,
+      "step": 10634
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5712154362734073e-05,
+      "loss": 0.4188,
+      "step": 10635
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5711336244326093e-05,
+      "loss": 0.3477,
+      "step": 10636
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571051806918128e-05,
+      "loss": 0.3942,
+      "step": 10637
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570969983730777e-05,
+      "loss": 0.4003,
+      "step": 10638
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570888154871369e-05,
+      "loss": 0.4534,
+      "step": 10639
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570806320340716e-05,
+      "loss": 0.3719,
+      "step": 10640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570724480139632e-05,
+      "loss": 0.317,
+      "step": 10641
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5706426342689294e-05,
+      "loss": 0.4138,
+      "step": 10642
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5705607827294217e-05,
+      "loss": 0.4332,
+      "step": 10643
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570478925521922e-05,
+      "loss": 0.4032,
+      "step": 10644
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5703970626472424e-05,
+      "loss": 0.4448,
+      "step": 10645
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5703151941061978e-05,
+      "loss": 0.4841,
+      "step": 10646
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5702333198996002e-05,
+      "loss": 0.3874,
+      "step": 10647
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5701514400282638e-05,
+      "loss": 0.3966,
+      "step": 10648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5700695544930017e-05,
+      "loss": 0.3857,
+      "step": 10649
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569987663294627e-05,
+      "loss": 0.3413,
+      "step": 10650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5699057664339537e-05,
+      "loss": 0.3945,
+      "step": 10651
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5698238639117952e-05,
+      "loss": 0.4141,
+      "step": 10652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569741955728965e-05,
+      "loss": 0.3841,
+      "step": 10653
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5696600418862768e-05,
+      "loss": 0.4459,
+      "step": 10654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5695781223845442e-05,
+      "loss": 0.346,
+      "step": 10655
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5694961972245814e-05,
+      "loss": 0.4516,
+      "step": 10656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5694142664072022e-05,
+      "loss": 0.3745,
+      "step": 10657
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.56933232993322e-05,
+      "loss": 0.3408,
+      "step": 10658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5692503878034496e-05,
+      "loss": 0.4104,
+      "step": 10659
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5691684400187045e-05,
+      "loss": 0.3626,
+      "step": 10660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5690864865797987e-05,
+      "loss": 0.4053,
+      "step": 10661
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5690045274875463e-05,
+      "loss": 0.3688,
+      "step": 10662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5689225627427618e-05,
+      "loss": 0.403,
+      "step": 10663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5688405923462594e-05,
+      "loss": 0.3974,
+      "step": 10664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568758616298853e-05,
+      "loss": 0.3604,
+      "step": 10665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5686766346013573e-05,
+      "loss": 0.4054,
+      "step": 10666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568594647254587e-05,
+      "loss": 0.4169,
+      "step": 10667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568512654259356e-05,
+      "loss": 0.3739,
+      "step": 10668
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568430655616479e-05,
+      "loss": 0.3607,
+      "step": 10669
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5683486513267707e-05,
+      "loss": 0.4367,
+      "step": 10670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568266641391045e-05,
+      "loss": 0.4142,
+      "step": 10671
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5681846258101183e-05,
+      "loss": 0.4245,
+      "step": 10672
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568102604584804e-05,
+      "loss": 0.4224,
+      "step": 10673
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5680205777159166e-05,
+      "loss": 0.3681,
+      "step": 10674
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5679385452042724e-05,
+      "loss": 0.408,
+      "step": 10675
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567856507050685e-05,
+      "loss": 0.4618,
+      "step": 10676
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5677744632559703e-05,
+      "loss": 0.4479,
+      "step": 10677
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5676924138209425e-05,
+      "loss": 0.3857,
+      "step": 10678
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567610358746417e-05,
+      "loss": 0.39,
+      "step": 10679
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567528298033209e-05,
+      "loss": 0.3734,
+      "step": 10680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567446231682134e-05,
+      "loss": 0.3638,
+      "step": 10681
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567364159694007e-05,
+      "loss": 0.4848,
+      "step": 10682
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567282082069643e-05,
+      "loss": 0.3498,
+      "step": 10683
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5671999988098582e-05,
+      "loss": 0.3989,
+      "step": 10684
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567117909915467e-05,
+      "loss": 0.3736,
+      "step": 10685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5670358153872852e-05,
+      "loss": 0.4425,
+      "step": 10686
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5669537152261285e-05,
+      "loss": 0.3503,
+      "step": 10687
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566871609432813e-05,
+      "loss": 0.3884,
+      "step": 10688
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566789498008153e-05,
+      "loss": 0.3549,
+      "step": 10689
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5667073809529656e-05,
+      "loss": 0.442,
+      "step": 10690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5666252582680657e-05,
+      "loss": 0.374,
+      "step": 10691
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5665431299542694e-05,
+      "loss": 0.3699,
+      "step": 10692
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5664609960123928e-05,
+      "loss": 0.4396,
+      "step": 10693
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5663788564432515e-05,
+      "loss": 0.4815,
+      "step": 10694
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5662967112476613e-05,
+      "loss": 0.3325,
+      "step": 10695
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5662145604264387e-05,
+      "loss": 0.4091,
+      "step": 10696
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566132403980399e-05,
+      "loss": 0.4958,
+      "step": 10697
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5660502419103597e-05,
+      "loss": 0.3695,
+      "step": 10698
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565968074217136e-05,
+      "loss": 0.3343,
+      "step": 10699
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565885900901544e-05,
+      "loss": 0.3851,
+      "step": 10700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565803721964401e-05,
+      "loss": 0.449,
+      "step": 10701
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5657215374065224e-05,
+      "loss": 0.3811,
+      "step": 10702
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565639347228725e-05,
+      "loss": 0.3447,
+      "step": 10703
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5655571514318257e-05,
+      "loss": 0.4518,
+      "step": 10704
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56547495001664e-05,
+      "loss": 0.3449,
+      "step": 10705
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5653927429839856e-05,
+      "loss": 0.4365,
+      "step": 10706
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5653105303346786e-05,
+      "loss": 0.3818,
+      "step": 10707
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5652283120695357e-05,
+      "loss": 0.468,
+      "step": 10708
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5651460881893735e-05,
+      "loss": 0.4216,
+      "step": 10709
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5650638586950095e-05,
+      "loss": 0.4252,
+      "step": 10710
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56498162358726e-05,
+      "loss": 0.3432,
+      "step": 10711
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5648993828669418e-05,
+      "loss": 0.3923,
+      "step": 10712
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5648171365348727e-05,
+      "loss": 0.4204,
+      "step": 10713
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5647348845918685e-05,
+      "loss": 0.3552,
+      "step": 10714
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5646526270387473e-05,
+      "loss": 0.4189,
+      "step": 10715
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564570363876326e-05,
+      "loss": 0.2948,
+      "step": 10716
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5644880951054216e-05,
+      "loss": 0.3598,
+      "step": 10717
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564405820726852e-05,
+      "loss": 0.4969,
+      "step": 10718
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5643235407414335e-05,
+      "loss": 0.4521,
+      "step": 10719
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5642412551499842e-05,
+      "loss": 0.4385,
+      "step": 10720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5641589639533212e-05,
+      "loss": 0.3175,
+      "step": 10721
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564076667152262e-05,
+      "loss": 0.3505,
+      "step": 10722
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5639943647476247e-05,
+      "loss": 0.393,
+      "step": 10723
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563912056740226e-05,
+      "loss": 0.3833,
+      "step": 10724
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5638297431308846e-05,
+      "loss": 0.4262,
+      "step": 10725
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563747423920417e-05,
+      "loss": 0.4712,
+      "step": 10726
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5636650991096418e-05,
+      "loss": 0.3368,
+      "step": 10727
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563582768699377e-05,
+      "loss": 0.3797,
+      "step": 10728
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5635004326904397e-05,
+      "loss": 0.3583,
+      "step": 10729
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5634180910836482e-05,
+      "loss": 0.4078,
+      "step": 10730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5633357438798206e-05,
+      "loss": 0.3418,
+      "step": 10731
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5632533910797747e-05,
+      "loss": 0.3424,
+      "step": 10732
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563171032684329e-05,
+      "loss": 0.3609,
+      "step": 10733
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5630886686943014e-05,
+      "loss": 0.4068,
+      "step": 10734
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56300629911051e-05,
+      "loss": 0.4158,
+      "step": 10735
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562923923933773e-05,
+      "loss": 0.3523,
+      "step": 10736
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562841543164909e-05,
+      "loss": 0.4351,
+      "step": 10737
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5627591568047364e-05,
+      "loss": 0.3864,
+      "step": 10738
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562676764854073e-05,
+      "loss": 0.3422,
+      "step": 10739
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5625943673137388e-05,
+      "loss": 0.3094,
+      "step": 10740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5625119641845507e-05,
+      "loss": 0.346,
+      "step": 10741
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562429555467328e-05,
+      "loss": 0.3675,
+      "step": 10742
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5623471411628896e-05,
+      "loss": 0.3739,
+      "step": 10743
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5622647212720534e-05,
+      "loss": 0.4285,
+      "step": 10744
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562182295795639e-05,
+      "loss": 0.3881,
+      "step": 10745
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5620998647344648e-05,
+      "loss": 0.3238,
+      "step": 10746
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56201742808935e-05,
+      "loss": 0.3312,
+      "step": 10747
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5619349858611127e-05,
+      "loss": 0.478,
+      "step": 10748
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5618525380505728e-05,
+      "loss": 0.3729,
+      "step": 10749
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561770084658549e-05,
+      "loss": 0.3851,
+      "step": 10750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5616876256858605e-05,
+      "loss": 0.4389,
+      "step": 10751
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5616051611333264e-05,
+      "loss": 0.4132,
+      "step": 10752
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5615226910017657e-05,
+      "loss": 0.3755,
+      "step": 10753
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561440215291998e-05,
+      "loss": 0.4261,
+      "step": 10754
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5613577340048423e-05,
+      "loss": 0.4244,
+      "step": 10755
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5612752471411185e-05,
+      "loss": 0.4149,
+      "step": 10756
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5611927547016455e-05,
+      "loss": 0.4408,
+      "step": 10757
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.561110256687243e-05,
+      "loss": 0.3764,
+      "step": 10758
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5610277530987305e-05,
+      "loss": 0.4075,
+      "step": 10759
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5609452439369274e-05,
+      "loss": 0.4492,
+      "step": 10760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5608627292026537e-05,
+      "loss": 0.4307,
+      "step": 10761
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5607802088967288e-05,
+      "loss": 0.379,
+      "step": 10762
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560697683019973e-05,
+      "loss": 0.4104,
+      "step": 10763
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5606151515732055e-05,
+      "loss": 0.3617,
+      "step": 10764
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560532614557246e-05,
+      "loss": 0.4187,
+      "step": 10765
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5604500719729153e-05,
+      "loss": 0.4221,
+      "step": 10766
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5603675238210333e-05,
+      "loss": 0.3619,
+      "step": 10767
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5602849701024186e-05,
+      "loss": 0.4586,
+      "step": 10768
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5602024108178935e-05,
+      "loss": 0.4245,
+      "step": 10769
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560119845968276e-05,
+      "loss": 0.5179,
+      "step": 10770
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5600372755543883e-05,
+      "loss": 0.3393,
+      "step": 10771
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559954699577049e-05,
+      "loss": 0.409,
+      "step": 10772
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559872118037079e-05,
+      "loss": 0.3903,
+      "step": 10773
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559789530935299e-05,
+      "loss": 0.4032,
+      "step": 10774
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5597069382725293e-05,
+      "loss": 0.4117,
+      "step": 10775
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5596243400495906e-05,
+      "loss": 0.4036,
+      "step": 10776
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5595417362673023e-05,
+      "loss": 0.474,
+      "step": 10777
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5594591269264866e-05,
+      "loss": 0.3758,
+      "step": 10778
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5593765120279628e-05,
+      "loss": 0.385,
+      "step": 10779
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5592938915725522e-05,
+      "loss": 0.3916,
+      "step": 10780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5592112655610757e-05,
+      "loss": 0.3173,
+      "step": 10781
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5591286339943537e-05,
+      "loss": 0.4528,
+      "step": 10782
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5590459968732073e-05,
+      "loss": 0.3776,
+      "step": 10783
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5589633541984576e-05,
+      "loss": 0.4037,
+      "step": 10784
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558880705970925e-05,
+      "loss": 0.4294,
+      "step": 10785
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5587980521914315e-05,
+      "loss": 0.4345,
+      "step": 10786
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5587153928607973e-05,
+      "loss": 0.5513,
+      "step": 10787
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558632727979844e-05,
+      "loss": 0.4271,
+      "step": 10788
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5585500575493922e-05,
+      "loss": 0.3819,
+      "step": 10789
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558467381570264e-05,
+      "loss": 0.3667,
+      "step": 10790
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5583847000432803e-05,
+      "loss": 0.4721,
+      "step": 10791
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5583020129692624e-05,
+      "loss": 0.399,
+      "step": 10792
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558219320349032e-05,
+      "loss": 0.3238,
+      "step": 10793
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55813662218341e-05,
+      "loss": 0.3,
+      "step": 10794
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5580539184732188e-05,
+      "loss": 0.3582,
+      "step": 10795
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557971209219279e-05,
+      "loss": 0.3686,
+      "step": 10796
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5578884944224127e-05,
+      "loss": 0.3889,
+      "step": 10797
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557805774083442e-05,
+      "loss": 0.453,
+      "step": 10798
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557723048203188e-05,
+      "loss": 0.4593,
+      "step": 10799
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557640316782473e-05,
+      "loss": 0.2906,
+      "step": 10800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5575575798221185e-05,
+      "loss": 0.3346,
+      "step": 10801
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5574748373229462e-05,
+      "loss": 0.4093,
+      "step": 10802
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557392089285779e-05,
+      "loss": 0.3726,
+      "step": 10803
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5573093357114382e-05,
+      "loss": 0.3509,
+      "step": 10804
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557226576600746e-05,
+      "loss": 0.4014,
+      "step": 10805
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5571438119545245e-05,
+      "loss": 0.409,
+      "step": 10806
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557061041773596e-05,
+      "loss": 0.3679,
+      "step": 10807
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5569782660587828e-05,
+      "loss": 0.3479,
+      "step": 10808
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5568954848109066e-05,
+      "loss": 0.5327,
+      "step": 10809
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556812698030791e-05,
+      "loss": 0.3345,
+      "step": 10810
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556729905719257e-05,
+      "loss": 0.3954,
+      "step": 10811
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556647107877128e-05,
+      "loss": 0.4192,
+      "step": 10812
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5565643045052263e-05,
+      "loss": 0.3824,
+      "step": 10813
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5564814956043742e-05,
+      "loss": 0.3458,
+      "step": 10814
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5563986811753948e-05,
+      "loss": 0.3217,
+      "step": 10815
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5563158612191106e-05,
+      "loss": 0.4819,
+      "step": 10816
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556233035736344e-05,
+      "loss": 0.3533,
+      "step": 10817
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556150204727918e-05,
+      "loss": 0.3045,
+      "step": 10818
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556067368194656e-05,
+      "loss": 0.3823,
+      "step": 10819
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55598452613738e-05,
+      "loss": 0.4879,
+      "step": 10820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5559016785569135e-05,
+      "loss": 0.3056,
+      "step": 10821
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5558188254540794e-05,
+      "loss": 0.4024,
+      "step": 10822
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5557359668297007e-05,
+      "loss": 0.3172,
+      "step": 10823
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5556531026846005e-05,
+      "loss": 0.3867,
+      "step": 10824
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.4791,
+      "step": 10825
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555487357835529e-05,
+      "loss": 0.4457,
+      "step": 10826
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555404477133204e-05,
+      "loss": 0.3703,
+      "step": 10827
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5553215909134506e-05,
+      "loss": 0.4264,
+      "step": 10828
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5552386991770924e-05,
+      "loss": 0.4057,
+      "step": 10829
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5551558019249525e-05,
+      "loss": 0.431,
+      "step": 10830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5550728991578544e-05,
+      "loss": 0.3746,
+      "step": 10831
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5549899908766222e-05,
+      "loss": 0.384,
+      "step": 10832
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554907077082079e-05,
+      "loss": 0.3909,
+      "step": 10833
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554824157775049e-05,
+      "loss": 0.4114,
+      "step": 10834
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554741232956355e-05,
+      "loss": 0.4058,
+      "step": 10835
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5546583026268218e-05,
+      "loss": 0.3752,
+      "step": 10836
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5545753667872724e-05,
+      "loss": 0.376,
+      "step": 10837
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5544924254385312e-05,
+      "loss": 0.4706,
+      "step": 10838
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5544094785814223e-05,
+      "loss": 0.3332,
+      "step": 10839
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554326526216769e-05,
+      "loss": 0.4314,
+      "step": 10840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554243568345396e-05,
+      "loss": 0.3062,
+      "step": 10841
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554160604968127e-05,
+      "loss": 0.4568,
+      "step": 10842
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554077636085786e-05,
+      "loss": 0.4326,
+      "step": 10843
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5539946616991985e-05,
+      "loss": 0.368,
+      "step": 10844
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5539116818091873e-05,
+      "loss": 0.4183,
+      "step": 10845
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5538286964165778e-05,
+      "loss": 0.3648,
+      "step": 10846
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553745705522193e-05,
+      "loss": 0.3876,
+      "step": 10847
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5536627091268586e-05,
+      "loss": 0.3147,
+      "step": 10848
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5535797072313987e-05,
+      "loss": 0.385,
+      "step": 10849
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5534966998366378e-05,
+      "loss": 0.377,
+      "step": 10850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5534136869434003e-05,
+      "loss": 0.3883,
+      "step": 10851
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5533306685525115e-05,
+      "loss": 0.358,
+      "step": 10852
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5532476446647956e-05,
+      "loss": 0.3847,
+      "step": 10853
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5531646152810773e-05,
+      "loss": 0.3776,
+      "step": 10854
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553081580402182e-05,
+      "loss": 0.3633,
+      "step": 10855
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5529985400289334e-05,
+      "loss": 0.3377,
+      "step": 10856
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5529154941621575e-05,
+      "loss": 0.4011,
+      "step": 10857
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552832442802679e-05,
+      "loss": 0.5496,
+      "step": 10858
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552749385951323e-05,
+      "loss": 0.4375,
+      "step": 10859
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5526663236089143e-05,
+      "loss": 0.3389,
+      "step": 10860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5525832557762787e-05,
+      "loss": 0.381,
+      "step": 10861
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5525001824542404e-05,
+      "loss": 0.3279,
+      "step": 10862
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5524171036436254e-05,
+      "loss": 0.3731,
+      "step": 10863
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5523340193452588e-05,
+      "loss": 0.444,
+      "step": 10864
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5522509295599662e-05,
+      "loss": 0.3702,
+      "step": 10865
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5521678342885724e-05,
+      "loss": 0.4002,
+      "step": 10866
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5520847335319037e-05,
+      "loss": 0.4459,
+      "step": 10867
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552001627290785e-05,
+      "loss": 0.3376,
+      "step": 10868
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551918515566042e-05,
+      "loss": 0.3973,
+      "step": 10869
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5518353983585008e-05,
+      "loss": 0.4618,
+      "step": 10870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5517522756689862e-05,
+      "loss": 0.3857,
+      "step": 10871
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5516691474983248e-05,
+      "loss": 0.3598,
+      "step": 10872
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551586013847342e-05,
+      "loss": 0.3706,
+      "step": 10873
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5515028747168633e-05,
+      "loss": 0.3786,
+      "step": 10874
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551419730107716e-05,
+      "loss": 0.3773,
+      "step": 10875
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5513365800207242e-05,
+      "loss": 0.3424,
+      "step": 10876
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551253424456715e-05,
+      "loss": 0.428,
+      "step": 10877
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5511702634165146e-05,
+      "loss": 0.4587,
+      "step": 10878
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5510870969009485e-05,
+      "loss": 0.4062,
+      "step": 10879
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551003924910843e-05,
+      "loss": 0.4043,
+      "step": 10880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5509207474470255e-05,
+      "loss": 0.3693,
+      "step": 10881
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5508375645103203e-05,
+      "loss": 0.3498,
+      "step": 10882
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5507543761015553e-05,
+      "loss": 0.3776,
+      "step": 10883
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.550671182221556e-05,
+      "loss": 0.4983,
+      "step": 10884
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5505879828711495e-05,
+      "loss": 0.4235,
+      "step": 10885
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5505047780511618e-05,
+      "loss": 0.3927,
+      "step": 10886
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5504215677624204e-05,
+      "loss": 0.3597,
+      "step": 10887
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5503383520057506e-05,
+      "loss": 0.3656,
+      "step": 10888
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.55025513078198e-05,
+      "loss": 0.372,
+      "step": 10889
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5501719040919345e-05,
+      "loss": 0.3946,
+      "step": 10890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5500886719364415e-05,
+      "loss": 0.4045,
+      "step": 10891
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.550005434316328e-05,
+      "loss": 0.4098,
+      "step": 10892
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5499221912324205e-05,
+      "loss": 0.4055,
+      "step": 10893
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549838942685546e-05,
+      "loss": 0.41,
+      "step": 10894
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5497556886765316e-05,
+      "loss": 0.3906,
+      "step": 10895
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5496724292062045e-05,
+      "loss": 0.4681,
+      "step": 10896
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5495891642753914e-05,
+      "loss": 0.3988,
+      "step": 10897
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5495058938849196e-05,
+      "loss": 0.3381,
+      "step": 10898
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5494226180356167e-05,
+      "loss": 0.386,
+      "step": 10899
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5493393367283092e-05,
+      "loss": 0.4539,
+      "step": 10900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5492560499638253e-05,
+      "loss": 0.3674,
+      "step": 10901
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5491727577429916e-05,
+      "loss": 0.4246,
+      "step": 10902
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549089460066636e-05,
+      "loss": 0.4131,
+      "step": 10903
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549006156935586e-05,
+      "loss": 0.3969,
+      "step": 10904
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548922848350669e-05,
+      "loss": 0.39,
+      "step": 10905
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5488395343127126e-05,
+      "loss": 0.3862,
+      "step": 10906
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5487562148225443e-05,
+      "loss": 0.3807,
+      "step": 10907
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548672889880992e-05,
+      "loss": 0.3473,
+      "step": 10908
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5485895594888833e-05,
+      "loss": 0.4374,
+      "step": 10909
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548506223647046e-05,
+      "loss": 0.3565,
+      "step": 10910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5484228823563087e-05,
+      "loss": 0.4111,
+      "step": 10911
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548339535617498e-05,
+      "loss": 0.3979,
+      "step": 10912
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5482561834314432e-05,
+      "loss": 0.3701,
+      "step": 10913
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5481728257989713e-05,
+      "loss": 0.3535,
+      "step": 10914
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548089462720911e-05,
+      "loss": 0.4045,
+      "step": 10915
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.54800609419809e-05,
+      "loss": 0.3416,
+      "step": 10916
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547922720231337e-05,
+      "loss": 0.4299,
+      "step": 10917
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5478393408214795e-05,
+      "loss": 0.3929,
+      "step": 10918
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5477559559693466e-05,
+      "loss": 0.4335,
+      "step": 10919
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5476725656757657e-05,
+      "loss": 0.4104,
+      "step": 10920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5475891699415664e-05,
+      "loss": 0.3666,
+      "step": 10921
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5475057687675766e-05,
+      "loss": 0.3905,
+      "step": 10922
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5474223621546246e-05,
+      "loss": 0.4057,
+      "step": 10923
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547338950103539e-05,
+      "loss": 0.3792,
+      "step": 10924
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5472555326151488e-05,
+      "loss": 0.3539,
+      "step": 10925
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5471721096902823e-05,
+      "loss": 0.3249,
+      "step": 10926
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5470886813297685e-05,
+      "loss": 0.4758,
+      "step": 10927
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5470052475344358e-05,
+      "loss": 0.365,
+      "step": 10928
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5469218083051133e-05,
+      "loss": 0.3977,
+      "step": 10929
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5468383636426302e-05,
+      "loss": 0.2786,
+      "step": 10930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5467549135478147e-05,
+      "loss": 0.3971,
+      "step": 10931
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5466714580214967e-05,
+      "loss": 0.3752,
+      "step": 10932
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5465879970645046e-05,
+      "loss": 0.3426,
+      "step": 10933
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5465045306776673e-05,
+      "loss": 0.2807,
+      "step": 10934
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5464210588618148e-05,
+      "loss": 0.3513,
+      "step": 10935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5463375816177756e-05,
+      "loss": 0.4427,
+      "step": 10936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5462540989463793e-05,
+      "loss": 0.4179,
+      "step": 10937
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5461706108484552e-05,
+      "loss": 0.4377,
+      "step": 10938
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5460871173248325e-05,
+      "loss": 0.4314,
+      "step": 10939
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546003618376341e-05,
+      "loss": 0.4409,
+      "step": 10940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.54592011400381e-05,
+      "loss": 0.4662,
+      "step": 10941
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5458366042080686e-05,
+      "loss": 0.3754,
+      "step": 10942
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545753088989947e-05,
+      "loss": 0.3168,
+      "step": 10943
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5456695683502747e-05,
+      "loss": 0.3861,
+      "step": 10944
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5455860422898813e-05,
+      "loss": 0.384,
+      "step": 10945
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545502510809596e-05,
+      "loss": 0.3843,
+      "step": 10946
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.54541897391025e-05,
+      "loss": 0.3004,
+      "step": 10947
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5453354315926722e-05,
+      "loss": 0.3549,
+      "step": 10948
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545251883857692e-05,
+      "loss": 0.3056,
+      "step": 10949
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5451683307061407e-05,
+      "loss": 0.3227,
+      "step": 10950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5450847721388474e-05,
+      "loss": 0.3673,
+      "step": 10951
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5450012081566422e-05,
+      "loss": 0.3889,
+      "step": 10952
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544917638760356e-05,
+      "loss": 0.3867,
+      "step": 10953
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544834063950818e-05,
+      "loss": 0.4036,
+      "step": 10954
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544750483728859e-05,
+      "loss": 0.4126,
+      "step": 10955
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5446668980953093e-05,
+      "loss": 0.3052,
+      "step": 10956
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544583307050999e-05,
+      "loss": 0.4426,
+      "step": 10957
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5444997105967586e-05,
+      "loss": 0.4149,
+      "step": 10958
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544416108733419e-05,
+      "loss": 0.4053,
+      "step": 10959
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5443325014618098e-05,
+      "loss": 0.4171,
+      "step": 10960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544248888782762e-05,
+      "loss": 0.4418,
+      "step": 10961
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5441652706971066e-05,
+      "loss": 0.3999,
+      "step": 10962
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5440816472056733e-05,
+      "loss": 0.3447,
+      "step": 10963
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543998018309294e-05,
+      "loss": 0.3708,
+      "step": 10964
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543914384008799e-05,
+      "loss": 0.3858,
+      "step": 10965
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5438307443050185e-05,
+      "loss": 0.4134,
+      "step": 10966
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5437470991987847e-05,
+      "loss": 0.4193,
+      "step": 10967
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5436634486909273e-05,
+      "loss": 0.37,
+      "step": 10968
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5435797927822774e-05,
+      "loss": 0.4571,
+      "step": 10969
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543496131473667e-05,
+      "loss": 0.3216,
+      "step": 10970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5434124647659265e-05,
+      "loss": 0.4267,
+      "step": 10971
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543328792659887e-05,
+      "loss": 0.3942,
+      "step": 10972
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5432451151563797e-05,
+      "loss": 0.4508,
+      "step": 10973
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5431614322562362e-05,
+      "loss": 0.3608,
+      "step": 10974
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5430777439602875e-05,
+      "loss": 0.4056,
+      "step": 10975
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5429940502693653e-05,
+      "loss": 0.423,
+      "step": 10976
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542910351184301e-05,
+      "loss": 0.3703,
+      "step": 10977
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5428266467059256e-05,
+      "loss": 0.3378,
+      "step": 10978
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5427429368350712e-05,
+      "loss": 0.3506,
+      "step": 10979
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5426592215725685e-05,
+      "loss": 0.3464,
+      "step": 10980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54257550091925e-05,
+      "loss": 0.3892,
+      "step": 10981
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542491774875947e-05,
+      "loss": 0.3006,
+      "step": 10982
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5424080434434917e-05,
+      "loss": 0.3945,
+      "step": 10983
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542324306622715e-05,
+      "loss": 0.3672,
+      "step": 10984
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54224056441445e-05,
+      "loss": 0.4864,
+      "step": 10985
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5421568168195275e-05,
+      "loss": 0.3942,
+      "step": 10986
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54207306383878e-05,
+      "loss": 0.36,
+      "step": 10987
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541989305473039e-05,
+      "loss": 0.3836,
+      "step": 10988
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5419055417231374e-05,
+      "loss": 0.387,
+      "step": 10989
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5418217725899064e-05,
+      "loss": 0.3332,
+      "step": 10990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541737998074179e-05,
+      "loss": 0.3702,
+      "step": 10991
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5416542181767865e-05,
+      "loss": 0.3262,
+      "step": 10992
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541570432898562e-05,
+      "loss": 0.3071,
+      "step": 10993
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5414866422403375e-05,
+      "loss": 0.3456,
+      "step": 10994
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5414028462029454e-05,
+      "loss": 0.4048,
+      "step": 10995
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5413190447872184e-05,
+      "loss": 0.4078,
+      "step": 10996
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5412352379939885e-05,
+      "loss": 0.4152,
+      "step": 10997
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5411514258240886e-05,
+      "loss": 0.3747,
+      "step": 10998
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541067608278351e-05,
+      "loss": 0.4118,
+      "step": 10999
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5409837853576088e-05,
+      "loss": 0.4525,
+      "step": 11000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5408999570626942e-05,
+      "loss": 0.4489,
+      "step": 11001
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54081612339444e-05,
+      "loss": 0.4071,
+      "step": 11002
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5407322843536796e-05,
+      "loss": 0.4005,
+      "step": 11003
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5406484399412452e-05,
+      "loss": 0.4477,
+      "step": 11004
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54056459015797e-05,
+      "loss": 0.4418,
+      "step": 11005
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540480735004687e-05,
+      "loss": 0.3718,
+      "step": 11006
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540396874482229e-05,
+      "loss": 0.3822,
+      "step": 11007
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5403130085914298e-05,
+      "loss": 0.3713,
+      "step": 11008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5402291373331213e-05,
+      "loss": 0.4099,
+      "step": 11009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5401452607081378e-05,
+      "loss": 0.3377,
+      "step": 11010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5400613787173114e-05,
+      "loss": 0.3567,
+      "step": 11011
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.539977491361477e-05,
+      "loss": 0.3894,
+      "step": 11012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5398935986414667e-05,
+      "loss": 0.4397,
+      "step": 11013
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.539809700558114e-05,
+      "loss": 0.4341,
+      "step": 11014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5397257971122527e-05,
+      "loss": 0.4012,
+      "step": 11015
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.539641888304716e-05,
+      "loss": 0.3663,
+      "step": 11016
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5395579741363383e-05,
+      "loss": 0.4191,
+      "step": 11017
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5394740546079518e-05,
+      "loss": 0.3576,
+      "step": 11018
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.539390129720391e-05,
+      "loss": 0.373,
+      "step": 11019
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53930619947449e-05,
+      "loss": 0.3761,
+      "step": 11020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5392222638710816e-05,
+      "loss": 0.4391,
+      "step": 11021
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5391383229110005e-05,
+      "loss": 0.3113,
+      "step": 11022
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53905437659508e-05,
+      "loss": 0.376,
+      "step": 11023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538970424924154e-05,
+      "loss": 0.3459,
+      "step": 11024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5388864678990568e-05,
+      "loss": 0.3247,
+      "step": 11025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5388025055206227e-05,
+      "loss": 0.4585,
+      "step": 11026
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538718537789685e-05,
+      "loss": 0.3448,
+      "step": 11027
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5386345647070784e-05,
+      "loss": 0.4435,
+      "step": 11028
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5385505862736366e-05,
+      "loss": 0.4496,
+      "step": 11029
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5384666024901948e-05,
+      "loss": 0.38,
+      "step": 11030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5383826133575867e-05,
+      "loss": 0.3942,
+      "step": 11031
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5382986188766463e-05,
+      "loss": 0.3492,
+      "step": 11032
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5382146190482083e-05,
+      "loss": 0.4193,
+      "step": 11033
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538130613873108e-05,
+      "loss": 0.3832,
+      "step": 11034
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5380466033521783e-05,
+      "loss": 0.336,
+      "step": 11035
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5379625874862546e-05,
+      "loss": 0.3543,
+      "step": 11036
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537878566276172e-05,
+      "loss": 0.3908,
+      "step": 11037
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5377945397227644e-05,
+      "loss": 0.3477,
+      "step": 11038
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537710507826867e-05,
+      "loss": 0.3795,
+      "step": 11039
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5376264705893144e-05,
+      "loss": 0.433,
+      "step": 11040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5375424280109412e-05,
+      "loss": 0.4021,
+      "step": 11041
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5374583800925824e-05,
+      "loss": 0.3459,
+      "step": 11042
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5373743268350737e-05,
+      "loss": 0.3883,
+      "step": 11043
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537290268239249e-05,
+      "loss": 0.4118,
+      "step": 11044
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5372062043059436e-05,
+      "loss": 0.3717,
+      "step": 11045
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5371221350359932e-05,
+      "loss": 0.3857,
+      "step": 11046
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5370380604302324e-05,
+      "loss": 0.4296,
+      "step": 11047
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5369539804894965e-05,
+      "loss": 0.3996,
+      "step": 11048
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536869895214621e-05,
+      "loss": 0.444,
+      "step": 11049
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5367858046064407e-05,
+      "loss": 0.3376,
+      "step": 11050
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5367017086657915e-05,
+      "loss": 0.3919,
+      "step": 11051
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5366176073935087e-05,
+      "loss": 0.3614,
+      "step": 11052
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5365335007904273e-05,
+      "loss": 0.3891,
+      "step": 11053
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5364493888573834e-05,
+      "loss": 0.3756,
+      "step": 11054
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536365271595212e-05,
+      "loss": 0.3879,
+      "step": 11055
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5362811490047498e-05,
+      "loss": 0.3536,
+      "step": 11056
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5361970210868314e-05,
+      "loss": 0.3507,
+      "step": 11057
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536112887842293e-05,
+      "loss": 0.348,
+      "step": 11058
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53602874927197e-05,
+      "loss": 0.423,
+      "step": 11059
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5359446053766985e-05,
+      "loss": 0.3471,
+      "step": 11060
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5358604561573147e-05,
+      "loss": 0.3469,
+      "step": 11061
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535776301614654e-05,
+      "loss": 0.3963,
+      "step": 11062
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5356921417495526e-05,
+      "loss": 0.3355,
+      "step": 11063
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535607976562847e-05,
+      "loss": 0.4943,
+      "step": 11064
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5355238060553727e-05,
+      "loss": 0.4682,
+      "step": 11065
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535439630227966e-05,
+      "loss": 0.3586,
+      "step": 11066
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535355449081463e-05,
+      "loss": 0.4104,
+      "step": 11067
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5352712626167005e-05,
+      "loss": 0.3643,
+      "step": 11068
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5351870708345142e-05,
+      "loss": 0.3744,
+      "step": 11069
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5351028737357406e-05,
+      "loss": 0.4476,
+      "step": 11070
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5350186713212166e-05,
+      "loss": 0.389,
+      "step": 11071
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534934463591778e-05,
+      "loss": 0.2959,
+      "step": 11072
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5348502505482618e-05,
+      "loss": 0.4378,
+      "step": 11073
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534766032191504e-05,
+      "loss": 0.4096,
+      "step": 11074
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5346818085223423e-05,
+      "loss": 0.4032,
+      "step": 11075
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534597579541612e-05,
+      "loss": 0.3186,
+      "step": 11076
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5345133452501514e-05,
+      "loss": 0.3976,
+      "step": 11077
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534429105648796e-05,
+      "loss": 0.4691,
+      "step": 11078
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5343448607383827e-05,
+      "loss": 0.3937,
+      "step": 11079
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534260610519749e-05,
+      "loss": 0.451,
+      "step": 11080
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.534176354993732e-05,
+      "loss": 0.405,
+      "step": 11081
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5340920941611683e-05,
+      "loss": 0.3467,
+      "step": 11082
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5340078280228947e-05,
+      "loss": 0.3378,
+      "step": 11083
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5339235565797486e-05,
+      "loss": 0.3636,
+      "step": 11084
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5338392798325673e-05,
+      "loss": 0.358,
+      "step": 11085
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5337549977821882e-05,
+      "loss": 0.3105,
+      "step": 11086
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533670710429448e-05,
+      "loss": 0.4,
+      "step": 11087
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5335864177751837e-05,
+      "loss": 0.3576,
+      "step": 11088
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5335021198202337e-05,
+      "loss": 0.4547,
+      "step": 11089
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5334178165654348e-05,
+      "loss": 0.4159,
+      "step": 11090
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533333508011625e-05,
+      "loss": 0.3715,
+      "step": 11091
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533249194159641e-05,
+      "loss": 0.3644,
+      "step": 11092
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533164875010321e-05,
+      "loss": 0.4753,
+      "step": 11093
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5330805505645025e-05,
+      "loss": 0.3848,
+      "step": 11094
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532996220823023e-05,
+      "loss": 0.2909,
+      "step": 11095
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.53291188578672e-05,
+      "loss": 0.3971,
+      "step": 11096
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5328275454564323e-05,
+      "loss": 0.3887,
+      "step": 11097
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532743199832997e-05,
+      "loss": 0.5127,
+      "step": 11098
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532658848917252e-05,
+      "loss": 0.4234,
+      "step": 11099
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532574492710035e-05,
+      "loss": 0.4098,
+      "step": 11100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532490131212185e-05,
+      "loss": 0.4121,
+      "step": 11101
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532405764424539e-05,
+      "loss": 0.4314,
+      "step": 11102
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5323213923479356e-05,
+      "loss": 0.3699,
+      "step": 11103
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5322370149832128e-05,
+      "loss": 0.3567,
+      "step": 11104
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532152632331209e-05,
+      "loss": 0.4013,
+      "step": 11105
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532068244392762e-05,
+      "loss": 0.3053,
+      "step": 11106
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5319838511687107e-05,
+      "loss": 0.5016,
+      "step": 11107
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5318994526598934e-05,
+      "loss": 0.4338,
+      "step": 11108
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5318150488671483e-05,
+      "loss": 0.4561,
+      "step": 11109
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317306397913137e-05,
+      "loss": 0.3722,
+      "step": 11110
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5316462254332285e-05,
+      "loss": 0.4197,
+      "step": 11111
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5315618057937313e-05,
+      "loss": 0.4237,
+      "step": 11112
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.53147738087366e-05,
+      "loss": 0.3894,
+      "step": 11113
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5313929506738543e-05,
+      "loss": 0.3478,
+      "step": 11114
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5313085151951527e-05,
+      "loss": 0.4075,
+      "step": 11115
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5312240744383933e-05,
+      "loss": 0.4604,
+      "step": 11116
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5311396284044153e-05,
+      "loss": 0.3531,
+      "step": 11117
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.531055177094058e-05,
+      "loss": 0.3426,
+      "step": 11118
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5309707205081597e-05,
+      "loss": 0.3548,
+      "step": 11119
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5308862586475602e-05,
+      "loss": 0.3617,
+      "step": 11120
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530801791513098e-05,
+      "loss": 0.3825,
+      "step": 11121
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5307173191056122e-05,
+      "loss": 0.3876,
+      "step": 11122
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5306328414259423e-05,
+      "loss": 0.3231,
+      "step": 11123
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5305483584749268e-05,
+      "loss": 0.3819,
+      "step": 11124
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5304638702534056e-05,
+      "loss": 0.3754,
+      "step": 11125
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530379376762218e-05,
+      "loss": 0.3117,
+      "step": 11126
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530294878002203e-05,
+      "loss": 0.4478,
+      "step": 11127
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5302103739742003e-05,
+      "loss": 0.396,
+      "step": 11128
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530125864679049e-05,
+      "loss": 0.3555,
+      "step": 11129
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5300413501175892e-05,
+      "loss": 0.3998,
+      "step": 11130
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5299568302906603e-05,
+      "loss": 0.3676,
+      "step": 11131
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5298723051991014e-05,
+      "loss": 0.3911,
+      "step": 11132
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529787774843753e-05,
+      "loss": 0.3603,
+      "step": 11133
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5297032392254538e-05,
+      "loss": 0.3363,
+      "step": 11134
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529618698345045e-05,
+      "loss": 0.3837,
+      "step": 11135
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529534152203365e-05,
+      "loss": 0.46,
+      "step": 11136
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529449600801255e-05,
+      "loss": 0.4539,
+      "step": 11137
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5293650441395532e-05,
+      "loss": 0.5197,
+      "step": 11138
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5292804822191014e-05,
+      "loss": 0.4323,
+      "step": 11139
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5291959150407386e-05,
+      "loss": 0.3987,
+      "step": 11140
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5291113426053055e-05,
+      "loss": 0.3698,
+      "step": 11141
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5290267649136413e-05,
+      "loss": 0.3995,
+      "step": 11142
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5289421819665876e-05,
+      "loss": 0.4317,
+      "step": 11143
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5288575937649836e-05,
+      "loss": 0.4399,
+      "step": 11144
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.52877300030967e-05,
+      "loss": 0.4372,
+      "step": 11145
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5286884016014866e-05,
+      "loss": 0.3862,
+      "step": 11146
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5286037976412744e-05,
+      "loss": 0.4229,
+      "step": 11147
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5285191884298744e-05,
+      "loss": 0.402,
+      "step": 11148
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528434573968126e-05,
+      "loss": 0.3656,
+      "step": 11149
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5283499542568697e-05,
+      "loss": 0.4151,
+      "step": 11150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5282653292969472e-05,
+      "loss": 0.4262,
+      "step": 11151
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5281806990891986e-05,
+      "loss": 0.4097,
+      "step": 11152
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528096063634464e-05,
+      "loss": 0.3874,
+      "step": 11153
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5280114229335856e-05,
+      "loss": 0.3649,
+      "step": 11154
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5279267769874034e-05,
+      "loss": 0.3718,
+      "step": 11155
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5278421257967574e-05,
+      "loss": 0.3269,
+      "step": 11156
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5277574693624902e-05,
+      "loss": 0.4391,
+      "step": 11157
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527672807685442e-05,
+      "loss": 0.3508,
+      "step": 11158
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5275881407664535e-05,
+      "loss": 0.4319,
+      "step": 11159
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5275034686063665e-05,
+      "loss": 0.4082,
+      "step": 11160
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5274187912060213e-05,
+      "loss": 0.4581,
+      "step": 11161
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5273341085662597e-05,
+      "loss": 0.3617,
+      "step": 11162
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5272494206879232e-05,
+      "loss": 0.448,
+      "step": 11163
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527164727571852e-05,
+      "loss": 0.3925,
+      "step": 11164
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527080029218889e-05,
+      "loss": 0.3817,
+      "step": 11165
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5269953256298742e-05,
+      "loss": 0.3781,
+      "step": 11166
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.52691061680565e-05,
+      "loss": 0.396,
+      "step": 11167
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5268259027470568e-05,
+      "loss": 0.4681,
+      "step": 11168
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5267411834549375e-05,
+      "loss": 0.4226,
+      "step": 11169
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526656458930133e-05,
+      "loss": 0.3768,
+      "step": 11170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5265717291734846e-05,
+      "loss": 0.3675,
+      "step": 11171
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526486994185835e-05,
+      "loss": 0.391,
+      "step": 11172
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526402253968025e-05,
+      "loss": 0.3145,
+      "step": 11173
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526317508520897e-05,
+      "loss": 0.3743,
+      "step": 11174
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5262327578452927e-05,
+      "loss": 0.3822,
+      "step": 11175
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526148001942054e-05,
+      "loss": 0.3072,
+      "step": 11176
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5260632408120225e-05,
+      "loss": 0.4676,
+      "step": 11177
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525978474456041e-05,
+      "loss": 0.3373,
+      "step": 11178
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5258937028749513e-05,
+      "loss": 0.3977,
+      "step": 11179
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525808926069595e-05,
+      "loss": 0.3347,
+      "step": 11180
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5257241440408148e-05,
+      "loss": 0.3638,
+      "step": 11181
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5256393567894531e-05,
+      "loss": 0.3465,
+      "step": 11182
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5255545643163516e-05,
+      "loss": 0.367,
+      "step": 11183
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525469766622353e-05,
+      "loss": 0.3268,
+      "step": 11184
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5253849637082997e-05,
+      "loss": 0.4489,
+      "step": 11185
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525300155575034e-05,
+      "loss": 0.3404,
+      "step": 11186
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5252153422233984e-05,
+      "loss": 0.3973,
+      "step": 11187
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5251305236542356e-05,
+      "loss": 0.4866,
+      "step": 11188
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5250456998683882e-05,
+      "loss": 0.4021,
+      "step": 11189
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5249608708666984e-05,
+      "loss": 0.3519,
+      "step": 11190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5248760366500093e-05,
+      "loss": 0.3177,
+      "step": 11191
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5247911972191639e-05,
+      "loss": 0.4176,
+      "step": 11192
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.524706352575004e-05,
+      "loss": 0.3878,
+      "step": 11193
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5246215027183738e-05,
+      "loss": 0.392,
+      "step": 11194
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5245366476501151e-05,
+      "loss": 0.437,
+      "step": 11195
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5244517873710715e-05,
+      "loss": 0.4277,
+      "step": 11196
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5243669218820856e-05,
+      "loss": 0.3616,
+      "step": 11197
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5242820511840011e-05,
+      "loss": 0.4232,
+      "step": 11198
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5241971752776601e-05,
+      "loss": 0.4287,
+      "step": 11199
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5241122941639064e-05,
+      "loss": 0.3965,
+      "step": 11200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5240274078435836e-05,
+      "loss": 0.358,
+      "step": 11201
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5239425163175341e-05,
+      "loss": 0.3529,
+      "step": 11202
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523857619586602e-05,
+      "loss": 0.3565,
+      "step": 11203
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.52377271765163e-05,
+      "loss": 0.3716,
+      "step": 11204
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5236878105134617e-05,
+      "loss": 0.3623,
+      "step": 11205
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523602898172941e-05,
+      "loss": 0.3509,
+      "step": 11206
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5235179806309111e-05,
+      "loss": 0.4696,
+      "step": 11207
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5234330578882155e-05,
+      "loss": 0.3499,
+      "step": 11208
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5233481299456978e-05,
+      "loss": 0.3718,
+      "step": 11209
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523263196804202e-05,
+      "loss": 0.4036,
+      "step": 11210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5231782584645716e-05,
+      "loss": 0.3767,
+      "step": 11211
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5230933149276503e-05,
+      "loss": 0.3625,
+      "step": 11212
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5230083661942825e-05,
+      "loss": 0.3523,
+      "step": 11213
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.522923412265311e-05,
+      "loss": 0.3493,
+      "step": 11214
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5228384531415809e-05,
+      "loss": 0.3284,
+      "step": 11215
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5227534888239354e-05,
+      "loss": 0.4224,
+      "step": 11216
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5226685193132189e-05,
+      "loss": 0.3472,
+      "step": 11217
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5225835446102753e-05,
+      "loss": 0.3445,
+      "step": 11218
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5224985647159489e-05,
+      "loss": 0.4111,
+      "step": 11219
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5224135796310839e-05,
+      "loss": 0.3465,
+      "step": 11220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5223285893565245e-05,
+      "loss": 0.4138,
+      "step": 11221
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.522243593893115e-05,
+      "loss": 0.4588,
+      "step": 11222
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5221585932416998e-05,
+      "loss": 0.4146,
+      "step": 11223
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5220735874031234e-05,
+      "loss": 0.3642,
+      "step": 11224
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5219885763782301e-05,
+      "loss": 0.289,
+      "step": 11225
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.521903560167864e-05,
+      "loss": 0.4076,
+      "step": 11226
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5218185387728703e-05,
+      "loss": 0.3184,
+      "step": 11227
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5217335121940933e-05,
+      "loss": 0.3628,
+      "step": 11228
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.521648480432378e-05,
+      "loss": 0.3717,
+      "step": 11229
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5215634434885684e-05,
+      "loss": 0.4758,
+      "step": 11230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.52147840136351e-05,
+      "loss": 0.3814,
+      "step": 11231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213933540580473e-05,
+      "loss": 0.3505,
+      "step": 11232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213083015730253e-05,
+      "loss": 0.4306,
+      "step": 11233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5212232439092885e-05,
+      "loss": 0.46,
+      "step": 11234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5211381810676826e-05,
+      "loss": 0.4518,
+      "step": 11235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5210531130490517e-05,
+      "loss": 0.4106,
+      "step": 11236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5209680398542418e-05,
+      "loss": 0.3321,
+      "step": 11237
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5208829614840976e-05,
+      "loss": 0.4146,
+      "step": 11238
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.520797877939464e-05,
+      "loss": 0.4184,
+      "step": 11239
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5207127892211868e-05,
+      "loss": 0.3733,
+      "step": 11240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5206276953301106e-05,
+      "loss": 0.4068,
+      "step": 11241
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5205425962670814e-05,
+      "loss": 0.3301,
+      "step": 11242
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5204574920329443e-05,
+      "loss": 0.4183,
+      "step": 11243
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5203723826285445e-05,
+      "loss": 0.4104,
+      "step": 11244
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5202872680547276e-05,
+      "loss": 0.3923,
+      "step": 11245
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5202021483123395e-05,
+      "loss": 0.3695,
+      "step": 11246
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5201170234022256e-05,
+      "loss": 0.3586,
+      "step": 11247
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.520031893325231e-05,
+      "loss": 0.4193,
+      "step": 11248
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5199467580822026e-05,
+      "loss": 0.4306,
+      "step": 11249
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5198616176739846e-05,
+      "loss": 0.3524,
+      "step": 11250
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5197764721014238e-05,
+      "loss": 0.362,
+      "step": 11251
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5196913213653662e-05,
+      "loss": 0.4092,
+      "step": 11252
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.519606165466657e-05,
+      "loss": 0.4324,
+      "step": 11253
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5195210044061423e-05,
+      "loss": 0.3631,
+      "step": 11254
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5194358381846686e-05,
+      "loss": 0.3458,
+      "step": 11255
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5193506668030815e-05,
+      "loss": 0.4294,
+      "step": 11256
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5192654902622268e-05,
+      "loss": 0.4045,
+      "step": 11257
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5191803085629516e-05,
+      "loss": 0.4354,
+      "step": 11258
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5190951217061014e-05,
+      "loss": 0.4394,
+      "step": 11259
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5190099296925222e-05,
+      "loss": 0.3512,
+      "step": 11260
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5189247325230611e-05,
+      "loss": 0.5123,
+      "step": 11261
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5188395301985641e-05,
+      "loss": 0.3868,
+      "step": 11262
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5187543227198776e-05,
+      "loss": 0.3807,
+      "step": 11263
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5186691100878478e-05,
+      "loss": 0.3404,
+      "step": 11264
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5185838923033216e-05,
+      "loss": 0.3847,
+      "step": 11265
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.518498669367145e-05,
+      "loss": 0.4564,
+      "step": 11266
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5184134412801655e-05,
+      "loss": 0.424,
+      "step": 11267
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5183282080432289e-05,
+      "loss": 0.4147,
+      "step": 11268
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5182429696571825e-05,
+      "loss": 0.4071,
+      "step": 11269
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5181577261228726e-05,
+      "loss": 0.349,
+      "step": 11270
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5180724774411469e-05,
+      "loss": 0.3478,
+      "step": 11271
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.517987223612851e-05,
+      "loss": 0.379,
+      "step": 11272
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5179019646388327e-05,
+      "loss": 0.4658,
+      "step": 11273
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5178167005199384e-05,
+      "loss": 0.325,
+      "step": 11274
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5177314312570155e-05,
+      "loss": 0.5253,
+      "step": 11275
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5176461568509111e-05,
+      "loss": 0.3685,
+      "step": 11276
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5175608773024724e-05,
+      "loss": 0.5464,
+      "step": 11277
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5174755926125465e-05,
+      "loss": 0.4131,
+      "step": 11278
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5173903027819798e-05,
+      "loss": 0.3315,
+      "step": 11279
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.517305007811621e-05,
+      "loss": 0.4143,
+      "step": 11280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5172197077023168e-05,
+      "loss": 0.3986,
+      "step": 11281
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.517134402454914e-05,
+      "loss": 0.3157,
+      "step": 11282
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5170490920702605e-05,
+      "loss": 0.3785,
+      "step": 11283
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5169637765492042e-05,
+      "loss": 0.4004,
+      "step": 11284
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516878455892592e-05,
+      "loss": 0.3825,
+      "step": 11285
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516793130101272e-05,
+      "loss": 0.3686,
+      "step": 11286
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516707799176091e-05,
+      "loss": 0.3954,
+      "step": 11287
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5166224631178977e-05,
+      "loss": 0.421,
+      "step": 11288
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5165371219275393e-05,
+      "loss": 0.4932,
+      "step": 11289
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5164517756058636e-05,
+      "loss": 0.4187,
+      "step": 11290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5163664241537184e-05,
+      "loss": 0.3313,
+      "step": 11291
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5162810675719518e-05,
+      "loss": 0.4052,
+      "step": 11292
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5161957058614116e-05,
+      "loss": 0.4056,
+      "step": 11293
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5161103390229457e-05,
+      "loss": 0.4014,
+      "step": 11294
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5160249670574026e-05,
+      "loss": 0.3976,
+      "step": 11295
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5159395899656298e-05,
+      "loss": 0.3777,
+      "step": 11296
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5158542077484755e-05,
+      "loss": 0.3987,
+      "step": 11297
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5157688204067882e-05,
+      "loss": 0.3699,
+      "step": 11298
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5156834279414162e-05,
+      "loss": 0.3068,
+      "step": 11299
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5155980303532074e-05,
+      "loss": 0.3555,
+      "step": 11300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5155126276430103e-05,
+      "loss": 0.366,
+      "step": 11301
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5154272198116737e-05,
+      "loss": 0.3432,
+      "step": 11302
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5153418068600454e-05,
+      "loss": 0.4759,
+      "step": 11303
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5152563887889742e-05,
+      "loss": 0.4762,
+      "step": 11304
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515170965599309e-05,
+      "loss": 0.4186,
+      "step": 11305
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5150855372918979e-05,
+      "loss": 0.3678,
+      "step": 11306
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5150001038675894e-05,
+      "loss": 0.3863,
+      "step": 11307
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5149146653272327e-05,
+      "loss": 0.3448,
+      "step": 11308
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5148292216716764e-05,
+      "loss": 0.3365,
+      "step": 11309
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5147437729017691e-05,
+      "loss": 0.3406,
+      "step": 11310
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.51465831901836e-05,
+      "loss": 0.3769,
+      "step": 11311
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5145728600222977e-05,
+      "loss": 0.3386,
+      "step": 11312
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5144873959144313e-05,
+      "loss": 0.4054,
+      "step": 11313
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.51440192669561e-05,
+      "loss": 0.4016,
+      "step": 11314
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5143164523666824e-05,
+      "loss": 0.4188,
+      "step": 11315
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5142309729284978e-05,
+      "loss": 0.3813,
+      "step": 11316
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5141454883819058e-05,
+      "loss": 0.3749,
+      "step": 11317
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5140599987277549e-05,
+      "loss": 0.389,
+      "step": 11318
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5139745039668945e-05,
+      "loss": 0.4563,
+      "step": 11319
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5138890041001744e-05,
+      "loss": 0.3879,
+      "step": 11320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5138034991284433e-05,
+      "loss": 0.417,
+      "step": 11321
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5137179890525513e-05,
+      "loss": 0.4013,
+      "step": 11322
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5136324738733475e-05,
+      "loss": 0.3843,
+      "step": 11323
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5135469535916815e-05,
+      "loss": 0.3754,
+      "step": 11324
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5134614282084025e-05,
+      "loss": 0.2863,
+      "step": 11325
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5133758977243606e-05,
+      "loss": 0.3273,
+      "step": 11326
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5132903621404053e-05,
+      "loss": 0.4465,
+      "step": 11327
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5132048214573862e-05,
+      "loss": 0.3485,
+      "step": 11328
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5131192756761534e-05,
+      "loss": 0.3929,
+      "step": 11329
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5130337247975564e-05,
+      "loss": 0.4304,
+      "step": 11330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5129481688224452e-05,
+      "loss": 0.4667,
+      "step": 11331
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5128626077516695e-05,
+      "loss": 0.3946,
+      "step": 11332
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5127770415860794e-05,
+      "loss": 0.3946,
+      "step": 11333
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5126914703265251e-05,
+      "loss": 0.3321,
+      "step": 11334
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5126058939738566e-05,
+      "loss": 0.3893,
+      "step": 11335
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5125203125289241e-05,
+      "loss": 0.344,
+      "step": 11336
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5124347259925772e-05,
+      "loss": 0.4166,
+      "step": 11337
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5123491343656666e-05,
+      "loss": 0.4671,
+      "step": 11338
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5122635376490432e-05,
+      "loss": 0.39,
+      "step": 11339
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.512177935843556e-05,
+      "loss": 0.4164,
+      "step": 11340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5120923289500561e-05,
+      "loss": 0.3976,
+      "step": 11341
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5120067169693941e-05,
+      "loss": 0.3886,
+      "step": 11342
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5119210999024199e-05,
+      "loss": 0.3297,
+      "step": 11343
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5118354777499847e-05,
+      "loss": 0.3669,
+      "step": 11344
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5117498505129384e-05,
+      "loss": 0.3765,
+      "step": 11345
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5116642181921322e-05,
+      "loss": 0.3355,
+      "step": 11346
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5115785807884163e-05,
+      "loss": 0.4355,
+      "step": 11347
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.511492938302642e-05,
+      "loss": 0.3601,
+      "step": 11348
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5114072907356596e-05,
+      "loss": 0.4015,
+      "step": 11349
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5113216380883198e-05,
+      "loss": 0.4024,
+      "step": 11350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5112359803614742e-05,
+      "loss": 0.4283,
+      "step": 11351
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5111503175559729e-05,
+      "loss": 0.5227,
+      "step": 11352
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5110646496726675e-05,
+      "loss": 0.2428,
+      "step": 11353
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5109789767124085e-05,
+      "loss": 0.395,
+      "step": 11354
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510893298676048e-05,
+      "loss": 0.3995,
+      "step": 11355
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5108076155644355e-05,
+      "loss": 0.475,
+      "step": 11356
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5107219273784236e-05,
+      "loss": 0.3458,
+      "step": 11357
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510636234118863e-05,
+      "loss": 0.2877,
+      "step": 11358
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5105505357866048e-05,
+      "loss": 0.4065,
+      "step": 11359
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5104648323825006e-05,
+      "loss": 0.3912,
+      "step": 11360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5103791239074017e-05,
+      "loss": 0.3958,
+      "step": 11361
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5102934103621596e-05,
+      "loss": 0.4269,
+      "step": 11362
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5102076917476258e-05,
+      "loss": 0.4075,
+      "step": 11363
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5101219680646519e-05,
+      "loss": 0.4337,
+      "step": 11364
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5100362393140891e-05,
+      "loss": 0.3885,
+      "step": 11365
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5099505054967893e-05,
+      "loss": 0.421,
+      "step": 11366
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5098647666136046e-05,
+      "loss": 0.4223,
+      "step": 11367
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5097790226653857e-05,
+      "loss": 0.3955,
+      "step": 11368
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5096932736529853e-05,
+      "loss": 0.297,
+      "step": 11369
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5096075195772551e-05,
+      "loss": 0.3388,
+      "step": 11370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5095217604390463e-05,
+      "loss": 0.4348,
+      "step": 11371
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509435996239212e-05,
+      "loss": 0.384,
+      "step": 11372
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509350226978603e-05,
+      "loss": 0.4113,
+      "step": 11373
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5092644526580722e-05,
+      "loss": 0.3726,
+      "step": 11374
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5091786732784717e-05,
+      "loss": 0.4324,
+      "step": 11375
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5090928888406527e-05,
+      "loss": 0.3548,
+      "step": 11376
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5090070993454684e-05,
+      "loss": 0.3874,
+      "step": 11377
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5089213047937703e-05,
+      "loss": 0.3632,
+      "step": 11378
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5088355051864115e-05,
+      "loss": 0.406,
+      "step": 11379
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5087497005242435e-05,
+      "loss": 0.3854,
+      "step": 11380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5086638908081193e-05,
+      "loss": 0.4076,
+      "step": 11381
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5085780760388911e-05,
+      "loss": 0.3841,
+      "step": 11382
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5084922562174109e-05,
+      "loss": 0.3115,
+      "step": 11383
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5084064313445326e-05,
+      "loss": 0.4246,
+      "step": 11384
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5083206014211071e-05,
+      "loss": 0.4208,
+      "step": 11385
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5082347664479885e-05,
+      "loss": 0.4206,
+      "step": 11386
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5081489264260284e-05,
+      "loss": 0.3289,
+      "step": 11387
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5080630813560802e-05,
+      "loss": 0.4204,
+      "step": 11388
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5079772312389965e-05,
+      "loss": 0.3928,
+      "step": 11389
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5078913760756302e-05,
+      "loss": 0.3734,
+      "step": 11390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5078055158668339e-05,
+      "loss": 0.4335,
+      "step": 11391
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5077196506134608e-05,
+      "loss": 0.3222,
+      "step": 11392
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5076337803163638e-05,
+      "loss": 0.4107,
+      "step": 11393
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5075479049763966e-05,
+      "loss": 0.3582,
+      "step": 11394
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5074620245944112e-05,
+      "loss": 0.402,
+      "step": 11395
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5073761391712612e-05,
+      "loss": 0.3493,
+      "step": 11396
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5072902487077999e-05,
+      "loss": 0.3692,
+      "step": 11397
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5072043532048804e-05,
+      "loss": 0.3258,
+      "step": 11398
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.507118452663356e-05,
+      "loss": 0.3622,
+      "step": 11399
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5070325470840803e-05,
+      "loss": 0.3804,
+      "step": 11400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5069466364679064e-05,
+      "loss": 0.3652,
+      "step": 11401
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5068607208156877e-05,
+      "loss": 0.459,
+      "step": 11402
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.506774800128278e-05,
+      "loss": 0.4599,
+      "step": 11403
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5066888744065308e-05,
+      "loss": 0.3619,
+      "step": 11404
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5066029436512992e-05,
+      "loss": 0.354,
+      "step": 11405
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5065170078634377e-05,
+      "loss": 0.3199,
+      "step": 11406
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5064310670437992e-05,
+      "loss": 0.3366,
+      "step": 11407
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5063451211932373e-05,
+      "loss": 0.3841,
+      "step": 11408
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5062591703126069e-05,
+      "loss": 0.32,
+      "step": 11409
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.506173214402761e-05,
+      "loss": 0.2866,
+      "step": 11410
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5060872534645535e-05,
+      "loss": 0.3628,
+      "step": 11411
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5060012874988386e-05,
+      "loss": 0.3737,
+      "step": 11412
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5059153165064703e-05,
+      "loss": 0.3887,
+      "step": 11413
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5058293404883025e-05,
+      "loss": 0.3876,
+      "step": 11414
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5057433594451891e-05,
+      "loss": 0.4038,
+      "step": 11415
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5056573733779848e-05,
+      "loss": 0.4251,
+      "step": 11416
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5055713822875433e-05,
+      "loss": 0.4419,
+      "step": 11417
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5054853861747192e-05,
+      "loss": 0.5064,
+      "step": 11418
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5053993850403663e-05,
+      "loss": 0.487,
+      "step": 11419
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5053133788853392e-05,
+      "loss": 0.3414,
+      "step": 11420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5052273677104925e-05,
+      "loss": 0.3739,
+      "step": 11421
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5051413515166806e-05,
+      "loss": 0.3582,
+      "step": 11422
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5050553303047576e-05,
+      "loss": 0.4223,
+      "step": 11423
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5049693040755787e-05,
+      "loss": 0.3953,
+      "step": 11424
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5048832728299978e-05,
+      "loss": 0.3684,
+      "step": 11425
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5047972365688698e-05,
+      "loss": 0.4057,
+      "step": 11426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5047111952930496e-05,
+      "loss": 0.4524,
+      "step": 11427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5046251490033918e-05,
+      "loss": 0.4452,
+      "step": 11428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.504539097700751e-05,
+      "loss": 0.3502,
+      "step": 11429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5044530413859822e-05,
+      "loss": 0.3848,
+      "step": 11430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5043669800599406e-05,
+      "loss": 0.4286,
+      "step": 11431
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5042809137234802e-05,
+      "loss": 0.415,
+      "step": 11432
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.504194842377457e-05,
+      "loss": 0.3766,
+      "step": 11433
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5041087660227258e-05,
+      "loss": 0.3957,
+      "step": 11434
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5040226846601413e-05,
+      "loss": 0.4542,
+      "step": 11435
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503936598290559e-05,
+      "loss": 0.3796,
+      "step": 11436
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503850506914834e-05,
+      "loss": 0.3798,
+      "step": 11437
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5037644105338217e-05,
+      "loss": 0.2813,
+      "step": 11438
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5036783091483766e-05,
+      "loss": 0.4009,
+      "step": 11439
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503592202759355e-05,
+      "loss": 0.3838,
+      "step": 11440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503506091367612e-05,
+      "loss": 0.413,
+      "step": 11441
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5034199749740029e-05,
+      "loss": 0.4019,
+      "step": 11442
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5033338535793831e-05,
+      "loss": 0.4607,
+      "step": 11443
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5032477271846084e-05,
+      "loss": 0.3882,
+      "step": 11444
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5031615957905343e-05,
+      "loss": 0.4422,
+      "step": 11445
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5030754593980161e-05,
+      "loss": 0.3952,
+      "step": 11446
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5029893180079102e-05,
+      "loss": 0.4631,
+      "step": 11447
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5029031716210719e-05,
+      "loss": 0.3463,
+      "step": 11448
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5028170202383567e-05,
+      "loss": 0.4524,
+      "step": 11449
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.502730863860621e-05,
+      "loss": 0.3789,
+      "step": 11450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5026447024887203e-05,
+      "loss": 0.4696,
+      "step": 11451
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.502558536123511e-05,
+      "loss": 0.3766,
+      "step": 11452
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5024723647658483e-05,
+      "loss": 0.3853,
+      "step": 11453
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5023861884165886e-05,
+      "loss": 0.3974,
+      "step": 11454
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5023000070765886e-05,
+      "loss": 0.3993,
+      "step": 11455
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5022138207467034e-05,
+      "loss": 0.3832,
+      "step": 11456
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5021276294277898e-05,
+      "loss": 0.3634,
+      "step": 11457
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5020414331207041e-05,
+      "loss": 0.4227,
+      "step": 11458
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5019552318263023e-05,
+      "loss": 0.3619,
+      "step": 11459
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5018690255454405e-05,
+      "loss": 0.4029,
+      "step": 11460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5017828142789755e-05,
+      "loss": 0.3414,
+      "step": 11461
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.501696598027764e-05,
+      "loss": 0.4446,
+      "step": 11462
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5016103767926616e-05,
+      "loss": 0.3632,
+      "step": 11463
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5015241505745257e-05,
+      "loss": 0.3397,
+      "step": 11464
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5014379193742123e-05,
+      "loss": 0.3365,
+      "step": 11465
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5013516831925783e-05,
+      "loss": 0.4195,
+      "step": 11466
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5012654420304804e-05,
+      "loss": 0.3968,
+      "step": 11467
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5011791958887752e-05,
+      "loss": 0.3803,
+      "step": 11468
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5010929447683191e-05,
+      "loss": 0.472,
+      "step": 11469
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.50100668866997e-05,
+      "loss": 0.3616,
+      "step": 11470
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5009204275945837e-05,
+      "loss": 0.4148,
+      "step": 11471
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5008341615430176e-05,
+      "loss": 0.3695,
+      "step": 11472
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5007478905161286e-05,
+      "loss": 0.3582,
+      "step": 11473
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5006616145147739e-05,
+      "loss": 0.469,
+      "step": 11474
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.50057533353981e-05,
+      "loss": 0.3551,
+      "step": 11475
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5004890475920947e-05,
+      "loss": 0.3811,
+      "step": 11476
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5004027566724849e-05,
+      "loss": 0.4382,
+      "step": 11477
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5003164607818377e-05,
+      "loss": 0.4154,
+      "step": 11478
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5002301599210106e-05,
+      "loss": 0.3637,
+      "step": 11479
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5001438540908607e-05,
+      "loss": 0.3527,
+      "step": 11480
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5000575432922453e-05,
+      "loss": 0.3595,
+      "step": 11481
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4999712275260222e-05,
+      "loss": 0.4262,
+      "step": 11482
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4998849067930487e-05,
+      "loss": 0.3944,
+      "step": 11483
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4997985810941824e-05,
+      "loss": 0.4485,
+      "step": 11484
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4997122504302803e-05,
+      "loss": 0.3877,
+      "step": 11485
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4996259148022008e-05,
+      "loss": 0.4075,
+      "step": 11486
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4995395742108012e-05,
+      "loss": 0.3398,
+      "step": 11487
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.499453228656939e-05,
+      "loss": 0.48,
+      "step": 11488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4993668781414724e-05,
+      "loss": 0.3178,
+      "step": 11489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.499280522665259e-05,
+      "loss": 0.3707,
+      "step": 11490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4991941622291564e-05,
+      "loss": 0.3781,
+      "step": 11491
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4991077968340229e-05,
+      "loss": 0.3771,
+      "step": 11492
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4990214264807166e-05,
+      "loss": 0.3629,
+      "step": 11493
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4989350511700949e-05,
+      "loss": 0.3911,
+      "step": 11494
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4988486709030166e-05,
+      "loss": 0.318,
+      "step": 11495
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4987622856803391e-05,
+      "loss": 0.3867,
+      "step": 11496
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4986758955029208e-05,
+      "loss": 0.3372,
+      "step": 11497
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4985895003716203e-05,
+      "loss": 0.4131,
+      "step": 11498
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4985031002872952e-05,
+      "loss": 0.3683,
+      "step": 11499
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4984166952508043e-05,
+      "loss": 0.3671,
+      "step": 11500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4983302852630058e-05,
+      "loss": 0.4141,
+      "step": 11501
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.498243870324758e-05,
+      "loss": 0.3872,
+      "step": 11502
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4981574504369193e-05,
+      "loss": 0.4622,
+      "step": 11503
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4980710256003485e-05,
+      "loss": 0.322,
+      "step": 11504
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4979845958159045e-05,
+      "loss": 0.4795,
+      "step": 11505
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4978981610844447e-05,
+      "loss": 0.3949,
+      "step": 11506
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4978117214068287e-05,
+      "loss": 0.4783,
+      "step": 11507
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497725276783915e-05,
+      "loss": 0.3764,
+      "step": 11508
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4976388272165623e-05,
+      "loss": 0.4479,
+      "step": 11509
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497552372705629e-05,
+      "loss": 0.3426,
+      "step": 11510
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4974659132519748e-05,
+      "loss": 0.3924,
+      "step": 11511
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497379448856458e-05,
+      "loss": 0.3355,
+      "step": 11512
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4972929795199375e-05,
+      "loss": 0.4438,
+      "step": 11513
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4972065052432726e-05,
+      "loss": 0.3482,
+      "step": 11514
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4971200260273223e-05,
+      "loss": 0.4454,
+      "step": 11515
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4970335418729454e-05,
+      "loss": 0.3844,
+      "step": 11516
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4969470527810015e-05,
+      "loss": 0.4337,
+      "step": 11517
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4968605587523495e-05,
+      "loss": 0.3988,
+      "step": 11518
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4967740597878482e-05,
+      "loss": 0.4273,
+      "step": 11519
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4966875558883578e-05,
+      "loss": 0.4185,
+      "step": 11520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4966010470547373e-05,
+      "loss": 0.4725,
+      "step": 11521
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4965145332878454e-05,
+      "loss": 0.3981,
+      "step": 11522
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4964280145885425e-05,
+      "loss": 0.345,
+      "step": 11523
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4963414909576877e-05,
+      "loss": 0.3677,
+      "step": 11524
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4962549623961401e-05,
+      "loss": 0.3383,
+      "step": 11525
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4961684289047602e-05,
+      "loss": 0.3886,
+      "step": 11526
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4960818904844073e-05,
+      "loss": 0.2803,
+      "step": 11527
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4959953471359402e-05,
+      "loss": 0.4471,
+      "step": 11528
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4959087988602199e-05,
+      "loss": 0.4906,
+      "step": 11529
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4958222456581056e-05,
+      "loss": 0.4053,
+      "step": 11530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4957356875304566e-05,
+      "loss": 0.3674,
+      "step": 11531
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.495649124478134e-05,
+      "loss": 0.3462,
+      "step": 11532
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4955625565019966e-05,
+      "loss": 0.3344,
+      "step": 11533
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.495475983602905e-05,
+      "loss": 0.3708,
+      "step": 11534
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.495389405781719e-05,
+      "loss": 0.4173,
+      "step": 11535
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4953028230392988e-05,
+      "loss": 0.4295,
+      "step": 11536
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4952162353765038e-05,
+      "loss": 0.3851,
+      "step": 11537
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4951296427941952e-05,
+      "loss": 0.413,
+      "step": 11538
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4950430452932333e-05,
+      "loss": 0.3616,
+      "step": 11539
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4949564428744772e-05,
+      "loss": 0.4181,
+      "step": 11540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4948698355387883e-05,
+      "loss": 0.3645,
+      "step": 11541
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4947832232870262e-05,
+      "loss": 0.4613,
+      "step": 11542
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4946966061200519e-05,
+      "loss": 0.385,
+      "step": 11543
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4946099840387259e-05,
+      "loss": 0.4361,
+      "step": 11544
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4945233570439079e-05,
+      "loss": 0.4011,
+      "step": 11545
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4944367251364591e-05,
+      "loss": 0.4246,
+      "step": 11546
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4943500883172401e-05,
+      "loss": 0.3977,
+      "step": 11547
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4942634465871116e-05,
+      "loss": 0.3817,
+      "step": 11548
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494176799946934e-05,
+      "loss": 0.3003,
+      "step": 11549
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494090148397568e-05,
+      "loss": 0.4267,
+      "step": 11550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4940034919398752e-05,
+      "loss": 0.311,
+      "step": 11551
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4939168305747154e-05,
+      "loss": 0.3401,
+      "step": 11552
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4938301643029502e-05,
+      "loss": 0.3854,
+      "step": 11553
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4937434931254402e-05,
+      "loss": 0.3589,
+      "step": 11554
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4936568170430465e-05,
+      "loss": 0.3468,
+      "step": 11555
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.49357013605663e-05,
+      "loss": 0.3729,
+      "step": 11556
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4934834501670523e-05,
+      "loss": 0.3837,
+      "step": 11557
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4933967593751742e-05,
+      "loss": 0.3591,
+      "step": 11558
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4933100636818568e-05,
+      "loss": 0.4054,
+      "step": 11559
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4932233630879613e-05,
+      "loss": 0.379,
+      "step": 11560
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493136657594349e-05,
+      "loss": 0.4461,
+      "step": 11561
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4930499472018818e-05,
+      "loss": 0.4006,
+      "step": 11562
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4929632319114207e-05,
+      "loss": 0.334,
+      "step": 11563
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4928765117238268e-05,
+      "loss": 0.3724,
+      "step": 11564
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4927897866399619e-05,
+      "loss": 0.4146,
+      "step": 11565
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4927030566606874e-05,
+      "loss": 0.4311,
+      "step": 11566
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4926163217868653e-05,
+      "loss": 0.3922,
+      "step": 11567
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4925295820193565e-05,
+      "loss": 0.371,
+      "step": 11568
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4924428373590235e-05,
+      "loss": 0.3628,
+      "step": 11569
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4923560878067275e-05,
+      "loss": 0.4017,
+      "step": 11570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4922693333633303e-05,
+      "loss": 0.3984,
+      "step": 11571
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4921825740296938e-05,
+      "loss": 0.3748,
+      "step": 11572
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.49209580980668e-05,
+      "loss": 0.4365,
+      "step": 11573
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4920090406951507e-05,
+      "loss": 0.29,
+      "step": 11574
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.491922266695968e-05,
+      "loss": 0.348,
+      "step": 11575
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4918354878099937e-05,
+      "loss": 0.3538,
+      "step": 11576
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4917487040380898e-05,
+      "loss": 0.4536,
+      "step": 11577
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.491661915381119e-05,
+      "loss": 0.3589,
+      "step": 11578
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4915751218399427e-05,
+      "loss": 0.3947,
+      "step": 11579
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4914883234154239e-05,
+      "loss": 0.3549,
+      "step": 11580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.491401520108424e-05,
+      "loss": 0.4151,
+      "step": 11581
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.491314711919806e-05,
+      "loss": 0.3717,
+      "step": 11582
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4912278988504318e-05,
+      "loss": 0.401,
+      "step": 11583
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4911410809011642e-05,
+      "loss": 0.3525,
+      "step": 11584
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4910542580728656e-05,
+      "loss": 0.3712,
+      "step": 11585
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4909674303663981e-05,
+      "loss": 0.4327,
+      "step": 11586
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4908805977826246e-05,
+      "loss": 0.3564,
+      "step": 11587
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490793760322408e-05,
+      "loss": 0.471,
+      "step": 11588
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4907069179866098e-05,
+      "loss": 0.3963,
+      "step": 11589
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490620070776094e-05,
+      "loss": 0.4075,
+      "step": 11590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4905332186917227e-05,
+      "loss": 0.3945,
+      "step": 11591
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490446361734359e-05,
+      "loss": 0.446,
+      "step": 11592
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4903594999048654e-05,
+      "loss": 0.434,
+      "step": 11593
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4902726332041049e-05,
+      "loss": 0.37,
+      "step": 11594
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4901857616329405e-05,
+      "loss": 0.4009,
+      "step": 11595
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490098885192235e-05,
+      "loss": 0.4035,
+      "step": 11596
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490012003882852e-05,
+      "loss": 0.4273,
+      "step": 11597
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4899251177056541e-05,
+      "loss": 0.3956,
+      "step": 11598
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4898382266615045e-05,
+      "loss": 0.3168,
+      "step": 11599
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4897513307512665e-05,
+      "loss": 0.4355,
+      "step": 11600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4896644299758029e-05,
+      "loss": 0.383,
+      "step": 11601
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4895775243359778e-05,
+      "loss": 0.4045,
+      "step": 11602
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4894906138326537e-05,
+      "loss": 0.3952,
+      "step": 11603
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4894036984666946e-05,
+      "loss": 0.3441,
+      "step": 11604
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4893167782389635e-05,
+      "loss": 0.4338,
+      "step": 11605
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4892298531503241e-05,
+      "loss": 0.3873,
+      "step": 11606
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4891429232016399e-05,
+      "loss": 0.3827,
+      "step": 11607
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4890559883937741e-05,
+      "loss": 0.3195,
+      "step": 11608
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488969048727591e-05,
+      "loss": 0.3538,
+      "step": 11609
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4888821042039536e-05,
+      "loss": 0.4275,
+      "step": 11610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488795154823726e-05,
+      "loss": 0.3571,
+      "step": 11611
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488708200587772e-05,
+      "loss": 0.3913,
+      "step": 11612
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4886212414969551e-05,
+      "loss": 0.431,
+      "step": 11613
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4885342775521394e-05,
+      "loss": 0.4071,
+      "step": 11614
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488447308754189e-05,
+      "loss": 0.4328,
+      "step": 11615
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4883603351039673e-05,
+      "loss": 0.3788,
+      "step": 11616
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4882733566023385e-05,
+      "loss": 0.4678,
+      "step": 11617
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488186373250167e-05,
+      "loss": 0.3571,
+      "step": 11618
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4880993850483166e-05,
+      "loss": 0.4605,
+      "step": 11619
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4880123919976512e-05,
+      "loss": 0.454,
+      "step": 11620
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487925394099036e-05,
+      "loss": 0.3112,
+      "step": 11621
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4878383913533338e-05,
+      "loss": 0.4022,
+      "step": 11622
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.48775138376141e-05,
+      "loss": 0.4393,
+      "step": 11623
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4876643713241286e-05,
+      "loss": 0.3213,
+      "step": 11624
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4875773540423538e-05,
+      "loss": 0.4426,
+      "step": 11625
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4874903319169501e-05,
+      "loss": 0.4468,
+      "step": 11626
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487403304948782e-05,
+      "loss": 0.3668,
+      "step": 11627
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4873162731387147e-05,
+      "loss": 0.4687,
+      "step": 11628
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4872292364876115e-05,
+      "loss": 0.4631,
+      "step": 11629
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487142194996338e-05,
+      "loss": 0.4613,
+      "step": 11630
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4870551486657588e-05,
+      "loss": 0.3977,
+      "step": 11631
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.486968097496738e-05,
+      "loss": 0.3095,
+      "step": 11632
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4868810414901411e-05,
+      "loss": 0.4131,
+      "step": 11633
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4867939806468328e-05,
+      "loss": 0.3959,
+      "step": 11634
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4867069149676774e-05,
+      "loss": 0.4254,
+      "step": 11635
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4866198444535403e-05,
+      "loss": 0.4281,
+      "step": 11636
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4865327691052862e-05,
+      "loss": 0.4279,
+      "step": 11637
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4864456889237807e-05,
+      "loss": 0.3897,
+      "step": 11638
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4863586039098881e-05,
+      "loss": 0.3834,
+      "step": 11639
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4862715140644739e-05,
+      "loss": 0.4402,
+      "step": 11640
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.486184419388403e-05,
+      "loss": 0.3848,
+      "step": 11641
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.486097319882541e-05,
+      "loss": 0.3774,
+      "step": 11642
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4860102155477533e-05,
+      "loss": 0.3958,
+      "step": 11643
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4859231063849045e-05,
+      "loss": 0.3772,
+      "step": 11644
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4858359923948604e-05,
+      "loss": 0.4296,
+      "step": 11645
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4857488735784862e-05,
+      "loss": 0.33,
+      "step": 11646
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4856617499366474e-05,
+      "loss": 0.3506,
+      "step": 11647
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4855746214702096e-05,
+      "loss": 0.4469,
+      "step": 11648
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4854874881800383e-05,
+      "loss": 0.4433,
+      "step": 11649
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4854003500669993e-05,
+      "loss": 0.4304,
+      "step": 11650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4853132071319575e-05,
+      "loss": 0.489,
+      "step": 11651
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4852260593757796e-05,
+      "loss": 0.4302,
+      "step": 11652
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4851389067993306e-05,
+      "loss": 0.4652,
+      "step": 11653
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4850517494034764e-05,
+      "loss": 0.362,
+      "step": 11654
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4849645871890832e-05,
+      "loss": 0.3939,
+      "step": 11655
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4848774201570165e-05,
+      "loss": 0.3425,
+      "step": 11656
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4847902483081422e-05,
+      "loss": 0.4073,
+      "step": 11657
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4847030716433264e-05,
+      "loss": 0.3158,
+      "step": 11658
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4846158901634353e-05,
+      "loss": 0.3709,
+      "step": 11659
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4845287038693345e-05,
+      "loss": 0.4175,
+      "step": 11660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4844415127618908e-05,
+      "loss": 0.373,
+      "step": 11661
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4843543168419698e-05,
+      "loss": 0.3862,
+      "step": 11662
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4842671161104376e-05,
+      "loss": 0.3891,
+      "step": 11663
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4841799105681612e-05,
+      "loss": 0.3882,
+      "step": 11664
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4840927002160062e-05,
+      "loss": 0.3689,
+      "step": 11665
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4840054850548391e-05,
+      "loss": 0.375,
+      "step": 11666
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4839182650855266e-05,
+      "loss": 0.3606,
+      "step": 11667
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4838310403089348e-05,
+      "loss": 0.4226,
+      "step": 11668
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4837438107259305e-05,
+      "loss": 0.4512,
+      "step": 11669
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.48365657633738e-05,
+      "loss": 0.3978,
+      "step": 11670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4835693371441498e-05,
+      "loss": 0.377,
+      "step": 11671
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483482093147107e-05,
+      "loss": 0.4041,
+      "step": 11672
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4833948443471177e-05,
+      "loss": 0.3558,
+      "step": 11673
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4833075907450493e-05,
+      "loss": 0.3687,
+      "step": 11674
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4832203323417679e-05,
+      "loss": 0.359,
+      "step": 11675
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4831330691381409e-05,
+      "loss": 0.3726,
+      "step": 11676
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483045801135035e-05,
+      "loss": 0.3496,
+      "step": 11677
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482958528333317e-05,
+      "loss": 0.3396,
+      "step": 11678
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482871250733854e-05,
+      "loss": 0.374,
+      "step": 11679
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4827839683375127e-05,
+      "loss": 0.4425,
+      "step": 11680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4826966811451606e-05,
+      "loss": 0.3938,
+      "step": 11681
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4826093891576646e-05,
+      "loss": 0.3771,
+      "step": 11682
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482522092375892e-05,
+      "loss": 0.4532,
+      "step": 11683
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4824347908007097e-05,
+      "loss": 0.4034,
+      "step": 11684
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4823474844329852e-05,
+      "loss": 0.3905,
+      "step": 11685
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4822601732735863e-05,
+      "loss": 0.3936,
+      "step": 11686
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4821728573233796e-05,
+      "loss": 0.369,
+      "step": 11687
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4820855365832323e-05,
+      "loss": 0.3606,
+      "step": 11688
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481998211054013e-05,
+      "loss": 0.3768,
+      "step": 11689
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4819108807365881e-05,
+      "loss": 0.399,
+      "step": 11690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4818235456318255e-05,
+      "loss": 0.3812,
+      "step": 11691
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481736205740593e-05,
+      "loss": 0.4279,
+      "step": 11692
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481648861063758e-05,
+      "loss": 0.3793,
+      "step": 11693
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4815615116021883e-05,
+      "loss": 0.4345,
+      "step": 11694
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4814741573567514e-05,
+      "loss": 0.3431,
+      "step": 11695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4813867983283159e-05,
+      "loss": 0.4011,
+      "step": 11696
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4812994345177485e-05,
+      "loss": 0.4018,
+      "step": 11697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4812120659259177e-05,
+      "loss": 0.4532,
+      "step": 11698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4811246925536913e-05,
+      "loss": 0.4368,
+      "step": 11699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4810373144019371e-05,
+      "loss": 0.3696,
+      "step": 11700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4809499314715239e-05,
+      "loss": 0.3393,
+      "step": 11701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4808625437633188e-05,
+      "loss": 0.3948,
+      "step": 11702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4807751512781903e-05,
+      "loss": 0.3935,
+      "step": 11703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4806877540170065e-05,
+      "loss": 0.4476,
+      "step": 11704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4806003519806363e-05,
+      "loss": 0.2961,
+      "step": 11705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4805129451699465e-05,
+      "loss": 0.4038,
+      "step": 11706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4804255335858064e-05,
+      "loss": 0.5114,
+      "step": 11707
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4803381172290848e-05,
+      "loss": 0.3981,
+      "step": 11708
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4802506961006486e-05,
+      "loss": 0.4526,
+      "step": 11709
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.480163270201368e-05,
+      "loss": 0.4111,
+      "step": 11710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4800758395321102e-05,
+      "loss": 0.4364,
+      "step": 11711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.479988404093744e-05,
+      "loss": 0.4303,
+      "step": 11712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4799009638871385e-05,
+      "loss": 0.3978,
+      "step": 11713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4798135189131619e-05,
+      "loss": 0.4268,
+      "step": 11714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4797260691726831e-05,
+      "loss": 0.3126,
+      "step": 11715
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4796386146665704e-05,
+      "loss": 0.3868,
+      "step": 11716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4795511553956931e-05,
+      "loss": 0.4154,
+      "step": 11717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4794636913609199e-05,
+      "loss": 0.4324,
+      "step": 11718
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4793762225631196e-05,
+      "loss": 0.348,
+      "step": 11719
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4792887490031613e-05,
+      "loss": 0.3614,
+      "step": 11720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4792012706819133e-05,
+      "loss": 0.37,
+      "step": 11721
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4791137876002454e-05,
+      "loss": 0.4769,
+      "step": 11722
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4790262997590266e-05,
+      "loss": 0.3811,
+      "step": 11723
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4789388071591255e-05,
+      "loss": 0.3168,
+      "step": 11724
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4788513098014114e-05,
+      "loss": 0.3916,
+      "step": 11725
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4787638076867541e-05,
+      "loss": 0.3823,
+      "step": 11726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4786763008160221e-05,
+      "loss": 0.4293,
+      "step": 11727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.478588789190085e-05,
+      "loss": 0.4625,
+      "step": 11728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4785012728098123e-05,
+      "loss": 0.4262,
+      "step": 11729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4784137516760733e-05,
+      "loss": 0.4104,
+      "step": 11730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4783262257897369e-05,
+      "loss": 0.3868,
+      "step": 11731
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4782386951516734e-05,
+      "loss": 0.3405,
+      "step": 11732
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.478151159762752e-05,
+      "loss": 0.3543,
+      "step": 11733
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4780636196238422e-05,
+      "loss": 0.4314,
+      "step": 11734
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4779760747358141e-05,
+      "loss": 0.3692,
+      "step": 11735
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4778885250995365e-05,
+      "loss": 0.4674,
+      "step": 11736
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4778009707158798e-05,
+      "loss": 0.2901,
+      "step": 11737
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4777134115857137e-05,
+      "loss": 0.3293,
+      "step": 11738
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.477625847709908e-05,
+      "loss": 0.4698,
+      "step": 11739
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.477538279089332e-05,
+      "loss": 0.3644,
+      "step": 11740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4774507057248565e-05,
+      "loss": 0.3913,
+      "step": 11741
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.477363127617351e-05,
+      "loss": 0.4139,
+      "step": 11742
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4772755447676852e-05,
+      "loss": 0.3505,
+      "step": 11743
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.47718795717673e-05,
+      "loss": 0.3865,
+      "step": 11744
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4771003648453548e-05,
+      "loss": 0.3165,
+      "step": 11745
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4770127677744297e-05,
+      "loss": 0.3991,
+      "step": 11746
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4769251659648257e-05,
+      "loss": 0.4665,
+      "step": 11747
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4768375594174123e-05,
+      "loss": 0.3501,
+      "step": 11748
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4767499481330598e-05,
+      "loss": 0.398,
+      "step": 11749
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476662332112639e-05,
+      "loss": 0.3576,
+      "step": 11750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.47657471135702e-05,
+      "loss": 0.378,
+      "step": 11751
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4764870858670727e-05,
+      "loss": 0.3725,
+      "step": 11752
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4763994556436687e-05,
+      "loss": 0.4002,
+      "step": 11753
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476311820687678e-05,
+      "loss": 0.3477,
+      "step": 11754
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4762241809999704e-05,
+      "loss": 0.3638,
+      "step": 11755
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476136536581418e-05,
+      "loss": 0.4169,
+      "step": 11756
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4760488874328905e-05,
+      "loss": 0.4229,
+      "step": 11757
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4759612335552586e-05,
+      "loss": 0.4332,
+      "step": 11758
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4758735749493936e-05,
+      "loss": 0.3288,
+      "step": 11759
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4757859116161657e-05,
+      "loss": 0.3522,
+      "step": 11760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4756982435564462e-05,
+      "loss": 0.4236,
+      "step": 11761
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4756105707711057e-05,
+      "loss": 0.3904,
+      "step": 11762
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4755228932610155e-05,
+      "loss": 0.4448,
+      "step": 11763
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475435211027046e-05,
+      "loss": 0.4041,
+      "step": 11764
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475347524070069e-05,
+      "loss": 0.3183,
+      "step": 11765
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475259832390955e-05,
+      "loss": 0.4092,
+      "step": 11766
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4751721359905755e-05,
+      "loss": 0.3329,
+      "step": 11767
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4750844348698014e-05,
+      "loss": 0.4074,
+      "step": 11768
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.474996729029504e-05,
+      "loss": 0.3146,
+      "step": 11769
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4749090184705548e-05,
+      "loss": 0.4112,
+      "step": 11770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4748213031938247e-05,
+      "loss": 0.4117,
+      "step": 11771
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4747335832001857e-05,
+      "loss": 0.4006,
+      "step": 11772
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4746458584905085e-05,
+      "loss": 0.4277,
+      "step": 11773
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4745581290656652e-05,
+      "loss": 0.4341,
+      "step": 11774
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4744703949265268e-05,
+      "loss": 0.3404,
+      "step": 11775
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4743826560739654e-05,
+      "loss": 0.3822,
+      "step": 11776
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4742949125088518e-05,
+      "loss": 0.3679,
+      "step": 11777
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4742071642320586e-05,
+      "loss": 0.358,
+      "step": 11778
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4741194112444566e-05,
+      "loss": 0.4826,
+      "step": 11779
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4740316535469181e-05,
+      "loss": 0.4162,
+      "step": 11780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4739438911403147e-05,
+      "loss": 0.3716,
+      "step": 11781
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4738561240255185e-05,
+      "loss": 0.3495,
+      "step": 11782
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4737683522034011e-05,
+      "loss": 0.3362,
+      "step": 11783
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4736805756748345e-05,
+      "loss": 0.4017,
+      "step": 11784
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4735927944406909e-05,
+      "loss": 0.4689,
+      "step": 11785
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4735050085018419e-05,
+      "loss": 0.4192,
+      "step": 11786
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4734172178591597e-05,
+      "loss": 0.3867,
+      "step": 11787
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.473329422513517e-05,
+      "loss": 0.3434,
+      "step": 11788
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4732416224657846e-05,
+      "loss": 0.3162,
+      "step": 11789
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4731538177168362e-05,
+      "loss": 0.3648,
+      "step": 11790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4730660082675433e-05,
+      "loss": 0.391,
+      "step": 11791
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4729781941187787e-05,
+      "loss": 0.3912,
+      "step": 11792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4728903752714142e-05,
+      "loss": 0.4197,
+      "step": 11793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4728025517263221e-05,
+      "loss": 0.4298,
+      "step": 11794
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4727147234843752e-05,
+      "loss": 0.4134,
+      "step": 11795
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4726268905464462e-05,
+      "loss": 0.3395,
+      "step": 11796
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4725390529134073e-05,
+      "loss": 0.3082,
+      "step": 11797
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4724512105861311e-05,
+      "loss": 0.3752,
+      "step": 11798
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4723633635654904e-05,
+      "loss": 0.4153,
+      "step": 11799
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4722755118523578e-05,
+      "loss": 0.3879,
+      "step": 11800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4721876554476058e-05,
+      "loss": 0.4196,
+      "step": 11801
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4720997943521076e-05,
+      "loss": 0.4388,
+      "step": 11802
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4720119285667357e-05,
+      "loss": 0.4093,
+      "step": 11803
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.471924058092363e-05,
+      "loss": 0.3245,
+      "step": 11804
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4718361829298626e-05,
+      "loss": 0.4547,
+      "step": 11805
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4717483030801072e-05,
+      "loss": 0.4333,
+      "step": 11806
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4716604185439698e-05,
+      "loss": 0.4598,
+      "step": 11807
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.471572529322324e-05,
+      "loss": 0.405,
+      "step": 11808
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4714846354160424e-05,
+      "loss": 0.3346,
+      "step": 11809
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 0.3123,
+      "step": 11810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.471308833553064e-05,
+      "loss": 0.2729,
+      "step": 11811
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4712209255981142e-05,
+      "loss": 0.395,
+      "step": 11812
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4711330129620217e-05,
+      "loss": 0.4343,
+      "step": 11813
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4710450956456593e-05,
+      "loss": 0.3739,
+      "step": 11814
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4709571736499008e-05,
+      "loss": 0.2582,
+      "step": 11815
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4708692469756195e-05,
+      "loss": 0.3968,
+      "step": 11816
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4707813156236891e-05,
+      "loss": 0.4034,
+      "step": 11817
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4706933795949832e-05,
+      "loss": 0.4005,
+      "step": 11818
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.470605438890375e-05,
+      "loss": 0.4395,
+      "step": 11819
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4705174935107379e-05,
+      "loss": 0.3858,
+      "step": 11820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4704295434569463e-05,
+      "loss": 0.4836,
+      "step": 11821
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.470341588729873e-05,
+      "loss": 0.3514,
+      "step": 11822
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4702536293303924e-05,
+      "loss": 0.3778,
+      "step": 11823
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4701656652593785e-05,
+      "loss": 0.307,
+      "step": 11824
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4700776965177044e-05,
+      "loss": 0.4184,
+      "step": 11825
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4699897231062443e-05,
+      "loss": 0.3609,
+      "step": 11826
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4699017450258724e-05,
+      "loss": 0.3986,
+      "step": 11827
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4698137622774624e-05,
+      "loss": 0.4044,
+      "step": 11828
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4697257748618886e-05,
+      "loss": 0.3917,
+      "step": 11829
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4696377827800245e-05,
+      "loss": 0.531,
+      "step": 11830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.469549786032745e-05,
+      "loss": 0.3578,
+      "step": 11831
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4694617846209235e-05,
+      "loss": 0.4153,
+      "step": 11832
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4693737785454349e-05,
+      "loss": 0.4162,
+      "step": 11833
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4692857678071528e-05,
+      "loss": 0.3226,
+      "step": 11834
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4691977524069518e-05,
+      "loss": 0.3528,
+      "step": 11835
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4691097323457069e-05,
+      "loss": 0.5462,
+      "step": 11836
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4690217076242915e-05,
+      "loss": 0.3362,
+      "step": 11837
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4689336782435801e-05,
+      "loss": 0.3716,
+      "step": 11838
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4688456442044479e-05,
+      "loss": 0.3885,
+      "step": 11839
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.468757605507769e-05,
+      "loss": 0.3525,
+      "step": 11840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.468669562154418e-05,
+      "loss": 0.4544,
+      "step": 11841
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4685815141452695e-05,
+      "loss": 0.392,
+      "step": 11842
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4684934614811984e-05,
+      "loss": 0.4192,
+      "step": 11843
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4684054041630792e-05,
+      "loss": 0.4092,
+      "step": 11844
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4683173421917865e-05,
+      "loss": 0.2976,
+      "step": 11845
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4682292755681956e-05,
+      "loss": 0.4058,
+      "step": 11846
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4681412042931813e-05,
+      "loss": 0.3894,
+      "step": 11847
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4680531283676178e-05,
+      "loss": 0.3985,
+      "step": 11848
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467965047792381e-05,
+      "loss": 0.4365,
+      "step": 11849
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4678769625683454e-05,
+      "loss": 0.3883,
+      "step": 11850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4677888726963858e-05,
+      "loss": 0.3934,
+      "step": 11851
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467700778177378e-05,
+      "loss": 0.4005,
+      "step": 11852
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4676126790121963e-05,
+      "loss": 0.3375,
+      "step": 11853
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4675245752017166e-05,
+      "loss": 0.427,
+      "step": 11854
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467436466746814e-05,
+      "loss": 0.4229,
+      "step": 11855
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4673483536483634e-05,
+      "loss": 0.4018,
+      "step": 11856
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4672602359072402e-05,
+      "loss": 0.4036,
+      "step": 11857
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4671721135243204e-05,
+      "loss": 0.4101,
+      "step": 11858
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4670839865004786e-05,
+      "loss": 0.3555,
+      "step": 11859
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4669958548365907e-05,
+      "loss": 0.3933,
+      "step": 11860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4669077185335319e-05,
+      "loss": 0.3897,
+      "step": 11861
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4668195775921781e-05,
+      "loss": 0.4005,
+      "step": 11862
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4667314320134047e-05,
+      "loss": 0.3827,
+      "step": 11863
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4666432817980876e-05,
+      "loss": 0.302,
+      "step": 11864
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4665551269471021e-05,
+      "loss": 0.4033,
+      "step": 11865
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.466466967461324e-05,
+      "loss": 0.417,
+      "step": 11866
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4663788033416296e-05,
+      "loss": 0.4334,
+      "step": 11867
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.466290634588894e-05,
+      "loss": 0.3353,
+      "step": 11868
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4662024612039931e-05,
+      "loss": 0.3499,
+      "step": 11869
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4661142831878036e-05,
+      "loss": 0.4391,
+      "step": 11870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4660261005412011e-05,
+      "loss": 0.3336,
+      "step": 11871
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4659379132650613e-05,
+      "loss": 0.3691,
+      "step": 11872
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4658497213602604e-05,
+      "loss": 0.4496,
+      "step": 11873
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4657615248276749e-05,
+      "loss": 0.3891,
+      "step": 11874
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4656733236681802e-05,
+      "loss": 0.4074,
+      "step": 11875
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465585117882653e-05,
+      "loss": 0.4205,
+      "step": 11876
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4654969074719696e-05,
+      "loss": 0.3206,
+      "step": 11877
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465408692437006e-05,
+      "loss": 0.4157,
+      "step": 11878
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4653204727786388e-05,
+      "loss": 0.3334,
+      "step": 11879
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4652322484977442e-05,
+      "loss": 0.3906,
+      "step": 11880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4651440195951985e-05,
+      "loss": 0.3085,
+      "step": 11881
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4650557860718788e-05,
+      "loss": 0.3959,
+      "step": 11882
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464967547928661e-05,
+      "loss": 0.3696,
+      "step": 11883
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464879305166422e-05,
+      "loss": 0.348,
+      "step": 11884
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4647910577860378e-05,
+      "loss": 0.3703,
+      "step": 11885
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4647028057883859e-05,
+      "loss": 0.3554,
+      "step": 11886
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4646145491743424e-05,
+      "loss": 0.3821,
+      "step": 11887
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4645262879447844e-05,
+      "loss": 0.384,
+      "step": 11888
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4644380221005887e-05,
+      "loss": 0.3409,
+      "step": 11889
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4643497516426315e-05,
+      "loss": 0.3615,
+      "step": 11890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4642614765717906e-05,
+      "loss": 0.2972,
+      "step": 11891
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4641731968889425e-05,
+      "loss": 0.3418,
+      "step": 11892
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4640849125949642e-05,
+      "loss": 0.3642,
+      "step": 11893
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4639966236907323e-05,
+      "loss": 0.3519,
+      "step": 11894
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4639083301771246e-05,
+      "loss": 0.4109,
+      "step": 11895
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.463820032055018e-05,
+      "loss": 0.3315,
+      "step": 11896
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4637317293252892e-05,
+      "loss": 0.3649,
+      "step": 11897
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.463643421988816e-05,
+      "loss": 0.4146,
+      "step": 11898
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4635551100464755e-05,
+      "loss": 0.3686,
+      "step": 11899
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4634667934991446e-05,
+      "loss": 0.4193,
+      "step": 11900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4633784723477014e-05,
+      "loss": 0.4089,
+      "step": 11901
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4632901465930225e-05,
+      "loss": 0.4392,
+      "step": 11902
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4632018162359857e-05,
+      "loss": 0.3651,
+      "step": 11903
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4631134812774685e-05,
+      "loss": 0.3768,
+      "step": 11904
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4630251417183483e-05,
+      "loss": 0.4732,
+      "step": 11905
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4629367975595028e-05,
+      "loss": 0.4102,
+      "step": 11906
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4628484488018097e-05,
+      "loss": 0.3403,
+      "step": 11907
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4627600954461465e-05,
+      "loss": 0.4752,
+      "step": 11908
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4626717374933908e-05,
+      "loss": 0.3958,
+      "step": 11909
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4625833749444205e-05,
+      "loss": 0.4163,
+      "step": 11910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4624950078001134e-05,
+      "loss": 0.3228,
+      "step": 11911
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4624066360613474e-05,
+      "loss": 0.376,
+      "step": 11912
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4623182597290004e-05,
+      "loss": 0.3977,
+      "step": 11913
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4622298788039501e-05,
+      "loss": 0.3369,
+      "step": 11914
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4621414932870746e-05,
+      "loss": 0.3628,
+      "step": 11915
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4620531031792523e-05,
+      "loss": 0.2929,
+      "step": 11916
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4619647084813606e-05,
+      "loss": 0.3505,
+      "step": 11917
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.461876309194278e-05,
+      "loss": 0.4255,
+      "step": 11918
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4617879053188828e-05,
+      "loss": 0.3809,
+      "step": 11919
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4616994968560531e-05,
+      "loss": 0.4011,
+      "step": 11920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4616110838066671e-05,
+      "loss": 0.3415,
+      "step": 11921
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.461522666171603e-05,
+      "loss": 0.4074,
+      "step": 11922
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4614342439517393e-05,
+      "loss": 0.3367,
+      "step": 11923
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.461345817147954e-05,
+      "loss": 0.4007,
+      "step": 11924
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4612573857611265e-05,
+      "loss": 0.3841,
+      "step": 11925
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4611689497921341e-05,
+      "loss": 0.3647,
+      "step": 11926
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4610805092418563e-05,
+      "loss": 0.389,
+      "step": 11927
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4609920641111713e-05,
+      "loss": 0.3825,
+      "step": 11928
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4609036144009574e-05,
+      "loss": 0.4304,
+      "step": 11929
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4608151601120935e-05,
+      "loss": 0.3891,
+      "step": 11930
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4607267012454584e-05,
+      "loss": 0.3674,
+      "step": 11931
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4606382378019312e-05,
+      "loss": 0.4089,
+      "step": 11932
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4605497697823899e-05,
+      "loss": 0.441,
+      "step": 11933
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4604612971877139e-05,
+      "loss": 0.3883,
+      "step": 11934
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4603728200187823e-05,
+      "loss": 0.4112,
+      "step": 11935
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4602843382764734e-05,
+      "loss": 0.366,
+      "step": 11936
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4601958519616663e-05,
+      "loss": 0.2945,
+      "step": 11937
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4601073610752405e-05,
+      "loss": 0.394,
+      "step": 11938
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4600188656180747e-05,
+      "loss": 0.3352,
+      "step": 11939
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4599303655910482e-05,
+      "loss": 0.3871,
+      "step": 11940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.45984186099504e-05,
+      "loss": 0.4057,
+      "step": 11941
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4597533518309295e-05,
+      "loss": 0.4343,
+      "step": 11942
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4596648380995956e-05,
+      "loss": 0.4059,
+      "step": 11943
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4595763198019182e-05,
+      "loss": 0.4209,
+      "step": 11944
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4594877969387761e-05,
+      "loss": 0.3627,
+      "step": 11945
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4593992695110486e-05,
+      "loss": 0.3999,
+      "step": 11946
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4593107375196161e-05,
+      "loss": 0.4211,
+      "step": 11947
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.459222200965357e-05,
+      "loss": 0.412,
+      "step": 11948
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4591336598491512e-05,
+      "loss": 0.4071,
+      "step": 11949
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4590451141718785e-05,
+      "loss": 0.3817,
+      "step": 11950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4589565639344181e-05,
+      "loss": 0.3573,
+      "step": 11951
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.45886800913765e-05,
+      "loss": 0.4311,
+      "step": 11952
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458779449782454e-05,
+      "loss": 0.5272,
+      "step": 11953
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4586908858697095e-05,
+      "loss": 0.4076,
+      "step": 11954
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4586023174002961e-05,
+      "loss": 0.3666,
+      "step": 11955
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4585137443750945e-05,
+      "loss": 0.4179,
+      "step": 11956
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458425166794984e-05,
+      "loss": 0.4033,
+      "step": 11957
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4583365846608444e-05,
+      "loss": 0.4152,
+      "step": 11958
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4582479979735562e-05,
+      "loss": 0.396,
+      "step": 11959
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4581594067339993e-05,
+      "loss": 0.3601,
+      "step": 11960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458070810943053e-05,
+      "loss": 0.3269,
+      "step": 11961
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4579822106015984e-05,
+      "loss": 0.3825,
+      "step": 11962
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4578936057105152e-05,
+      "loss": 0.4201,
+      "step": 11963
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457804996270684e-05,
+      "loss": 0.4271,
+      "step": 11964
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4577163822829845e-05,
+      "loss": 0.3172,
+      "step": 11965
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4576277637482975e-05,
+      "loss": 0.2992,
+      "step": 11966
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4575391406675026e-05,
+      "loss": 0.3251,
+      "step": 11967
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457450513041481e-05,
+      "loss": 0.3516,
+      "step": 11968
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457361880871113e-05,
+      "loss": 0.3376,
+      "step": 11969
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4572732441572783e-05,
+      "loss": 0.4538,
+      "step": 11970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4571846029008585e-05,
+      "loss": 0.3769,
+      "step": 11971
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4570959571027337e-05,
+      "loss": 0.3391,
+      "step": 11972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4570073067637842e-05,
+      "loss": 0.4363,
+      "step": 11973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4569186518848913e-05,
+      "loss": 0.3567,
+      "step": 11974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4568299924669351e-05,
+      "loss": 0.4454,
+      "step": 11975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4567413285107967e-05,
+      "loss": 0.3136,
+      "step": 11976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.456652660017357e-05,
+      "loss": 0.3935,
+      "step": 11977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4565639869874964e-05,
+      "loss": 0.339,
+      "step": 11978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4564753094220963e-05,
+      "loss": 0.3068,
+      "step": 11979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4563866273220369e-05,
+      "loss": 0.3488,
+      "step": 11980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4562979406881999e-05,
+      "loss": 0.3057,
+      "step": 11981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4562092495214662e-05,
+      "loss": 0.4618,
+      "step": 11982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4561205538227163e-05,
+      "loss": 0.4581,
+      "step": 11983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.456031853592832e-05,
+      "loss": 0.397,
+      "step": 11984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4559431488326943e-05,
+      "loss": 0.4068,
+      "step": 11985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.455854439543184e-05,
+      "loss": 0.4029,
+      "step": 11986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4557657257251828e-05,
+      "loss": 0.3094,
+      "step": 11987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4556770073795718e-05,
+      "loss": 0.4896,
+      "step": 11988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4555882845072322e-05,
+      "loss": 0.3781,
+      "step": 11989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4554995571090459e-05,
+      "loss": 0.4228,
+      "step": 11990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4554108251858937e-05,
+      "loss": 0.363,
+      "step": 11991
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.455322088738657e-05,
+      "loss": 0.3436,
+      "step": 11992
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.455233347768218e-05,
+      "loss": 0.4056,
+      "step": 11993
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4551446022754581e-05,
+      "loss": 0.3382,
+      "step": 11994
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4550558522612584e-05,
+      "loss": 0.3829,
+      "step": 11995
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4549670977265009e-05,
+      "loss": 0.3907,
+      "step": 11996
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4548783386720673e-05,
+      "loss": 0.3829,
+      "step": 11997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547895750988393e-05,
+      "loss": 0.3442,
+      "step": 11998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547008070076984e-05,
+      "loss": 0.2786,
+      "step": 11999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4546120343995271e-05,
+      "loss": 0.4157,
+      "step": 12000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4545232572752065e-05,
+      "loss": 0.3772,
+      "step": 12001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4544344756356193e-05,
+      "loss": 0.3702,
+      "step": 12002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4543456894816466e-05,
+      "loss": 0.3533,
+      "step": 12003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.454256898814171e-05,
+      "loss": 0.3728,
+      "step": 12004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4541681036340744e-05,
+      "loss": 0.3561,
+      "step": 12005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4540793039422391e-05,
+      "loss": 0.4082,
+      "step": 12006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.3592,
+      "step": 12007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4539016910268801e-05,
+      "loss": 0.3851,
+      "step": 12008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4538128778051213e-05,
+      "loss": 0.334,
+      "step": 12009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453724060075152e-05,
+      "loss": 0.4171,
+      "step": 12010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4536352378378551e-05,
+      "loss": 0.4255,
+      "step": 12011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4535464110941131e-05,
+      "loss": 0.4093,
+      "step": 12012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4534575798448076e-05,
+      "loss": 0.3701,
+      "step": 12013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4533687440908222e-05,
+      "loss": 0.3427,
+      "step": 12014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4532799038330385e-05,
+      "loss": 0.3511,
+      "step": 12015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4531910590723394e-05,
+      "loss": 0.4137,
+      "step": 12016
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4531022098096075e-05,
+      "loss": 0.4269,
+      "step": 12017
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4530133560457253e-05,
+      "loss": 0.4095,
+      "step": 12018
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4529244977815756e-05,
+      "loss": 0.3205,
+      "step": 12019
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4528356350180411e-05,
+      "loss": 0.3592,
+      "step": 12020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4527467677560043e-05,
+      "loss": 0.3963,
+      "step": 12021
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4526578959963485e-05,
+      "loss": 0.3881,
+      "step": 12022
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4525690197399563e-05,
+      "loss": 0.3919,
+      "step": 12023
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4524801389877109e-05,
+      "loss": 0.469,
+      "step": 12024
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4523912537404945e-05,
+      "loss": 0.4065,
+      "step": 12025
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4523023639991904e-05,
+      "loss": 0.4444,
+      "step": 12026
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4522134697646824e-05,
+      "loss": 0.3508,
+      "step": 12027
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4521245710378527e-05,
+      "loss": 0.4056,
+      "step": 12028
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4520356678195849e-05,
+      "loss": 0.4162,
+      "step": 12029
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4519467601107617e-05,
+      "loss": 0.381,
+      "step": 12030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4518578479122667e-05,
+      "loss": 0.3634,
+      "step": 12031
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4517689312249829e-05,
+      "loss": 0.3571,
+      "step": 12032
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4516800100497941e-05,
+      "loss": 0.3731,
+      "step": 12033
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4515910843875834e-05,
+      "loss": 0.387,
+      "step": 12034
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4515021542392335e-05,
+      "loss": 0.5084,
+      "step": 12035
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.451413219605629e-05,
+      "loss": 0.4604,
+      "step": 12036
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4513242804876526e-05,
+      "loss": 0.3455,
+      "step": 12037
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4512353368861878e-05,
+      "loss": 0.4307,
+      "step": 12038
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4511463888021189e-05,
+      "loss": 0.464,
+      "step": 12039
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4510574362363287e-05,
+      "loss": 0.2949,
+      "step": 12040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509684791897014e-05,
+      "loss": 0.4087,
+      "step": 12041
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4508795176631205e-05,
+      "loss": 0.3909,
+      "step": 12042
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4507905516574699e-05,
+      "loss": 0.328,
+      "step": 12043
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4507015811736328e-05,
+      "loss": 0.3427,
+      "step": 12044
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4506126062124936e-05,
+      "loss": 0.346,
+      "step": 12045
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4505236267749366e-05,
+      "loss": 0.3828,
+      "step": 12046
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4504346428618445e-05,
+      "loss": 0.3187,
+      "step": 12047
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4503456544741026e-05,
+      "loss": 0.3001,
+      "step": 12048
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.450256661612594e-05,
+      "loss": 0.388,
+      "step": 12049
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4501676642782028e-05,
+      "loss": 0.3712,
+      "step": 12050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4500786624718139e-05,
+      "loss": 0.3847,
+      "step": 12051
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4499896561943105e-05,
+      "loss": 0.3164,
+      "step": 12052
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4499006454465774e-05,
+      "loss": 0.4577,
+      "step": 12053
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4498116302294986e-05,
+      "loss": 0.3938,
+      "step": 12054
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4497226105439586e-05,
+      "loss": 0.406,
+      "step": 12055
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4496335863908413e-05,
+      "loss": 0.3792,
+      "step": 12056
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4495445577710314e-05,
+      "loss": 0.4231,
+      "step": 12057
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4494555246854134e-05,
+      "loss": 0.3655,
+      "step": 12058
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4493664871348717e-05,
+      "loss": 0.44,
+      "step": 12059
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4492774451202903e-05,
+      "loss": 0.3472,
+      "step": 12060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4491883986425544e-05,
+      "loss": 0.3603,
+      "step": 12061
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4490993477025483e-05,
+      "loss": 0.3081,
+      "step": 12062
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4490102923011566e-05,
+      "loss": 0.4112,
+      "step": 12063
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4489212324392642e-05,
+      "loss": 0.3166,
+      "step": 12064
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4488321681177558e-05,
+      "loss": 0.3128,
+      "step": 12065
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.448743099337516e-05,
+      "loss": 0.3612,
+      "step": 12066
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4486540260994297e-05,
+      "loss": 0.3844,
+      "step": 12067
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4485649484043818e-05,
+      "loss": 0.3701,
+      "step": 12068
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.448475866253257e-05,
+      "loss": 0.4176,
+      "step": 12069
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4483867796469407e-05,
+      "loss": 0.3441,
+      "step": 12070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4482976885863173e-05,
+      "loss": 0.379,
+      "step": 12071
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4482085930722722e-05,
+      "loss": 0.4288,
+      "step": 12072
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4481194931056906e-05,
+      "loss": 0.3939,
+      "step": 12073
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.448030388687457e-05,
+      "loss": 0.4409,
+      "step": 12074
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4479412798184576e-05,
+      "loss": 0.2884,
+      "step": 12075
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4478521664995766e-05,
+      "loss": 0.3302,
+      "step": 12076
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4477630487317e-05,
+      "loss": 0.3374,
+      "step": 12077
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4476739265157125e-05,
+      "loss": 0.3146,
+      "step": 12078
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4475847998525e-05,
+      "loss": 0.3681,
+      "step": 12079
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4474956687429476e-05,
+      "loss": 0.4551,
+      "step": 12080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4474065331879403e-05,
+      "loss": 0.3821,
+      "step": 12081
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4473173931883646e-05,
+      "loss": 0.4124,
+      "step": 12082
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4472282487451052e-05,
+      "loss": 0.4365,
+      "step": 12083
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4471390998590478e-05,
+      "loss": 0.375,
+      "step": 12084
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4470499465310786e-05,
+      "loss": 0.3907,
+      "step": 12085
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4469607887620825e-05,
+      "loss": 0.3902,
+      "step": 12086
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4468716265529455e-05,
+      "loss": 0.398,
+      "step": 12087
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4467824599045532e-05,
+      "loss": 0.3393,
+      "step": 12088
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446693288817792e-05,
+      "loss": 0.3551,
+      "step": 12089
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4466041132935467e-05,
+      "loss": 0.2965,
+      "step": 12090
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446514933332704e-05,
+      "loss": 0.425,
+      "step": 12091
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4464257489361497e-05,
+      "loss": 0.4459,
+      "step": 12092
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4463365601047692e-05,
+      "loss": 0.3941,
+      "step": 12093
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4462473668394495e-05,
+      "loss": 0.3465,
+      "step": 12094
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4461581691410757e-05,
+      "loss": 0.3743,
+      "step": 12095
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4460689670105344e-05,
+      "loss": 0.3464,
+      "step": 12096
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4459797604487116e-05,
+      "loss": 0.4225,
+      "step": 12097
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4458905494564935e-05,
+      "loss": 0.4035,
+      "step": 12098
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.445801334034766e-05,
+      "loss": 0.3984,
+      "step": 12099
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4457121141844161e-05,
+      "loss": 0.3222,
+      "step": 12100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4456228899063298e-05,
+      "loss": 0.3047,
+      "step": 12101
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4455336612013931e-05,
+      "loss": 0.3409,
+      "step": 12102
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4454444280704927e-05,
+      "loss": 0.4097,
+      "step": 12103
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4453551905145156e-05,
+      "loss": 0.3427,
+      "step": 12104
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4452659485343473e-05,
+      "loss": 0.3268,
+      "step": 12105
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4451767021308744e-05,
+      "loss": 0.3557,
+      "step": 12106
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4450874513049841e-05,
+      "loss": 0.3733,
+      "step": 12107
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444998196057563e-05,
+      "loss": 0.3694,
+      "step": 12108
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4449089363894975e-05,
+      "loss": 0.3744,
+      "step": 12109
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444819672301674e-05,
+      "loss": 0.3522,
+      "step": 12110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4447304037949799e-05,
+      "loss": 0.382,
+      "step": 12111
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4446411308703017e-05,
+      "loss": 0.2842,
+      "step": 12112
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4445518535285263e-05,
+      "loss": 0.3657,
+      "step": 12113
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4444625717705405e-05,
+      "loss": 0.3898,
+      "step": 12114
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4443732855972307e-05,
+      "loss": 0.4523,
+      "step": 12115
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4442839950094851e-05,
+      "loss": 0.3886,
+      "step": 12116
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.44419470000819e-05,
+      "loss": 0.3827,
+      "step": 12117
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4441054005942325e-05,
+      "loss": 0.4131,
+      "step": 12118
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4440160967684994e-05,
+      "loss": 0.4466,
+      "step": 12119
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4439267885318786e-05,
+      "loss": 0.3454,
+      "step": 12120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4438374758852566e-05,
+      "loss": 0.4594,
+      "step": 12121
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.443748158829521e-05,
+      "loss": 0.3393,
+      "step": 12122
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4436588373655593e-05,
+      "loss": 0.378,
+      "step": 12123
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4435695114942579e-05,
+      "loss": 0.3413,
+      "step": 12124
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4434801812165052e-05,
+      "loss": 0.3962,
+      "step": 12125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4433908465331882e-05,
+      "loss": 0.3796,
+      "step": 12126
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4433015074451941e-05,
+      "loss": 0.3807,
+      "step": 12127
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4432121639534112e-05,
+      "loss": 0.4494,
+      "step": 12128
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.443122816058726e-05,
+      "loss": 0.4201,
+      "step": 12129
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4430334637620266e-05,
+      "loss": 0.4106,
+      "step": 12130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4429441070642009e-05,
+      "loss": 0.3604,
+      "step": 12131
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4428547459661361e-05,
+      "loss": 0.3903,
+      "step": 12132
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4427653804687203e-05,
+      "loss": 0.388,
+      "step": 12133
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442676010572841e-05,
+      "loss": 0.3465,
+      "step": 12134
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4425866362793863e-05,
+      "loss": 0.4195,
+      "step": 12135
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4424972575892432e-05,
+      "loss": 0.438,
+      "step": 12136
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4424078745033007e-05,
+      "loss": 0.342,
+      "step": 12137
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4423184870224468e-05,
+      "loss": 0.3953,
+      "step": 12138
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4422290951475681e-05,
+      "loss": 0.3823,
+      "step": 12139
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442139698879554e-05,
+      "loss": 0.3797,
+      "step": 12140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4420502982192916e-05,
+      "loss": 0.3459,
+      "step": 12141
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4419608931676699e-05,
+      "loss": 0.3541,
+      "step": 12142
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4418714837255766e-05,
+      "loss": 0.3802,
+      "step": 12143
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4417820698938995e-05,
+      "loss": 0.3761,
+      "step": 12144
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4416926516735274e-05,
+      "loss": 0.3892,
+      "step": 12145
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4416032290653487e-05,
+      "loss": 0.3979,
+      "step": 12146
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4415138020702518e-05,
+      "loss": 0.463,
+      "step": 12147
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.441424370689124e-05,
+      "loss": 0.3441,
+      "step": 12148
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4413349349228547e-05,
+      "loss": 0.3839,
+      "step": 12149
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4412454947723324e-05,
+      "loss": 0.3073,
+      "step": 12150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4411560502384452e-05,
+      "loss": 0.3475,
+      "step": 12151
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4410666013220818e-05,
+      "loss": 0.4542,
+      "step": 12152
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4409771480241307e-05,
+      "loss": 0.4069,
+      "step": 12153
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4408876903454808e-05,
+      "loss": 0.4443,
+      "step": 12154
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4407982282870203e-05,
+      "loss": 0.3454,
+      "step": 12155
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4407087618496384e-05,
+      "loss": 0.4341,
+      "step": 12156
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4406192910342239e-05,
+      "loss": 0.4036,
+      "step": 12157
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4405298158416649e-05,
+      "loss": 0.3731,
+      "step": 12158
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4404403362728513e-05,
+      "loss": 0.4274,
+      "step": 12159
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4403508523286711e-05,
+      "loss": 0.3599,
+      "step": 12160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4402613640100136e-05,
+      "loss": 0.4235,
+      "step": 12161
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.440171871317768e-05,
+      "loss": 0.372,
+      "step": 12162
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4400823742528229e-05,
+      "loss": 0.448,
+      "step": 12163
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4399928728160675e-05,
+      "loss": 0.4394,
+      "step": 12164
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4399033670083911e-05,
+      "loss": 0.3916,
+      "step": 12165
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4398138568306828e-05,
+      "loss": 0.3928,
+      "step": 12166
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4397243422838316e-05,
+      "loss": 0.3165,
+      "step": 12167
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.439634823368727e-05,
+      "loss": 0.3903,
+      "step": 12168
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.439545300086258e-05,
+      "loss": 0.3698,
+      "step": 12169
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4394557724373142e-05,
+      "loss": 0.4105,
+      "step": 12170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4393662404227851e-05,
+      "loss": 0.3228,
+      "step": 12171
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4392767040435596e-05,
+      "loss": 0.3253,
+      "step": 12172
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4391871633005273e-05,
+      "loss": 0.3345,
+      "step": 12173
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4390976181945782e-05,
+      "loss": 0.3532,
+      "step": 12174
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4390080687266013e-05,
+      "loss": 0.3457,
+      "step": 12175
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4389185148974864e-05,
+      "loss": 0.4285,
+      "step": 12176
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4388289567081232e-05,
+      "loss": 0.4268,
+      "step": 12177
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4387393941594014e-05,
+      "loss": 0.4224,
+      "step": 12178
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4386498272522105e-05,
+      "loss": 0.3782,
+      "step": 12179
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4385602559874406e-05,
+      "loss": 0.3867,
+      "step": 12180
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4384706803659811e-05,
+      "loss": 0.4302,
+      "step": 12181
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4383811003887218e-05,
+      "loss": 0.3472,
+      "step": 12182
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4382915160565534e-05,
+      "loss": 0.4829,
+      "step": 12183
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.438201927370365e-05,
+      "loss": 0.3978,
+      "step": 12184
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4381123343310468e-05,
+      "loss": 0.3494,
+      "step": 12185
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.438022736939489e-05,
+      "loss": 0.3365,
+      "step": 12186
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4379331351965815e-05,
+      "loss": 0.4014,
+      "step": 12187
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4378435291032144e-05,
+      "loss": 0.39,
+      "step": 12188
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.437753918660278e-05,
+      "loss": 0.3397,
+      "step": 12189
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4376643038686624e-05,
+      "loss": 0.4151,
+      "step": 12190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4375746847292578e-05,
+      "loss": 0.4075,
+      "step": 12191
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4374850612429546e-05,
+      "loss": 0.3162,
+      "step": 12192
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4373954334106433e-05,
+      "loss": 0.5208,
+      "step": 12193
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4373058012332137e-05,
+      "loss": 0.2989,
+      "step": 12194
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4372161647115563e-05,
+      "loss": 0.372,
+      "step": 12195
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4371265238465624e-05,
+      "loss": 0.3599,
+      "step": 12196
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4370368786391215e-05,
+      "loss": 0.3807,
+      "step": 12197
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4369472290901247e-05,
+      "loss": 0.4206,
+      "step": 12198
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4368575752004622e-05,
+      "loss": 0.3506,
+      "step": 12199
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4367679169710253e-05,
+      "loss": 0.3917,
+      "step": 12200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4366782544027038e-05,
+      "loss": 0.4098,
+      "step": 12201
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4365885874963889e-05,
+      "loss": 0.3539,
+      "step": 12202
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4364989162529717e-05,
+      "loss": 0.4388,
+      "step": 12203
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.436409240673342e-05,
+      "loss": 0.4097,
+      "step": 12204
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4363195607583919e-05,
+      "loss": 0.3675,
+      "step": 12205
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4362298765090112e-05,
+      "loss": 0.3818,
+      "step": 12206
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4361401879260912e-05,
+      "loss": 0.3287,
+      "step": 12207
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4360504950105232e-05,
+      "loss": 0.3687,
+      "step": 12208
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4359607977631978e-05,
+      "loss": 0.4289,
+      "step": 12209
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4358710961850062e-05,
+      "loss": 0.3389,
+      "step": 12210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4357813902768393e-05,
+      "loss": 0.3497,
+      "step": 12211
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4356916800395891e-05,
+      "loss": 0.4055,
+      "step": 12212
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4356019654741453e-05,
+      "loss": 0.384,
+      "step": 12213
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4355122465814005e-05,
+      "loss": 0.3936,
+      "step": 12214
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4354225233622457e-05,
+      "loss": 0.4901,
+      "step": 12215
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.435332795817571e-05,
+      "loss": 0.3882,
+      "step": 12216
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4352430639482698e-05,
+      "loss": 0.3759,
+      "step": 12217
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4351533277552319e-05,
+      "loss": 0.3937,
+      "step": 12218
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4350635872393492e-05,
+      "loss": 0.4462,
+      "step": 12219
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4349738424015134e-05,
+      "loss": 0.3747,
+      "step": 12220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4348840932426159e-05,
+      "loss": 0.3298,
+      "step": 12221
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4347943397635481e-05,
+      "loss": 0.4059,
+      "step": 12222
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.434704581965202e-05,
+      "loss": 0.364,
+      "step": 12223
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4346148198484688e-05,
+      "loss": 0.3918,
+      "step": 12224
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345250534142408e-05,
+      "loss": 0.4379,
+      "step": 12225
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4344352826634091e-05,
+      "loss": 0.3047,
+      "step": 12226
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.434345507596866e-05,
+      "loss": 0.4031,
+      "step": 12227
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.434255728215503e-05,
+      "loss": 0.3796,
+      "step": 12228
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4341659445202121e-05,
+      "loss": 0.3699,
+      "step": 12229
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4340761565118852e-05,
+      "loss": 0.4817,
+      "step": 12230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4339863641914143e-05,
+      "loss": 0.3208,
+      "step": 12231
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4338965675596915e-05,
+      "loss": 0.3972,
+      "step": 12232
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4338067666176084e-05,
+      "loss": 0.3983,
+      "step": 12233
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4337169613660575e-05,
+      "loss": 0.4022,
+      "step": 12234
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.433627151805931e-05,
+      "loss": 0.4721,
+      "step": 12235
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4335373379381208e-05,
+      "loss": 0.3012,
+      "step": 12236
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4334475197635194e-05,
+      "loss": 0.3752,
+      "step": 12237
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4333576972830185e-05,
+      "loss": 0.3945,
+      "step": 12238
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4332678704975114e-05,
+      "loss": 0.3574,
+      "step": 12239
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4331780394078893e-05,
+      "loss": 0.4554,
+      "step": 12240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4330882040150454e-05,
+      "loss": 0.3625,
+      "step": 12241
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4329983643198717e-05,
+      "loss": 0.4203,
+      "step": 12242
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4329085203232608e-05,
+      "loss": 0.4148,
+      "step": 12243
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4328186720261054e-05,
+      "loss": 0.326,
+      "step": 12244
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4327288194292978e-05,
+      "loss": 0.3753,
+      "step": 12245
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4326389625337312e-05,
+      "loss": 0.3744,
+      "step": 12246
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432549101340297e-05,
+      "loss": 0.3853,
+      "step": 12247
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432459235849889e-05,
+      "loss": 0.3416,
+      "step": 12248
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4323693660633997e-05,
+      "loss": 0.4322,
+      "step": 12249
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4322794919817216e-05,
+      "loss": 0.4277,
+      "step": 12250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4321896136057478e-05,
+      "loss": 0.4084,
+      "step": 12251
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4320997309363706e-05,
+      "loss": 0.4525,
+      "step": 12252
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4320098439744835e-05,
+      "loss": 0.3478,
+      "step": 12253
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4319199527209797e-05,
+      "loss": 0.3669,
+      "step": 12254
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4318300571767514e-05,
+      "loss": 0.4111,
+      "step": 12255
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4317401573426922e-05,
+      "loss": 0.3855,
+      "step": 12256
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4316502532196945e-05,
+      "loss": 0.4027,
+      "step": 12257
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4315603448086527e-05,
+      "loss": 0.3426,
+      "step": 12258
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4314704321104583e-05,
+      "loss": 0.3813,
+      "step": 12259
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4313805151260054e-05,
+      "loss": 0.4,
+      "step": 12260
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4312905938561876e-05,
+      "loss": 0.4133,
+      "step": 12261
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4312006683018974e-05,
+      "loss": 0.3843,
+      "step": 12262
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4311107384640286e-05,
+      "loss": 0.4295,
+      "step": 12263
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4310208043434744e-05,
+      "loss": 0.4891,
+      "step": 12264
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4309308659411282e-05,
+      "loss": 0.3536,
+      "step": 12265
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4308409232578837e-05,
+      "loss": 0.3735,
+      "step": 12266
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4307509762946341e-05,
+      "loss": 0.2958,
+      "step": 12267
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4306610250522732e-05,
+      "loss": 0.3575,
+      "step": 12268
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4305710695316944e-05,
+      "loss": 0.4634,
+      "step": 12269
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4304811097337913e-05,
+      "loss": 0.363,
+      "step": 12270
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4303911456594573e-05,
+      "loss": 0.4513,
+      "step": 12271
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4303011773095867e-05,
+      "loss": 0.328,
+      "step": 12272
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4302112046850733e-05,
+      "loss": 0.4515,
+      "step": 12273
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4301212277868102e-05,
+      "loss": 0.3802,
+      "step": 12274
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4300312466156917e-05,
+      "loss": 0.3835,
+      "step": 12275
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4299412611726116e-05,
+      "loss": 0.4651,
+      "step": 12276
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4298512714584638e-05,
+      "loss": 0.4109,
+      "step": 12277
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4297612774741426e-05,
+      "loss": 0.4076,
+      "step": 12278
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4296712792205412e-05,
+      "loss": 0.2937,
+      "step": 12279
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4295812766985544e-05,
+      "loss": 0.3562,
+      "step": 12280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4294912699090758e-05,
+      "loss": 0.5394,
+      "step": 12281
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4294012588530001e-05,
+      "loss": 0.4,
+      "step": 12282
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.429311243531221e-05,
+      "loss": 0.3989,
+      "step": 12283
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4292212239446326e-05,
+      "loss": 0.3621,
+      "step": 12284
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.42913120009413e-05,
+      "loss": 0.4001,
+      "step": 12285
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4290411719806066e-05,
+      "loss": 0.3947,
+      "step": 12286
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428951139604957e-05,
+      "loss": 0.4273,
+      "step": 12287
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4288611029680757e-05,
+      "loss": 0.3749,
+      "step": 12288
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428771062070857e-05,
+      "loss": 0.3814,
+      "step": 12289
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4286810169141955e-05,
+      "loss": 0.3946,
+      "step": 12290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4285909674989857e-05,
+      "loss": 0.3992,
+      "step": 12291
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4285009138261226e-05,
+      "loss": 0.4178,
+      "step": 12292
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4284108558964995e-05,
+      "loss": 0.4242,
+      "step": 12293
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4283207937110124e-05,
+      "loss": 0.3809,
+      "step": 12294
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4282307272705555e-05,
+      "loss": 0.4281,
+      "step": 12295
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428140656576023e-05,
+      "loss": 0.4063,
+      "step": 12296
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4280505816283107e-05,
+      "loss": 0.3697,
+      "step": 12297
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4279605024283127e-05,
+      "loss": 0.4754,
+      "step": 12298
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427870418976924e-05,
+      "loss": 0.4268,
+      "step": 12299
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4277803312750394e-05,
+      "loss": 0.284,
+      "step": 12300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4276902393235541e-05,
+      "loss": 0.3891,
+      "step": 12301
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4276001431233628e-05,
+      "loss": 0.3928,
+      "step": 12302
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4275100426753608e-05,
+      "loss": 0.4546,
+      "step": 12303
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4274199379804432e-05,
+      "loss": 0.3894,
+      "step": 12304
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4273298290395043e-05,
+      "loss": 0.407,
+      "step": 12305
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4272397158534404e-05,
+      "loss": 0.3673,
+      "step": 12306
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427149598423146e-05,
+      "loss": 0.416,
+      "step": 12307
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4270594767495167e-05,
+      "loss": 0.3936,
+      "step": 12308
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4269693508334473e-05,
+      "loss": 0.456,
+      "step": 12309
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4268792206758335e-05,
+      "loss": 0.3742,
+      "step": 12310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4267890862775704e-05,
+      "loss": 0.3317,
+      "step": 12311
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4266989476395538e-05,
+      "loss": 0.3323,
+      "step": 12312
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4266088047626787e-05,
+      "loss": 0.4537,
+      "step": 12313
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.426518657647841e-05,
+      "loss": 0.4145,
+      "step": 12314
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4264285062959358e-05,
+      "loss": 0.393,
+      "step": 12315
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4263383507078591e-05,
+      "loss": 0.4037,
+      "step": 12316
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4262481908845062e-05,
+      "loss": 0.3431,
+      "step": 12317
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4261580268267726e-05,
+      "loss": 0.3522,
+      "step": 12318
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4260678585355546e-05,
+      "loss": 0.4046,
+      "step": 12319
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4259776860117476e-05,
+      "loss": 0.3733,
+      "step": 12320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425887509256247e-05,
+      "loss": 0.4477,
+      "step": 12321
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4257973282699493e-05,
+      "loss": 0.3886,
+      "step": 12322
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.42570714305375e-05,
+      "loss": 0.3944,
+      "step": 12323
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4256169536085449e-05,
+      "loss": 0.4286,
+      "step": 12324
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4255267599352301e-05,
+      "loss": 0.4133,
+      "step": 12325
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425436562034702e-05,
+      "loss": 0.4296,
+      "step": 12326
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4253463599078557e-05,
+      "loss": 0.3012,
+      "step": 12327
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425256153555588e-05,
+      "loss": 0.3156,
+      "step": 12328
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425165942978795e-05,
+      "loss": 0.3801,
+      "step": 12329
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4250757281783723e-05,
+      "loss": 0.4492,
+      "step": 12330
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4249855091552167e-05,
+      "loss": 0.4252,
+      "step": 12331
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4248952859102242e-05,
+      "loss": 0.3627,
+      "step": 12332
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4248050584442911e-05,
+      "loss": 0.3888,
+      "step": 12333
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4247148267583137e-05,
+      "loss": 0.4015,
+      "step": 12334
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4246245908531883e-05,
+      "loss": 0.4027,
+      "step": 12335
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4245343507298117e-05,
+      "loss": 0.3796,
+      "step": 12336
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.42444410638908e-05,
+      "loss": 0.3947,
+      "step": 12337
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4243538578318898e-05,
+      "loss": 0.3738,
+      "step": 12338
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4242636050591372e-05,
+      "loss": 0.5093,
+      "step": 12339
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4241733480717196e-05,
+      "loss": 0.371,
+      "step": 12340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424083086870533e-05,
+      "loss": 0.4295,
+      "step": 12341
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423992821456474e-05,
+      "loss": 0.4079,
+      "step": 12342
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.42390255183044e-05,
+      "loss": 0.3904,
+      "step": 12343
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423812277993327e-05,
+      "loss": 0.379,
+      "step": 12344
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4237219999460323e-05,
+      "loss": 0.375,
+      "step": 12345
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4236317176894523e-05,
+      "loss": 0.3702,
+      "step": 12346
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4235414312244841e-05,
+      "loss": 0.289,
+      "step": 12347
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4234511405520248e-05,
+      "loss": 0.3561,
+      "step": 12348
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423360845672971e-05,
+      "loss": 0.3362,
+      "step": 12349
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4232705465882199e-05,
+      "loss": 0.3698,
+      "step": 12350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231802432986683e-05,
+      "loss": 0.4653,
+      "step": 12351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423089935805214e-05,
+      "loss": 0.379,
+      "step": 12352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4229996241087529e-05,
+      "loss": 0.3669,
+      "step": 12353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4229093082101833e-05,
+      "loss": 0.3576,
+      "step": 12354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4228189881104016e-05,
+      "loss": 0.4162,
+      "step": 12355
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4227286638103057e-05,
+      "loss": 0.3975,
+      "step": 12356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4226383353107924e-05,
+      "loss": 0.4105,
+      "step": 12357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4225480026127592e-05,
+      "loss": 0.456,
+      "step": 12358
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4224576657171036e-05,
+      "loss": 0.3907,
+      "step": 12359
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4223673246247232e-05,
+      "loss": 0.4168,
+      "step": 12360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4222769793365144e-05,
+      "loss": 0.4009,
+      "step": 12361
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4221866298533763e-05,
+      "loss": 0.3787,
+      "step": 12362
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.422096276176205e-05,
+      "loss": 0.3128,
+      "step": 12363
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220059183058986e-05,
+      "loss": 0.3356,
+      "step": 12364
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4219155562433551e-05,
+      "loss": 0.5073,
+      "step": 12365
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4218251899894717e-05,
+      "loss": 0.3605,
+      "step": 12366
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4217348195451463e-05,
+      "loss": 0.4174,
+      "step": 12367
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4216444449112768e-05,
+      "loss": 0.33,
+      "step": 12368
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4215540660887609e-05,
+      "loss": 0.4863,
+      "step": 12369
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4214636830784958e-05,
+      "loss": 0.4005,
+      "step": 12370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.42137329588138e-05,
+      "loss": 0.442,
+      "step": 12371
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4212829044983116e-05,
+      "loss": 0.3497,
+      "step": 12372
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421192508930188e-05,
+      "loss": 0.4237,
+      "step": 12373
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4211021091779078e-05,
+      "loss": 0.3571,
+      "step": 12374
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4210117052423683e-05,
+      "loss": 0.4144,
+      "step": 12375
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4209212971244682e-05,
+      "loss": 0.4224,
+      "step": 12376
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4208308848251053e-05,
+      "loss": 0.4899,
+      "step": 12377
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.420740468345178e-05,
+      "loss": 0.3827,
+      "step": 12378
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4206500476855843e-05,
+      "loss": 0.4018,
+      "step": 12379
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4205596228472222e-05,
+      "loss": 0.4156,
+      "step": 12380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4204691938309909e-05,
+      "loss": 0.4058,
+      "step": 12381
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4203787606377877e-05,
+      "loss": 0.3774,
+      "step": 12382
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4202883232685113e-05,
+      "loss": 0.3371,
+      "step": 12383
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4201978817240607e-05,
+      "loss": 0.3493,
+      "step": 12384
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4201074360053332e-05,
+      "loss": 0.3761,
+      "step": 12385
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4200169861132285e-05,
+      "loss": 0.3938,
+      "step": 12386
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4199265320486444e-05,
+      "loss": 0.3717,
+      "step": 12387
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4198360738124797e-05,
+      "loss": 0.4324,
+      "step": 12388
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419745611405633e-05,
+      "loss": 0.3568,
+      "step": 12389
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4196551448290027e-05,
+      "loss": 0.3625,
+      "step": 12390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4195646740834881e-05,
+      "loss": 0.3075,
+      "step": 12391
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4194741991699875e-05,
+      "loss": 0.3963,
+      "step": 12392
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4193837200893995e-05,
+      "loss": 0.4771,
+      "step": 12393
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4192932368426234e-05,
+      "loss": 0.3698,
+      "step": 12394
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4192027494305578e-05,
+      "loss": 0.3886,
+      "step": 12395
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4191122578541018e-05,
+      "loss": 0.3492,
+      "step": 12396
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4190217621141542e-05,
+      "loss": 0.3663,
+      "step": 12397
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4189312622116141e-05,
+      "loss": 0.4249,
+      "step": 12398
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4188407581473802e-05,
+      "loss": 0.2752,
+      "step": 12399
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.418750249922352e-05,
+      "loss": 0.3438,
+      "step": 12400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4186597375374283e-05,
+      "loss": 0.3338,
+      "step": 12401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4185692209935085e-05,
+      "loss": 0.326,
+      "step": 12402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4184787002914914e-05,
+      "loss": 0.3917,
+      "step": 12403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.418388175432277e-05,
+      "loss": 0.3804,
+      "step": 12404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4182976464167637e-05,
+      "loss": 0.4314,
+      "step": 12405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4182071132458518e-05,
+      "loss": 0.4384,
+      "step": 12406
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4181165759204394e-05,
+      "loss": 0.3175,
+      "step": 12407
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4180260344414272e-05,
+      "loss": 0.3227,
+      "step": 12408
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4179354888097136e-05,
+      "loss": 0.4723,
+      "step": 12409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4178449390261988e-05,
+      "loss": 0.4347,
+      "step": 12410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4177543850917818e-05,
+      "loss": 0.3336,
+      "step": 12411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4176638270073628e-05,
+      "loss": 0.3995,
+      "step": 12412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4175732647738407e-05,
+      "loss": 0.3738,
+      "step": 12413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4174826983921157e-05,
+      "loss": 0.3796,
+      "step": 12414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4173921278630874e-05,
+      "loss": 0.3143,
+      "step": 12415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.417301553187655e-05,
+      "loss": 0.4636,
+      "step": 12416
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4172109743667188e-05,
+      "loss": 0.3567,
+      "step": 12417
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4171203914011789e-05,
+      "loss": 0.383,
+      "step": 12418
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4170298042919344e-05,
+      "loss": 0.4354,
+      "step": 12419
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4169392130398859e-05,
+      "loss": 0.4299,
+      "step": 12420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4168486176459326e-05,
+      "loss": 0.35,
+      "step": 12421
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4167580181109747e-05,
+      "loss": 0.3826,
+      "step": 12422
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4166674144359132e-05,
+      "loss": 0.2694,
+      "step": 12423
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416576806621647e-05,
+      "loss": 0.3888,
+      "step": 12424
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4164861946690765e-05,
+      "loss": 0.392,
+      "step": 12425
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4163955785791019e-05,
+      "loss": 0.424,
+      "step": 12426
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4163049583526238e-05,
+      "loss": 0.3658,
+      "step": 12427
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4162143339905415e-05,
+      "loss": 0.3528,
+      "step": 12428
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416123705493756e-05,
+      "loss": 0.4126,
+      "step": 12429
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4160330728631678e-05,
+      "loss": 0.3414,
+      "step": 12430
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4159424360996761e-05,
+      "loss": 0.4237,
+      "step": 12431
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4158517952041826e-05,
+      "loss": 0.3648,
+      "step": 12432
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4157611501775872e-05,
+      "loss": 0.3985,
+      "step": 12433
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4156705010207902e-05,
+      "loss": 0.3791,
+      "step": 12434
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4155798477346922e-05,
+      "loss": 0.3881,
+      "step": 12435
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4154891903201941e-05,
+      "loss": 0.2994,
+      "step": 12436
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4153985287781962e-05,
+      "loss": 0.3943,
+      "step": 12437
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4153078631095988e-05,
+      "loss": 0.3894,
+      "step": 12438
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4152171933153035e-05,
+      "loss": 0.2785,
+      "step": 12439
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4151265193962106e-05,
+      "loss": 0.4147,
+      "step": 12440
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.41503584135322e-05,
+      "loss": 0.3639,
+      "step": 12441
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4149451591872336e-05,
+      "loss": 0.3579,
+      "step": 12442
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.414854472899152e-05,
+      "loss": 0.4022,
+      "step": 12443
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.414763782489876e-05,
+      "loss": 0.3857,
+      "step": 12444
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4146730879603063e-05,
+      "loss": 0.4287,
+      "step": 12445
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414582389311344e-05,
+      "loss": 0.3655,
+      "step": 12446
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4144916865438903e-05,
+      "loss": 0.3525,
+      "step": 12447
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4144009796588463e-05,
+      "loss": 0.423,
+      "step": 12448
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4143102686571126e-05,
+      "loss": 0.3623,
+      "step": 12449
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4142195535395907e-05,
+      "loss": 0.3808,
+      "step": 12450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414128834307182e-05,
+      "loss": 0.4377,
+      "step": 12451
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414038110960787e-05,
+      "loss": 0.3276,
+      "step": 12452
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4139473835013074e-05,
+      "loss": 0.4955,
+      "step": 12453
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4138566519296446e-05,
+      "loss": 0.3542,
+      "step": 12454
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4137659162466999e-05,
+      "loss": 0.3567,
+      "step": 12455
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4136751764533745e-05,
+      "loss": 0.4196,
+      "step": 12456
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4135844325505698e-05,
+      "loss": 0.3785,
+      "step": 12457
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4134936845391873e-05,
+      "loss": 0.3571,
+      "step": 12458
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4134029324201286e-05,
+      "loss": 0.4231,
+      "step": 12459
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4133121761942953e-05,
+      "loss": 0.2736,
+      "step": 12460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4132214158625888e-05,
+      "loss": 0.3331,
+      "step": 12461
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4131306514259104e-05,
+      "loss": 0.3903,
+      "step": 12462
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4130398828851626e-05,
+      "loss": 0.3236,
+      "step": 12463
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4129491102412466e-05,
+      "loss": 0.362,
+      "step": 12464
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4128583334950638e-05,
+      "loss": 0.3619,
+      "step": 12465
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4127675526475168e-05,
+      "loss": 0.4334,
+      "step": 12466
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412676767699507e-05,
+      "loss": 0.3677,
+      "step": 12467
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4125859786519361e-05,
+      "loss": 0.3562,
+      "step": 12468
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412495185505706e-05,
+      "loss": 0.389,
+      "step": 12469
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412404388261719e-05,
+      "loss": 0.4562,
+      "step": 12470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4123135869208766e-05,
+      "loss": 0.3919,
+      "step": 12471
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4122227814840814e-05,
+      "loss": 0.3882,
+      "step": 12472
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4121319719522352e-05,
+      "loss": 0.4218,
+      "step": 12473
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.41204115832624e-05,
+      "loss": 0.3664,
+      "step": 12474
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.411950340606998e-05,
+      "loss": 0.3496,
+      "step": 12475
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4118595187954116e-05,
+      "loss": 0.3444,
+      "step": 12476
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4117686928923829e-05,
+      "loss": 0.3897,
+      "step": 12477
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4116778628988139e-05,
+      "loss": 0.3838,
+      "step": 12478
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4115870288156071e-05,
+      "loss": 0.3745,
+      "step": 12479
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.411496190643665e-05,
+      "loss": 0.4467,
+      "step": 12480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4114053483838898e-05,
+      "loss": 0.3457,
+      "step": 12481
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.411314502037184e-05,
+      "loss": 0.339,
+      "step": 12482
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4112236516044505e-05,
+      "loss": 0.3849,
+      "step": 12483
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4111327970865909e-05,
+      "loss": 0.4487,
+      "step": 12484
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4110419384845086e-05,
+      "loss": 0.3704,
+      "step": 12485
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4109510757991058e-05,
+      "loss": 0.3772,
+      "step": 12486
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4108602090312851e-05,
+      "loss": 0.4545,
+      "step": 12487
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4107693381819494e-05,
+      "loss": 0.4524,
+      "step": 12488
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4106784632520012e-05,
+      "loss": 0.3775,
+      "step": 12489
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4105875842423433e-05,
+      "loss": 0.4241,
+      "step": 12490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4104967011538785e-05,
+      "loss": 0.3538,
+      "step": 12491
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4104058139875099e-05,
+      "loss": 0.434,
+      "step": 12492
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4103149227441399e-05,
+      "loss": 0.4202,
+      "step": 12493
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4102240274246718e-05,
+      "loss": 0.3429,
+      "step": 12494
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.410133128030009e-05,
+      "loss": 0.4156,
+      "step": 12495
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4100422245610532e-05,
+      "loss": 0.4479,
+      "step": 12496
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4099513170187086e-05,
+      "loss": 0.4065,
+      "step": 12497
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4098604054038778e-05,
+      "loss": 0.348,
+      "step": 12498
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4097694897174637e-05,
+      "loss": 0.4537,
+      "step": 12499
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4096785699603701e-05,
+      "loss": 0.321,
+      "step": 12500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4095876461334998e-05,
+      "loss": 0.3751,
+      "step": 12501
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4094967182377558e-05,
+      "loss": 0.3856,
+      "step": 12502
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.409405786274042e-05,
+      "loss": 0.3112,
+      "step": 12503
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4093148502432614e-05,
+      "loss": 0.4588,
+      "step": 12504
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4092239101463174e-05,
+      "loss": 0.4443,
+      "step": 12505
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4091329659841134e-05,
+      "loss": 0.3927,
+      "step": 12506
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4090420177575527e-05,
+      "loss": 0.3862,
+      "step": 12507
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4089510654675389e-05,
+      "loss": 0.3075,
+      "step": 12508
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4088601091149758e-05,
+      "loss": 0.3772,
+      "step": 12509
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4087691487007666e-05,
+      "loss": 0.2817,
+      "step": 12510
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4086781842258148e-05,
+      "loss": 0.3578,
+      "step": 12511
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4085872156910245e-05,
+      "loss": 0.4323,
+      "step": 12512
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4084962430972992e-05,
+      "loss": 0.4123,
+      "step": 12513
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4084052664455424e-05,
+      "loss": 0.3556,
+      "step": 12514
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.408314285736658e-05,
+      "loss": 0.4702,
+      "step": 12515
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.40822330097155e-05,
+      "loss": 0.28,
+      "step": 12516
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4081323121511221e-05,
+      "loss": 0.361,
+      "step": 12517
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4080413192762783e-05,
+      "loss": 0.435,
+      "step": 12518
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4079503223479226e-05,
+      "loss": 0.4432,
+      "step": 12519
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4078593213669582e-05,
+      "loss": 0.4031,
+      "step": 12520
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4077683163342903e-05,
+      "loss": 0.3498,
+      "step": 12521
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407677307250822e-05,
+      "loss": 0.4059,
+      "step": 12522
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4075862941174579e-05,
+      "loss": 0.3601,
+      "step": 12523
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407495276935102e-05,
+      "loss": 0.2627,
+      "step": 12524
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4074042557046584e-05,
+      "loss": 0.352,
+      "step": 12525
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4073132304270313e-05,
+      "loss": 0.487,
+      "step": 12526
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4072222011031251e-05,
+      "loss": 0.3547,
+      "step": 12527
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407131167733844e-05,
+      "loss": 0.3422,
+      "step": 12528
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4070401303200925e-05,
+      "loss": 0.4021,
+      "step": 12529
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4069490888627744e-05,
+      "loss": 0.3642,
+      "step": 12530
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.406858043362795e-05,
+      "loss": 0.3631,
+      "step": 12531
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.406766993821058e-05,
+      "loss": 0.3742,
+      "step": 12532
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.406675940238468e-05,
+      "loss": 0.3953,
+      "step": 12533
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4065848826159302e-05,
+      "loss": 0.4187,
+      "step": 12534
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064938209543483e-05,
+      "loss": 0.4202,
+      "step": 12535
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064027552546275e-05,
+      "loss": 0.4127,
+      "step": 12536
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4063116855176721e-05,
+      "loss": 0.4444,
+      "step": 12537
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4062206117443871e-05,
+      "loss": 0.4568,
+      "step": 12538
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4061295339356769e-05,
+      "loss": 0.4121,
+      "step": 12539
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4060384520924464e-05,
+      "loss": 0.4501,
+      "step": 12540
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4059473662156008e-05,
+      "loss": 0.3373,
+      "step": 12541
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4058562763060441e-05,
+      "loss": 0.5041,
+      "step": 12542
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4057651823646824e-05,
+      "loss": 0.498,
+      "step": 12543
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4056740843924192e-05,
+      "loss": 0.2627,
+      "step": 12544
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4055829823901604e-05,
+      "loss": 0.3802,
+      "step": 12545
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4054918763588112e-05,
+      "loss": 0.3509,
+      "step": 12546
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4054007662992761e-05,
+      "loss": 0.3724,
+      "step": 12547
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4053096522124604e-05,
+      "loss": 0.4251,
+      "step": 12548
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.405218534099269e-05,
+      "loss": 0.4009,
+      "step": 12549
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4051274119606076e-05,
+      "loss": 0.4658,
+      "step": 12550
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4050362857973807e-05,
+      "loss": 0.3693,
+      "step": 12551
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.404945155610494e-05,
+      "loss": 0.5009,
+      "step": 12552
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4048540214008533e-05,
+      "loss": 0.3048,
+      "step": 12553
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4047628831693627e-05,
+      "loss": 0.4143,
+      "step": 12554
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4046717409169288e-05,
+      "loss": 0.4077,
+      "step": 12555
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4045805946444561e-05,
+      "loss": 0.3541,
+      "step": 12556
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4044894443528505e-05,
+      "loss": 0.432,
+      "step": 12557
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4043982900430176e-05,
+      "loss": 0.4505,
+      "step": 12558
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4043071317158626e-05,
+      "loss": 0.4206,
+      "step": 12559
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.404215969372291e-05,
+      "loss": 0.3457,
+      "step": 12560
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.404124803013209e-05,
+      "loss": 0.3291,
+      "step": 12561
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4040336326395217e-05,
+      "loss": 0.3313,
+      "step": 12562
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4039424582521348e-05,
+      "loss": 0.4154,
+      "step": 12563
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4038512798519546e-05,
+      "loss": 0.3693,
+      "step": 12564
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4037600974398863e-05,
+      "loss": 0.3911,
+      "step": 12565
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403668911016836e-05,
+      "loss": 0.3834,
+      "step": 12566
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4035777205837093e-05,
+      "loss": 0.3507,
+      "step": 12567
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4034865261414123e-05,
+      "loss": 0.4163,
+      "step": 12568
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4033953276908509e-05,
+      "loss": 0.3884,
+      "step": 12569
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403304125232931e-05,
+      "loss": 0.4176,
+      "step": 12570
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4032129187685588e-05,
+      "loss": 0.3807,
+      "step": 12571
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4031217082986402e-05,
+      "loss": 0.3037,
+      "step": 12572
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403030493824081e-05,
+      "loss": 0.3438,
+      "step": 12573
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.402939275345788e-05,
+      "loss": 0.326,
+      "step": 12574
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4028480528646669e-05,
+      "loss": 0.3899,
+      "step": 12575
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4027568263816238e-05,
+      "loss": 0.3667,
+      "step": 12576
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4026655958975655e-05,
+      "loss": 0.4673,
+      "step": 12577
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4025743614133977e-05,
+      "loss": 0.4055,
+      "step": 12578
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.402483122930027e-05,
+      "loss": 0.3523,
+      "step": 12579
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4023918804483598e-05,
+      "loss": 0.3386,
+      "step": 12580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4023006339693027e-05,
+      "loss": 0.3693,
+      "step": 12581
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4022093834937618e-05,
+      "loss": 0.3563,
+      "step": 12582
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4021181290226436e-05,
+      "loss": 0.3505,
+      "step": 12583
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.402026870556855e-05,
+      "loss": 0.3614,
+      "step": 12584
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4019356080973017e-05,
+      "loss": 0.3388,
+      "step": 12585
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4018443416448912e-05,
+      "loss": 0.4147,
+      "step": 12586
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.40175307120053e-05,
+      "loss": 0.4537,
+      "step": 12587
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4016617967651244e-05,
+      "loss": 0.4224,
+      "step": 12588
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4015705183395819e-05,
+      "loss": 0.4088,
+      "step": 12589
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4014792359248082e-05,
+      "loss": 0.4111,
+      "step": 12590
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4013879495217109e-05,
+      "loss": 0.3556,
+      "step": 12591
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4012966591311962e-05,
+      "loss": 0.371,
+      "step": 12592
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4012053647541718e-05,
+      "loss": 0.42,
+      "step": 12593
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4011140663915442e-05,
+      "loss": 0.3447,
+      "step": 12594
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4010227640442202e-05,
+      "loss": 0.3818,
+      "step": 12595
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.400931457713107e-05,
+      "loss": 0.4419,
+      "step": 12596
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4008401473991115e-05,
+      "loss": 0.4312,
+      "step": 12597
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4007488331031409e-05,
+      "loss": 0.3014,
+      "step": 12598
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4006575148261028e-05,
+      "loss": 0.3739,
+      "step": 12599
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4005661925689032e-05,
+      "loss": 0.4149,
+      "step": 12600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4004748663324506e-05,
+      "loss": 0.3458,
+      "step": 12601
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4003835361176513e-05,
+      "loss": 0.3435,
+      "step": 12602
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4002922019254127e-05,
+      "loss": 0.3818,
+      "step": 12603
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4002008637566427e-05,
+      "loss": 0.3207,
+      "step": 12604
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.400109521612248e-05,
+      "loss": 0.3851,
+      "step": 12605
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4000181754931363e-05,
+      "loss": 0.3637,
+      "step": 12606
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3999268254002153e-05,
+      "loss": 0.3723,
+      "step": 12607
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3998354713343922e-05,
+      "loss": 0.3114,
+      "step": 12608
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3997441132965743e-05,
+      "loss": 0.3314,
+      "step": 12609
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3996527512876692e-05,
+      "loss": 0.3569,
+      "step": 12610
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3995613853085852e-05,
+      "loss": 0.3642,
+      "step": 12611
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.399470015360229e-05,
+      "loss": 0.3728,
+      "step": 12612
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.399378641443509e-05,
+      "loss": 0.3827,
+      "step": 12613
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3992872635593321e-05,
+      "loss": 0.3579,
+      "step": 12614
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3991958817086068e-05,
+      "loss": 0.3972,
+      "step": 12615
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3991044958922409e-05,
+      "loss": 0.4575,
+      "step": 12616
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3990131061111418e-05,
+      "loss": 0.3846,
+      "step": 12617
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3989217123662178e-05,
+      "loss": 0.3454,
+      "step": 12618
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3988303146583759e-05,
+      "loss": 0.3796,
+      "step": 12619
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3987389129885254e-05,
+      "loss": 0.3595,
+      "step": 12620
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3986475073575734e-05,
+      "loss": 0.3416,
+      "step": 12621
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3985560977664279e-05,
+      "loss": 0.3995,
+      "step": 12622
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3984646842159974e-05,
+      "loss": 0.3867,
+      "step": 12623
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.39837326670719e-05,
+      "loss": 0.3528,
+      "step": 12624
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3982818452409132e-05,
+      "loss": 0.3801,
+      "step": 12625
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.398190419818076e-05,
+      "loss": 0.4198,
+      "step": 12626
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3980989904395864e-05,
+      "loss": 0.3803,
+      "step": 12627
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3980075571063523e-05,
+      "loss": 0.3633,
+      "step": 12628
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3979161198192821e-05,
+      "loss": 0.5167,
+      "step": 12629
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3978246785792848e-05,
+      "loss": 0.354,
+      "step": 12630
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3977332333872678e-05,
+      "loss": 0.3527,
+      "step": 12631
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.39764178424414e-05,
+      "loss": 0.3595,
+      "step": 12632
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3975503311508103e-05,
+      "loss": 0.4215,
+      "step": 12633
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3974588741081862e-05,
+      "loss": 0.4013,
+      "step": 12634
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3973674131171775e-05,
+      "loss": 0.3346,
+      "step": 12635
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3972759481786917e-05,
+      "loss": 0.3819,
+      "step": 12636
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.397184479293638e-05,
+      "loss": 0.367,
+      "step": 12637
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3970930064629245e-05,
+      "loss": 0.4472,
+      "step": 12638
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3970015296874607e-05,
+      "loss": 0.465,
+      "step": 12639
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3969100489681544e-05,
+      "loss": 0.4023,
+      "step": 12640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3968185643059152e-05,
+      "loss": 0.3956,
+      "step": 12641
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3967270757016515e-05,
+      "loss": 0.3493,
+      "step": 12642
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3966355831562721e-05,
+      "loss": 0.3815,
+      "step": 12643
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3965440866706863e-05,
+      "loss": 0.4027,
+      "step": 12644
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3964525862458029e-05,
+      "loss": 0.4143,
+      "step": 12645
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3963610818825305e-05,
+      "loss": 0.3786,
+      "step": 12646
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3962695735817785e-05,
+      "loss": 0.351,
+      "step": 12647
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3961780613444558e-05,
+      "loss": 0.4299,
+      "step": 12648
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3960865451714712e-05,
+      "loss": 0.4079,
+      "step": 12649
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3959950250637344e-05,
+      "loss": 0.315,
+      "step": 12650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3959035010221542e-05,
+      "loss": 0.302,
+      "step": 12651
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3958119730476399e-05,
+      "loss": 0.3401,
+      "step": 12652
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3957204411411007e-05,
+      "loss": 0.3715,
+      "step": 12653
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3956289053034461e-05,
+      "loss": 0.4252,
+      "step": 12654
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3955373655355852e-05,
+      "loss": 0.4233,
+      "step": 12655
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3954458218384272e-05,
+      "loss": 0.3653,
+      "step": 12656
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.395354274212882e-05,
+      "loss": 0.3643,
+      "step": 12657
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3952627226598586e-05,
+      "loss": 0.3052,
+      "step": 12658
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3951711671802666e-05,
+      "loss": 0.3843,
+      "step": 12659
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3950796077750156e-05,
+      "loss": 0.3417,
+      "step": 12660
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3949880444450153e-05,
+      "loss": 0.3337,
+      "step": 12661
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3948964771911748e-05,
+      "loss": 0.3634,
+      "step": 12662
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3948049060144042e-05,
+      "loss": 0.3591,
+      "step": 12663
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.394713330915613e-05,
+      "loss": 0.3474,
+      "step": 12664
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.394621751895711e-05,
+      "loss": 0.3263,
+      "step": 12665
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.394530168955608e-05,
+      "loss": 0.3486,
+      "step": 12666
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3944385820962134e-05,
+      "loss": 0.3526,
+      "step": 12667
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3943469913184374e-05,
+      "loss": 0.4743,
+      "step": 12668
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3942553966231898e-05,
+      "loss": 0.3938,
+      "step": 12669
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3941637980113805e-05,
+      "loss": 0.3929,
+      "step": 12670
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3940721954839194e-05,
+      "loss": 0.3502,
+      "step": 12671
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3939805890417167e-05,
+      "loss": 0.4159,
+      "step": 12672
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.393888978685682e-05,
+      "loss": 0.4003,
+      "step": 12673
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3937973644167256e-05,
+      "loss": 0.4018,
+      "step": 12674
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3937057462357578e-05,
+      "loss": 0.4132,
+      "step": 12675
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3936141241436885e-05,
+      "loss": 0.314,
+      "step": 12676
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3935224981414278e-05,
+      "loss": 0.4298,
+      "step": 12677
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3934308682298863e-05,
+      "loss": 0.3732,
+      "step": 12678
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.393339234409974e-05,
+      "loss": 0.4801,
+      "step": 12679
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3932475966826007e-05,
+      "loss": 0.3937,
+      "step": 12680
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3931559550486777e-05,
+      "loss": 0.3751,
+      "step": 12681
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3930643095091147e-05,
+      "loss": 0.3727,
+      "step": 12682
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3929726600648226e-05,
+      "loss": 0.4192,
+      "step": 12683
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3928810067167113e-05,
+      "loss": 0.3809,
+      "step": 12684
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.392789349465692e-05,
+      "loss": 0.405,
+      "step": 12685
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3926976883126744e-05,
+      "loss": 0.4445,
+      "step": 12686
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3926060232585697e-05,
+      "loss": 0.3629,
+      "step": 12687
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3925143543042883e-05,
+      "loss": 0.3562,
+      "step": 12688
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3924226814507406e-05,
+      "loss": 0.3903,
+      "step": 12689
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3923310046988377e-05,
+      "loss": 0.3631,
+      "step": 12690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.39223932404949e-05,
+      "loss": 0.4838,
+      "step": 12691
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3921476395036085e-05,
+      "loss": 0.3571,
+      "step": 12692
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3920559510621037e-05,
+      "loss": 0.3996,
+      "step": 12693
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3919642587258869e-05,
+      "loss": 0.4195,
+      "step": 12694
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3918725624958684e-05,
+      "loss": 0.3836,
+      "step": 12695
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3917808623729597e-05,
+      "loss": 0.4052,
+      "step": 12696
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3916891583580714e-05,
+      "loss": 0.3278,
+      "step": 12697
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3915974504521147e-05,
+      "loss": 0.3445,
+      "step": 12698
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3915057386560005e-05,
+      "loss": 0.3709,
+      "step": 12699
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3914140229706398e-05,
+      "loss": 0.4207,
+      "step": 12700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3913223033969439e-05,
+      "loss": 0.3817,
+      "step": 12701
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3912305799358234e-05,
+      "loss": 0.404,
+      "step": 12702
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3911388525881905e-05,
+      "loss": 0.3951,
+      "step": 12703
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3910471213549556e-05,
+      "loss": 0.3027,
+      "step": 12704
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3909553862370305e-05,
+      "loss": 0.3645,
+      "step": 12705
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3908636472353259e-05,
+      "loss": 0.4303,
+      "step": 12706
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.390771904350754e-05,
+      "loss": 0.3613,
+      "step": 12707
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3906801575842249e-05,
+      "loss": 0.3481,
+      "step": 12708
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3905884069366511e-05,
+      "loss": 0.3759,
+      "step": 12709
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3904966524089439e-05,
+      "loss": 0.3197,
+      "step": 12710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3904048940020144e-05,
+      "loss": 0.3534,
+      "step": 12711
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3903131317167744e-05,
+      "loss": 0.3435,
+      "step": 12712
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3902213655541354e-05,
+      "loss": 0.4375,
+      "step": 12713
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.390129595515009e-05,
+      "loss": 0.4666,
+      "step": 12714
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3900378216003073e-05,
+      "loss": 0.3297,
+      "step": 12715
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3899460438109412e-05,
+      "loss": 0.3794,
+      "step": 12716
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3898542621478228e-05,
+      "loss": 0.3497,
+      "step": 12717
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.389762476611864e-05,
+      "loss": 0.3286,
+      "step": 12718
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3896706872039767e-05,
+      "loss": 0.3831,
+      "step": 12719
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3895788939250719e-05,
+      "loss": 0.3205,
+      "step": 12720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3894870967760625e-05,
+      "loss": 0.3362,
+      "step": 12721
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.38939529575786e-05,
+      "loss": 0.3246,
+      "step": 12722
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3893034908713763e-05,
+      "loss": 0.4071,
+      "step": 12723
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3892116821175236e-05,
+      "loss": 0.41,
+      "step": 12724
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3891198694972139e-05,
+      "loss": 0.4168,
+      "step": 12725
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3890280530113591e-05,
+      "loss": 0.3675,
+      "step": 12726
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3889362326608714e-05,
+      "loss": 0.3616,
+      "step": 12727
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3888444084466628e-05,
+      "loss": 0.3288,
+      "step": 12728
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3887525803696455e-05,
+      "loss": 0.3453,
+      "step": 12729
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3886607484307321e-05,
+      "loss": 0.4383,
+      "step": 12730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3885689126308348e-05,
+      "loss": 0.3609,
+      "step": 12731
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3884770729708654e-05,
+      "loss": 0.3041,
+      "step": 12732
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3883852294517369e-05,
+      "loss": 0.4203,
+      "step": 12733
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3882933820743612e-05,
+      "loss": 0.3889,
+      "step": 12734
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3882015308396508e-05,
+      "loss": 0.4193,
+      "step": 12735
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3881096757485184e-05,
+      "loss": 0.3502,
+      "step": 12736
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3880178168018761e-05,
+      "loss": 0.3409,
+      "step": 12737
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.387925954000637e-05,
+      "loss": 0.3579,
+      "step": 12738
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3878340873457128e-05,
+      "loss": 0.2965,
+      "step": 12739
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.387742216838017e-05,
+      "loss": 0.4153,
+      "step": 12740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3876503424784617e-05,
+      "loss": 0.357,
+      "step": 12741
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.38755846426796e-05,
+      "loss": 0.4174,
+      "step": 12742
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3874665822074243e-05,
+      "loss": 0.3692,
+      "step": 12743
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3873746962977675e-05,
+      "loss": 0.3635,
+      "step": 12744
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3872828065399023e-05,
+      "loss": 0.3981,
+      "step": 12745
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3871909129347416e-05,
+      "loss": 0.3464,
+      "step": 12746
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3870990154831982e-05,
+      "loss": 0.4004,
+      "step": 12747
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3870071141861852e-05,
+      "loss": 0.3509,
+      "step": 12748
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3869152090446156e-05,
+      "loss": 0.3645,
+      "step": 12749
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386823300059402e-05,
+      "loss": 0.4595,
+      "step": 12750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3867313872314578e-05,
+      "loss": 0.4191,
+      "step": 12751
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3866394705616956e-05,
+      "loss": 0.3218,
+      "step": 12752
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3865475500510294e-05,
+      "loss": 0.4919,
+      "step": 12753
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3864556257003714e-05,
+      "loss": 0.3841,
+      "step": 12754
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3863636975106354e-05,
+      "loss": 0.3471,
+      "step": 12755
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3862717654827345e-05,
+      "loss": 0.4035,
+      "step": 12756
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3861798296175815e-05,
+      "loss": 0.3465,
+      "step": 12757
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3860878899160903e-05,
+      "loss": 0.4198,
+      "step": 12758
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3859959463791739e-05,
+      "loss": 0.344,
+      "step": 12759
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3859039990077458e-05,
+      "loss": 0.3759,
+      "step": 12760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3858120478027193e-05,
+      "loss": 0.3789,
+      "step": 12761
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.385720092765008e-05,
+      "loss": 0.3846,
+      "step": 12762
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3856281338955253e-05,
+      "loss": 0.4041,
+      "step": 12763
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3855361711951847e-05,
+      "loss": 0.3669,
+      "step": 12764
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3854442046649002e-05,
+      "loss": 0.3659,
+      "step": 12765
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3853522343055842e-05,
+      "loss": 0.3656,
+      "step": 12766
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3852602601181518e-05,
+      "loss": 0.5282,
+      "step": 12767
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3851682821035159e-05,
+      "loss": 0.3989,
+      "step": 12768
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3850763002625902e-05,
+      "loss": 0.3624,
+      "step": 12769
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3849843145962888e-05,
+      "loss": 0.4167,
+      "step": 12770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3848923251055251e-05,
+      "loss": 0.3896,
+      "step": 12771
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3848003317912131e-05,
+      "loss": 0.3683,
+      "step": 12772
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3847083346542666e-05,
+      "loss": 0.2978,
+      "step": 12773
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3846163336956e-05,
+      "loss": 0.3614,
+      "step": 12774
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3845243289161263e-05,
+      "loss": 0.3825,
+      "step": 12775
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3844323203167605e-05,
+      "loss": 0.4194,
+      "step": 12776
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384340307898416e-05,
+      "loss": 0.3891,
+      "step": 12777
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384248291662007e-05,
+      "loss": 0.2973,
+      "step": 12778
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3841562716084473e-05,
+      "loss": 0.3318,
+      "step": 12779
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384064247738652e-05,
+      "loss": 0.3785,
+      "step": 12780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3839722200535339e-05,
+      "loss": 0.3922,
+      "step": 12781
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3838801885540081e-05,
+      "loss": 0.4191,
+      "step": 12782
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3837881532409889e-05,
+      "loss": 0.4273,
+      "step": 12783
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.38369611411539e-05,
+      "loss": 0.3953,
+      "step": 12784
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3836040711781263e-05,
+      "loss": 0.3795,
+      "step": 12785
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.383512024430112e-05,
+      "loss": 0.3712,
+      "step": 12786
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3834199738722613e-05,
+      "loss": 0.3234,
+      "step": 12787
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3833279195054885e-05,
+      "loss": 0.3712,
+      "step": 12788
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3832358613307087e-05,
+      "loss": 0.385,
+      "step": 12789
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.383143799348836e-05,
+      "loss": 0.2941,
+      "step": 12790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3830517335607848e-05,
+      "loss": 0.3457,
+      "step": 12791
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.38295966396747e-05,
+      "loss": 0.3862,
+      "step": 12792
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3828675905698061e-05,
+      "loss": 0.4205,
+      "step": 12793
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.382775513368708e-05,
+      "loss": 0.4301,
+      "step": 12794
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.3739,
+      "step": 12795
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.382591347559867e-05,
+      "loss": 0.3689,
+      "step": 12796
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3824992589539536e-05,
+      "loss": 0.3517,
+      "step": 12797
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3824071665482652e-05,
+      "loss": 0.3532,
+      "step": 12798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3823150703437163e-05,
+      "loss": 0.3714,
+      "step": 12799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3822229703412213e-05,
+      "loss": 0.371,
+      "step": 12800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3821308665416957e-05,
+      "loss": 0.3996,
+      "step": 12801
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3820387589460547e-05,
+      "loss": 0.4031,
+      "step": 12802
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3819466475552126e-05,
+      "loss": 0.3538,
+      "step": 12803
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3818545323700849e-05,
+      "loss": 0.3996,
+      "step": 12804
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3817624133915865e-05,
+      "loss": 0.3553,
+      "step": 12805
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3816702906206327e-05,
+      "loss": 0.4188,
+      "step": 12806
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3815781640581388e-05,
+      "loss": 0.4226,
+      "step": 12807
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3814860337050193e-05,
+      "loss": 0.4195,
+      "step": 12808
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.38139389956219e-05,
+      "loss": 0.3767,
+      "step": 12809
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381301761630566e-05,
+      "loss": 0.3668,
+      "step": 12810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3812096199110629e-05,
+      "loss": 0.3236,
+      "step": 12811
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3811174744045955e-05,
+      "loss": 0.4168,
+      "step": 12812
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3810253251120793e-05,
+      "loss": 0.3689,
+      "step": 12813
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3809331720344303e-05,
+      "loss": 0.366,
+      "step": 12814
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3808410151725633e-05,
+      "loss": 0.3383,
+      "step": 12815
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3807488545273941e-05,
+      "loss": 0.3482,
+      "step": 12816
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3806566900998382e-05,
+      "loss": 0.3701,
+      "step": 12817
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3805645218908108e-05,
+      "loss": 0.3456,
+      "step": 12818
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3804723499012283e-05,
+      "loss": 0.438,
+      "step": 12819
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3803801741320055e-05,
+      "loss": 0.4208,
+      "step": 12820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3802879945840586e-05,
+      "loss": 0.3935,
+      "step": 12821
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3801958112583031e-05,
+      "loss": 0.4126,
+      "step": 12822
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3801036241556552e-05,
+      "loss": 0.4434,
+      "step": 12823
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.38001143327703e-05,
+      "loss": 0.345,
+      "step": 12824
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3799192386233434e-05,
+      "loss": 0.45,
+      "step": 12825
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3798270401955119e-05,
+      "loss": 0.4019,
+      "step": 12826
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3797348379944508e-05,
+      "loss": 0.3483,
+      "step": 12827
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3796426320210765e-05,
+      "loss": 0.4011,
+      "step": 12828
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3795504222763046e-05,
+      "loss": 0.37,
+      "step": 12829
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3794582087610513e-05,
+      "loss": 0.4003,
+      "step": 12830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3793659914762324e-05,
+      "loss": 0.3722,
+      "step": 12831
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3792737704227645e-05,
+      "loss": 0.4104,
+      "step": 12832
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3791815456015635e-05,
+      "loss": 0.3365,
+      "step": 12833
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3790893170135452e-05,
+      "loss": 0.3268,
+      "step": 12834
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3789970846596268e-05,
+      "loss": 0.3074,
+      "step": 12835
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3789048485407232e-05,
+      "loss": 0.4085,
+      "step": 12836
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3788126086577513e-05,
+      "loss": 0.3797,
+      "step": 12837
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3787203650116282e-05,
+      "loss": 0.3608,
+      "step": 12838
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3786281176032688e-05,
+      "loss": 0.3821,
+      "step": 12839
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3785358664335905e-05,
+      "loss": 0.3641,
+      "step": 12840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3784436115035092e-05,
+      "loss": 0.4846,
+      "step": 12841
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3783513528139418e-05,
+      "loss": 0.3795,
+      "step": 12842
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3782590903658047e-05,
+      "loss": 0.4861,
+      "step": 12843
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.378166824160014e-05,
+      "loss": 0.3281,
+      "step": 12844
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3780745541974871e-05,
+      "loss": 0.3858,
+      "step": 12845
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3779822804791399e-05,
+      "loss": 0.3716,
+      "step": 12846
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3778900030058895e-05,
+      "loss": 0.3563,
+      "step": 12847
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3777977217786522e-05,
+      "loss": 0.3822,
+      "step": 12848
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3777054367983448e-05,
+      "loss": 0.3558,
+      "step": 12849
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3776131480658843e-05,
+      "loss": 0.3252,
+      "step": 12850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3775208555821875e-05,
+      "loss": 0.4381,
+      "step": 12851
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.377428559348171e-05,
+      "loss": 0.3876,
+      "step": 12852
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3773362593647519e-05,
+      "loss": 0.4009,
+      "step": 12853
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.377243955632847e-05,
+      "loss": 0.3604,
+      "step": 12854
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3771516481533733e-05,
+      "loss": 0.395,
+      "step": 12855
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3770593369272476e-05,
+      "loss": 0.3925,
+      "step": 12856
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3769670219553875e-05,
+      "loss": 0.329,
+      "step": 12857
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3768747032387094e-05,
+      "loss": 0.3567,
+      "step": 12858
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3767823807781303e-05,
+      "loss": 0.3513,
+      "step": 12859
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3766900545745681e-05,
+      "loss": 0.4229,
+      "step": 12860
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3765977246289395e-05,
+      "loss": 0.3761,
+      "step": 12861
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3765053909421618e-05,
+      "loss": 0.3799,
+      "step": 12862
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3764130535151523e-05,
+      "loss": 0.3717,
+      "step": 12863
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3763207123488281e-05,
+      "loss": 0.3805,
+      "step": 12864
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3762283674441066e-05,
+      "loss": 0.4205,
+      "step": 12865
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3761360188019053e-05,
+      "loss": 0.4177,
+      "step": 12866
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3760436664231416e-05,
+      "loss": 0.29,
+      "step": 12867
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3759513103087325e-05,
+      "loss": 0.4544,
+      "step": 12868
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.375858950459596e-05,
+      "loss": 0.3963,
+      "step": 12869
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3757665868766496e-05,
+      "loss": 0.3775,
+      "step": 12870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3756742195608102e-05,
+      "loss": 0.4136,
+      "step": 12871
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3755818485129964e-05,
+      "loss": 0.4259,
+      "step": 12872
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3754894737341248e-05,
+      "loss": 0.4262,
+      "step": 12873
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3753970952251138e-05,
+      "loss": 0.3748,
+      "step": 12874
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3753047129868808e-05,
+      "loss": 0.4164,
+      "step": 12875
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3752123270203437e-05,
+      "loss": 0.3896,
+      "step": 12876
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3751199373264196e-05,
+      "loss": 0.447,
+      "step": 12877
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3750275439060273e-05,
+      "loss": 0.403,
+      "step": 12878
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3749351467600842e-05,
+      "loss": 0.3701,
+      "step": 12879
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3748427458895077e-05,
+      "loss": 0.4581,
+      "step": 12880
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3747503412952169e-05,
+      "loss": 0.4288,
+      "step": 12881
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3746579329781283e-05,
+      "loss": 0.327,
+      "step": 12882
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3745655209391606e-05,
+      "loss": 0.3216,
+      "step": 12883
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3744731051792323e-05,
+      "loss": 0.3978,
+      "step": 12884
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.374380685699261e-05,
+      "loss": 0.3946,
+      "step": 12885
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3742882625001645e-05,
+      "loss": 0.3224,
+      "step": 12886
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3741958355828616e-05,
+      "loss": 0.3986,
+      "step": 12887
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.37410340494827e-05,
+      "loss": 0.3709,
+      "step": 12888
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3740109705973076e-05,
+      "loss": 0.326,
+      "step": 12889
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3739185325308935e-05,
+      "loss": 0.3819,
+      "step": 12890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3738260907499457e-05,
+      "loss": 0.3214,
+      "step": 12891
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3737336452553818e-05,
+      "loss": 0.3264,
+      "step": 12892
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3736411960481213e-05,
+      "loss": 0.3192,
+      "step": 12893
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3735487431290818e-05,
+      "loss": 0.3951,
+      "step": 12894
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3734562864991823e-05,
+      "loss": 0.3719,
+      "step": 12895
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3733638261593406e-05,
+      "loss": 0.3964,
+      "step": 12896
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3732713621104756e-05,
+      "loss": 0.3493,
+      "step": 12897
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3731788943535059e-05,
+      "loss": 0.449,
+      "step": 12898
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.37308642288935e-05,
+      "loss": 0.3566,
+      "step": 12899
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3729939477189264e-05,
+      "loss": 0.3693,
+      "step": 12900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3729014688431539e-05,
+      "loss": 0.3866,
+      "step": 12901
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3728089862629512e-05,
+      "loss": 0.3606,
+      "step": 12902
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3727164999792372e-05,
+      "loss": 0.319,
+      "step": 12903
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3726240099929302e-05,
+      "loss": 0.3704,
+      "step": 12904
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3725315163049495e-05,
+      "loss": 0.3789,
+      "step": 12905
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3724390189162133e-05,
+      "loss": 0.4131,
+      "step": 12906
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3723465178276411e-05,
+      "loss": 0.46,
+      "step": 12907
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3722540130401516e-05,
+      "loss": 0.4216,
+      "step": 12908
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3721615045546634e-05,
+      "loss": 0.3412,
+      "step": 12909
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3720689923720965e-05,
+      "loss": 0.4246,
+      "step": 12910
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3719764764933684e-05,
+      "loss": 0.397,
+      "step": 12911
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3718839569193994e-05,
+      "loss": 0.4393,
+      "step": 12912
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3717914336511085e-05,
+      "loss": 0.3914,
+      "step": 12913
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.371698906689414e-05,
+      "loss": 0.3369,
+      "step": 12914
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3716063760352359e-05,
+      "loss": 0.4031,
+      "step": 12915
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3715138416894932e-05,
+      "loss": 0.3792,
+      "step": 12916
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3714213036531048e-05,
+      "loss": 0.38,
+      "step": 12917
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3713287619269902e-05,
+      "loss": 0.4169,
+      "step": 12918
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3712362165120688e-05,
+      "loss": 0.3598,
+      "step": 12919
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3711436674092598e-05,
+      "loss": 0.3454,
+      "step": 12920
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3710511146194828e-05,
+      "loss": 0.3805,
+      "step": 12921
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3709585581436574e-05,
+      "loss": 0.4141,
+      "step": 12922
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3708659979827021e-05,
+      "loss": 0.3314,
+      "step": 12923
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3707734341375375e-05,
+      "loss": 0.3492,
+      "step": 12924
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3706808666090827e-05,
+      "loss": 0.4059,
+      "step": 12925
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370588295398257e-05,
+      "loss": 0.365,
+      "step": 12926
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3704957205059808e-05,
+      "loss": 0.4093,
+      "step": 12927
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3704031419331729e-05,
+      "loss": 0.3517,
+      "step": 12928
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3703105596807534e-05,
+      "loss": 0.3091,
+      "step": 12929
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3702179737496417e-05,
+      "loss": 0.3744,
+      "step": 12930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370125384140758e-05,
+      "loss": 0.4179,
+      "step": 12931
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370032790855022e-05,
+      "loss": 0.3773,
+      "step": 12932
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3699401938933532e-05,
+      "loss": 0.3491,
+      "step": 12933
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369847593256672e-05,
+      "loss": 0.4152,
+      "step": 12934
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3697549889458974e-05,
+      "loss": 0.3237,
+      "step": 12935
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3696623809619502e-05,
+      "loss": 0.3978,
+      "step": 12936
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3695697693057504e-05,
+      "loss": 0.382,
+      "step": 12937
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3694771539782173e-05,
+      "loss": 0.33,
+      "step": 12938
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3693845349802714e-05,
+      "loss": 0.4211,
+      "step": 12939
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3692919123128327e-05,
+      "loss": 0.4866,
+      "step": 12940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3691992859768215e-05,
+      "loss": 0.3209,
+      "step": 12941
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3691066559731578e-05,
+      "loss": 0.4362,
+      "step": 12942
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3690140223027618e-05,
+      "loss": 0.3314,
+      "step": 12943
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3689213849665537e-05,
+      "loss": 0.331,
+      "step": 12944
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3688287439654537e-05,
+      "loss": 0.3735,
+      "step": 12945
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3687360993003825e-05,
+      "loss": 0.3729,
+      "step": 12946
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.36864345097226e-05,
+      "loss": 0.3629,
+      "step": 12947
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3685507989820063e-05,
+      "loss": 0.3697,
+      "step": 12948
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3684581433305426e-05,
+      "loss": 0.4307,
+      "step": 12949
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368365484018789e-05,
+      "loss": 0.4171,
+      "step": 12950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368272821047666e-05,
+      "loss": 0.3935,
+      "step": 12951
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368180154418094e-05,
+      "loss": 0.3975,
+      "step": 12952
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3680874841309934e-05,
+      "loss": 0.3707,
+      "step": 12953
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3679948101872853e-05,
+      "loss": 0.3763,
+      "step": 12954
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3679021325878901e-05,
+      "loss": 0.393,
+      "step": 12955
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3678094513337283e-05,
+      "loss": 0.3361,
+      "step": 12956
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3677167664257205e-05,
+      "loss": 0.3348,
+      "step": 12957
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3676240778647882e-05,
+      "loss": 0.3847,
+      "step": 12958
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3675313856518513e-05,
+      "loss": 0.3632,
+      "step": 12959
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3674386897878306e-05,
+      "loss": 0.3674,
+      "step": 12960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3673459902736478e-05,
+      "loss": 0.2513,
+      "step": 12961
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3672532871102233e-05,
+      "loss": 0.3479,
+      "step": 12962
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.367160580298478e-05,
+      "loss": 0.3954,
+      "step": 12963
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3670678698393326e-05,
+      "loss": 0.3495,
+      "step": 12964
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3669751557337085e-05,
+      "loss": 0.3642,
+      "step": 12965
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3668824379825265e-05,
+      "loss": 0.3475,
+      "step": 12966
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.366789716586708e-05,
+      "loss": 0.4053,
+      "step": 12967
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3666969915471737e-05,
+      "loss": 0.3776,
+      "step": 12968
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3666042628648445e-05,
+      "loss": 0.373,
+      "step": 12969
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3665115305406423e-05,
+      "loss": 0.3797,
+      "step": 12970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3664187945754881e-05,
+      "loss": 0.3668,
+      "step": 12971
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3663260549703026e-05,
+      "loss": 0.3874,
+      "step": 12972
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.366233311726008e-05,
+      "loss": 0.4131,
+      "step": 12973
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3661405648435249e-05,
+      "loss": 0.4772,
+      "step": 12974
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3660478143237748e-05,
+      "loss": 0.3882,
+      "step": 12975
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3659550601676792e-05,
+      "loss": 0.343,
+      "step": 12976
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3658623023761595e-05,
+      "loss": 0.4106,
+      "step": 12977
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3657695409501371e-05,
+      "loss": 0.4163,
+      "step": 12978
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3656767758905335e-05,
+      "loss": 0.4607,
+      "step": 12979
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3655840071982705e-05,
+      "loss": 0.3575,
+      "step": 12980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.365491234874269e-05,
+      "loss": 0.3459,
+      "step": 12981
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3653984589194513e-05,
+      "loss": 0.429,
+      "step": 12982
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.365305679334739e-05,
+      "loss": 0.3758,
+      "step": 12983
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3652128961210536e-05,
+      "loss": 0.4278,
+      "step": 12984
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3651201092793164e-05,
+      "loss": 0.3609,
+      "step": 12985
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3650273188104498e-05,
+      "loss": 0.4558,
+      "step": 12986
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3649345247153753e-05,
+      "loss": 0.4697,
+      "step": 12987
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3648417269950147e-05,
+      "loss": 0.2753,
+      "step": 12988
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3647489256502899e-05,
+      "loss": 0.3672,
+      "step": 12989
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3646561206821231e-05,
+      "loss": 0.3301,
+      "step": 12990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3645633120914355e-05,
+      "loss": 0.3349,
+      "step": 12991
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3644704998791501e-05,
+      "loss": 0.3862,
+      "step": 12992
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3643776840461878e-05,
+      "loss": 0.4185,
+      "step": 12993
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.364284864593471e-05,
+      "loss": 0.4244,
+      "step": 12994
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3641920415219224e-05,
+      "loss": 0.454,
+      "step": 12995
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3640992148324634e-05,
+      "loss": 0.3962,
+      "step": 12996
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3640063845260164e-05,
+      "loss": 0.4019,
+      "step": 12997
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3639135506035035e-05,
+      "loss": 0.3957,
+      "step": 12998
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3638207130658471e-05,
+      "loss": 0.3149,
+      "step": 12999
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3637278719139695e-05,
+      "loss": 0.2608,
+      "step": 13000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3636350271487926e-05,
+      "loss": 0.2448,
+      "step": 13001
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3635421787712391e-05,
+      "loss": 0.2837,
+      "step": 13002
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3634493267822308e-05,
+      "loss": 0.2712,
+      "step": 13003
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3633564711826911e-05,
+      "loss": 0.2127,
+      "step": 13004
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3632636119735416e-05,
+      "loss": 0.2341,
+      "step": 13005
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3631707491557049e-05,
+      "loss": 0.2296,
+      "step": 13006
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3630778827301038e-05,
+      "loss": 0.2116,
+      "step": 13007
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3629850126976606e-05,
+      "loss": 0.2293,
+      "step": 13008
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3628921390592982e-05,
+      "loss": 0.2547,
+      "step": 13009
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3627992618159387e-05,
+      "loss": 0.2228,
+      "step": 13010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3627063809685051e-05,
+      "loss": 0.2139,
+      "step": 13011
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3626134965179198e-05,
+      "loss": 0.2675,
+      "step": 13012
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362520608465106e-05,
+      "loss": 0.2407,
+      "step": 13013
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3624277168109861e-05,
+      "loss": 0.2529,
+      "step": 13014
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3623348215564827e-05,
+      "loss": 0.2567,
+      "step": 13015
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3622419227025195e-05,
+      "loss": 0.2322,
+      "step": 13016
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3621490202500183e-05,
+      "loss": 0.2089,
+      "step": 13017
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3620561141999025e-05,
+      "loss": 0.2109,
+      "step": 13018
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3619632045530948e-05,
+      "loss": 0.2532,
+      "step": 13019
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3618702913105186e-05,
+      "loss": 0.2812,
+      "step": 13020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3617773744730967e-05,
+      "loss": 0.2706,
+      "step": 13021
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.361684454041752e-05,
+      "loss": 0.2232,
+      "step": 13022
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.361591530017408e-05,
+      "loss": 0.2676,
+      "step": 13023
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3614986024009872e-05,
+      "loss": 0.2428,
+      "step": 13024
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.361405671193413e-05,
+      "loss": 0.2703,
+      "step": 13025
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3613127363956089e-05,
+      "loss": 0.2502,
+      "step": 13026
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3612197980084977e-05,
+      "loss": 0.2482,
+      "step": 13027
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3611268560330028e-05,
+      "loss": 0.2246,
+      "step": 13028
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3610339104700476e-05,
+      "loss": 0.2127,
+      "step": 13029
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3609409613205553e-05,
+      "loss": 0.2495,
+      "step": 13030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3608480085854494e-05,
+      "loss": 0.2256,
+      "step": 13031
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.360755052265653e-05,
+      "loss": 0.2292,
+      "step": 13032
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.36066209236209e-05,
+      "loss": 0.301,
+      "step": 13033
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3605691288756833e-05,
+      "loss": 0.2446,
+      "step": 13034
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3604761618073567e-05,
+      "loss": 0.2603,
+      "step": 13035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3603831911580342e-05,
+      "loss": 0.2269,
+      "step": 13036
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3602902169286386e-05,
+      "loss": 0.2155,
+      "step": 13037
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.360197239120094e-05,
+      "loss": 0.2283,
+      "step": 13038
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.360104257733324e-05,
+      "loss": 0.2565,
+      "step": 13039
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3600112727692515e-05,
+      "loss": 0.2238,
+      "step": 13040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3599182842288017e-05,
+      "loss": 0.2574,
+      "step": 13041
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3598252921128972e-05,
+      "loss": 0.2057,
+      "step": 13042
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3597322964224622e-05,
+      "loss": 0.2703,
+      "step": 13043
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3596392971584205e-05,
+      "loss": 0.2026,
+      "step": 13044
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3595462943216959e-05,
+      "loss": 0.269,
+      "step": 13045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.359453287913212e-05,
+      "loss": 0.1959,
+      "step": 13046
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3593602779338935e-05,
+      "loss": 0.2399,
+      "step": 13047
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3592672643846641e-05,
+      "loss": 0.2621,
+      "step": 13048
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3591742472664468e-05,
+      "loss": 0.266,
+      "step": 13049
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3590812265801675e-05,
+      "loss": 0.2544,
+      "step": 13050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3589882023267484e-05,
+      "loss": 0.17,
+      "step": 13051
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3588951745071145e-05,
+      "loss": 0.2189,
+      "step": 13052
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3588021431221905e-05,
+      "loss": 0.2321,
+      "step": 13053
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3587091081728997e-05,
+      "loss": 0.2112,
+      "step": 13054
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3586160696601667e-05,
+      "loss": 0.1835,
+      "step": 13055
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3585230275849153e-05,
+      "loss": 0.2183,
+      "step": 13056
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3584299819480706e-05,
+      "loss": 0.2362,
+      "step": 13057
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.358336932750556e-05,
+      "loss": 0.239,
+      "step": 13058
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3582438799932965e-05,
+      "loss": 0.2176,
+      "step": 13059
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3581508236772166e-05,
+      "loss": 0.2541,
+      "step": 13060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.35805776380324e-05,
+      "loss": 0.2497,
+      "step": 13061
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3579647003722918e-05,
+      "loss": 0.2136,
+      "step": 13062
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3578716333852962e-05,
+      "loss": 0.2876,
+      "step": 13063
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3577785628431778e-05,
+      "loss": 0.2693,
+      "step": 13064
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3576854887468614e-05,
+      "loss": 0.2263,
+      "step": 13065
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3575924110972711e-05,
+      "loss": 0.2326,
+      "step": 13066
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.357499329895332e-05,
+      "loss": 0.252,
+      "step": 13067
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3574062451419687e-05,
+      "loss": 0.2179,
+      "step": 13068
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3573131568381058e-05,
+      "loss": 0.3031,
+      "step": 13069
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3572200649846683e-05,
+      "loss": 0.2618,
+      "step": 13070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3571269695825801e-05,
+      "loss": 0.2324,
+      "step": 13071
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3570338706327673e-05,
+      "loss": 0.2062,
+      "step": 13072
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356940768136154e-05,
+      "loss": 0.2774,
+      "step": 13073
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356847662093665e-05,
+      "loss": 0.2595,
+      "step": 13074
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3567545525062256e-05,
+      "loss": 0.2565,
+      "step": 13075
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3566614393747606e-05,
+      "loss": 0.2462,
+      "step": 13076
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356568322700195e-05,
+      "loss": 0.301,
+      "step": 13077
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3564752024834535e-05,
+      "loss": 0.2616,
+      "step": 13078
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356382078725462e-05,
+      "loss": 0.2164,
+      "step": 13079
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3562889514271444e-05,
+      "loss": 0.2724,
+      "step": 13080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356195820589427e-05,
+      "loss": 0.2384,
+      "step": 13081
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3561026862132347e-05,
+      "loss": 0.1863,
+      "step": 13082
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356009548299492e-05,
+      "loss": 0.2174,
+      "step": 13083
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3559164068491249e-05,
+      "loss": 0.2456,
+      "step": 13084
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3558232618630582e-05,
+      "loss": 0.2052,
+      "step": 13085
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3557301133422174e-05,
+      "loss": 0.3009,
+      "step": 13086
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355636961287528e-05,
+      "loss": 0.207,
+      "step": 13087
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355543805699915e-05,
+      "loss": 0.2221,
+      "step": 13088
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3554506465803044e-05,
+      "loss": 0.2453,
+      "step": 13089
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3553574839296208e-05,
+      "loss": 0.2378,
+      "step": 13090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3552643177487908e-05,
+      "loss": 0.2395,
+      "step": 13091
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3551711480387387e-05,
+      "loss": 0.2858,
+      "step": 13092
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3550779748003908e-05,
+      "loss": 0.2569,
+      "step": 13093
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3549847980346728e-05,
+      "loss": 0.2333,
+      "step": 13094
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3548916177425095e-05,
+      "loss": 0.2296,
+      "step": 13095
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3547984339248279e-05,
+      "loss": 0.236,
+      "step": 13096
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3547052465825525e-05,
+      "loss": 0.2604,
+      "step": 13097
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3546120557166094e-05,
+      "loss": 0.2176,
+      "step": 13098
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3545188613279246e-05,
+      "loss": 0.2654,
+      "step": 13099
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3544256634174235e-05,
+      "loss": 0.2449,
+      "step": 13100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3543324619860324e-05,
+      "loss": 0.1931,
+      "step": 13101
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3542392570346764e-05,
+      "loss": 0.2627,
+      "step": 13102
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3541460485642825e-05,
+      "loss": 0.252,
+      "step": 13103
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3540528365757756e-05,
+      "loss": 0.252,
+      "step": 13104
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3539596210700825e-05,
+      "loss": 0.2281,
+      "step": 13105
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3538664020481288e-05,
+      "loss": 0.2417,
+      "step": 13106
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3537731795108407e-05,
+      "loss": 0.2419,
+      "step": 13107
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3536799534591439e-05,
+      "loss": 0.2377,
+      "step": 13108
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3535867238939648e-05,
+      "loss": 0.2889,
+      "step": 13109
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3534934908162296e-05,
+      "loss": 0.241,
+      "step": 13110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3534002542268643e-05,
+      "loss": 0.2359,
+      "step": 13111
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3533070141267954e-05,
+      "loss": 0.3025,
+      "step": 13112
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3532137705169487e-05,
+      "loss": 0.2059,
+      "step": 13113
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3531205233982512e-05,
+      "loss": 0.3034,
+      "step": 13114
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3530272727716286e-05,
+      "loss": 0.2465,
+      "step": 13115
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3529340186380075e-05,
+      "loss": 0.2501,
+      "step": 13116
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352840760998314e-05,
+      "loss": 0.2434,
+      "step": 13117
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3527474998534754e-05,
+      "loss": 0.2749,
+      "step": 13118
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3526542352044172e-05,
+      "loss": 0.2314,
+      "step": 13119
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352560967052066e-05,
+      "loss": 0.2705,
+      "step": 13120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3524676953973488e-05,
+      "loss": 0.2454,
+      "step": 13121
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3523744202411919e-05,
+      "loss": 0.2508,
+      "step": 13122
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3522811415845219e-05,
+      "loss": 0.2701,
+      "step": 13123
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3521878594282656e-05,
+      "loss": 0.23,
+      "step": 13124
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3520945737733496e-05,
+      "loss": 0.2311,
+      "step": 13125
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3520012846207003e-05,
+      "loss": 0.2334,
+      "step": 13126
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351907991971245e-05,
+      "loss": 0.2494,
+      "step": 13127
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.35181469582591e-05,
+      "loss": 0.1939,
+      "step": 13128
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351721396185622e-05,
+      "loss": 0.2183,
+      "step": 13129
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3516280930513087e-05,
+      "loss": 0.2229,
+      "step": 13130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351534786423896e-05,
+      "loss": 0.2983,
+      "step": 13131
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3514414763043116e-05,
+      "loss": 0.2523,
+      "step": 13132
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3513481626934817e-05,
+      "loss": 0.2124,
+      "step": 13133
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351254845592334e-05,
+      "loss": 0.2212,
+      "step": 13134
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3511615250017948e-05,
+      "loss": 0.2437,
+      "step": 13135
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3510682009227917e-05,
+      "loss": 0.2383,
+      "step": 13136
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.350974873356252e-05,
+      "loss": 0.1612,
+      "step": 13137
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3508815423031018e-05,
+      "loss": 0.2991,
+      "step": 13138
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3507882077642691e-05,
+      "loss": 0.3062,
+      "step": 13139
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3506948697406812e-05,
+      "loss": 0.2271,
+      "step": 13140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3506015282332648e-05,
+      "loss": 0.1805,
+      "step": 13141
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3505081832429476e-05,
+      "loss": 0.2824,
+      "step": 13142
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3504148347706565e-05,
+      "loss": 0.2444,
+      "step": 13143
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3503214828173187e-05,
+      "loss": 0.2578,
+      "step": 13144
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3502281273838622e-05,
+      "loss": 0.2285,
+      "step": 13145
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.350134768471214e-05,
+      "loss": 0.2406,
+      "step": 13146
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3500414060803017e-05,
+      "loss": 0.2464,
+      "step": 13147
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3499480402120527e-05,
+      "loss": 0.2462,
+      "step": 13148
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3498546708673946e-05,
+      "loss": 0.3016,
+      "step": 13149
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3497612980472543e-05,
+      "loss": 0.2266,
+      "step": 13150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3496679217525603e-05,
+      "loss": 0.2238,
+      "step": 13151
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3495745419842399e-05,
+      "loss": 0.2243,
+      "step": 13152
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3494811587432204e-05,
+      "loss": 0.2484,
+      "step": 13153
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3493877720304296e-05,
+      "loss": 0.2124,
+      "step": 13154
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3492943818467956e-05,
+      "loss": 0.2703,
+      "step": 13155
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3492009881932458e-05,
+      "loss": 0.2652,
+      "step": 13156
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3491075910707079e-05,
+      "loss": 0.1799,
+      "step": 13157
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.34901419048011e-05,
+      "loss": 0.2531,
+      "step": 13158
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3489207864223798e-05,
+      "loss": 0.2514,
+      "step": 13159
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3488273788984452e-05,
+      "loss": 0.2453,
+      "step": 13160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3487339679092342e-05,
+      "loss": 0.2532,
+      "step": 13161
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3486405534556744e-05,
+      "loss": 0.2558,
+      "step": 13162
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3485471355386941e-05,
+      "loss": 0.2513,
+      "step": 13163
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3484537141592215e-05,
+      "loss": 0.2567,
+      "step": 13164
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3483602893181841e-05,
+      "loss": 0.2269,
+      "step": 13165
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3482668610165106e-05,
+      "loss": 0.2258,
+      "step": 13166
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3481734292551287e-05,
+      "loss": 0.2179,
+      "step": 13167
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.348079994034967e-05,
+      "loss": 0.2257,
+      "step": 13168
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3479865553569528e-05,
+      "loss": 0.2579,
+      "step": 13169
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3478931132220153e-05,
+      "loss": 0.2255,
+      "step": 13170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3477996676310825e-05,
+      "loss": 0.2519,
+      "step": 13171
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3477062185850822e-05,
+      "loss": 0.2566,
+      "step": 13172
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3476127660849434e-05,
+      "loss": 0.2468,
+      "step": 13173
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.347519310131594e-05,
+      "loss": 0.2155,
+      "step": 13174
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3474258507259623e-05,
+      "loss": 0.2634,
+      "step": 13175
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3473323878689775e-05,
+      "loss": 0.2666,
+      "step": 13176
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3472389215615674e-05,
+      "loss": 0.2645,
+      "step": 13177
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3471454518046605e-05,
+      "loss": 0.2506,
+      "step": 13178
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3470519785991856e-05,
+      "loss": 0.2236,
+      "step": 13179
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3469585019460712e-05,
+      "loss": 0.1823,
+      "step": 13180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3468650218462457e-05,
+      "loss": 0.2772,
+      "step": 13181
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3467715383006382e-05,
+      "loss": 0.2684,
+      "step": 13182
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.346678051310177e-05,
+      "loss": 0.1919,
+      "step": 13183
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3465845608757906e-05,
+      "loss": 0.2613,
+      "step": 13184
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3464910669984084e-05,
+      "loss": 0.2254,
+      "step": 13185
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3463975696789586e-05,
+      "loss": 0.21,
+      "step": 13186
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3463040689183701e-05,
+      "loss": 0.2122,
+      "step": 13187
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3462105647175721e-05,
+      "loss": 0.2551,
+      "step": 13188
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.346117057077493e-05,
+      "loss": 0.2038,
+      "step": 13189
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3460235459990623e-05,
+      "loss": 0.2546,
+      "step": 13190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3459300314832082e-05,
+      "loss": 0.2398,
+      "step": 13191
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3458365135308601e-05,
+      "loss": 0.2401,
+      "step": 13192
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3457429921429473e-05,
+      "loss": 0.2268,
+      "step": 13193
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3456494673203983e-05,
+      "loss": 0.2561,
+      "step": 13194
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3455559390641426e-05,
+      "loss": 0.2803,
+      "step": 13195
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.345462407375109e-05,
+      "loss": 0.2644,
+      "step": 13196
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3453688722542269e-05,
+      "loss": 0.2679,
+      "step": 13197
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3452753337024251e-05,
+      "loss": 0.2371,
+      "step": 13198
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3451817917206334e-05,
+      "loss": 0.2449,
+      "step": 13199
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3450882463097806e-05,
+      "loss": 0.3053,
+      "step": 13200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3449946974707959e-05,
+      "loss": 0.2783,
+      "step": 13201
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.344901145204609e-05,
+      "loss": 0.1995,
+      "step": 13202
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3448075895121494e-05,
+      "loss": 0.286,
+      "step": 13203
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3447140303943458e-05,
+      "loss": 0.2458,
+      "step": 13204
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3446204678521281e-05,
+      "loss": 0.2384,
+      "step": 13205
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3445269018864257e-05,
+      "loss": 0.1627,
+      "step": 13206
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3444333324981683e-05,
+      "loss": 0.219,
+      "step": 13207
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.344339759688285e-05,
+      "loss": 0.195,
+      "step": 13208
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3442461834577052e-05,
+      "loss": 0.2376,
+      "step": 13209
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3441526038073593e-05,
+      "loss": 0.2381,
+      "step": 13210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3440590207381762e-05,
+      "loss": 0.2195,
+      "step": 13211
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3439654342510859e-05,
+      "loss": 0.2377,
+      "step": 13212
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3438718443470183e-05,
+      "loss": 0.236,
+      "step": 13213
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3437782510269027e-05,
+      "loss": 0.2549,
+      "step": 13214
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3436846542916686e-05,
+      "loss": 0.1975,
+      "step": 13215
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3435910541422466e-05,
+      "loss": 0.2368,
+      "step": 13216
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3434974505795663e-05,
+      "loss": 0.2242,
+      "step": 13217
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.343403843604557e-05,
+      "loss": 0.1838,
+      "step": 13218
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3433102332181494e-05,
+      "loss": 0.2086,
+      "step": 13219
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3432166194212727e-05,
+      "loss": 0.1943,
+      "step": 13220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3431230022148571e-05,
+      "loss": 0.2591,
+      "step": 13221
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3430293815998333e-05,
+      "loss": 0.2598,
+      "step": 13222
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3429357575771303e-05,
+      "loss": 0.2608,
+      "step": 13223
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3428421301476786e-05,
+      "loss": 0.276,
+      "step": 13224
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3427484993124082e-05,
+      "loss": 0.2693,
+      "step": 13225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3426548650722498e-05,
+      "loss": 0.2525,
+      "step": 13226
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3425612274281325e-05,
+      "loss": 0.2438,
+      "step": 13227
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3424675863809875e-05,
+      "loss": 0.2316,
+      "step": 13228
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3423739419317448e-05,
+      "loss": 0.2289,
+      "step": 13229
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3422802940813342e-05,
+      "loss": 0.1487,
+      "step": 13230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3421866428306865e-05,
+      "loss": 0.2636,
+      "step": 13231
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3420929881807319e-05,
+      "loss": 0.2476,
+      "step": 13232
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3419993301324007e-05,
+      "loss": 0.246,
+      "step": 13233
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3419056686866232e-05,
+      "loss": 0.2402,
+      "step": 13234
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.34181200384433e-05,
+      "loss": 0.245,
+      "step": 13235
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3417183356064515e-05,
+      "loss": 0.2682,
+      "step": 13236
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3416246639739183e-05,
+      "loss": 0.2635,
+      "step": 13237
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3415309889476611e-05,
+      "loss": 0.2425,
+      "step": 13238
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3414373105286103e-05,
+      "loss": 0.194,
+      "step": 13239
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3413436287176962e-05,
+      "loss": 0.2606,
+      "step": 13240
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3412499435158499e-05,
+      "loss": 0.2506,
+      "step": 13241
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.341156254924002e-05,
+      "loss": 0.2407,
+      "step": 13242
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3410625629430829e-05,
+      "loss": 0.2131,
+      "step": 13243
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3409688675740236e-05,
+      "loss": 0.2811,
+      "step": 13244
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3408751688177551e-05,
+      "loss": 0.2341,
+      "step": 13245
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3407814666752075e-05,
+      "loss": 0.2678,
+      "step": 13246
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3406877611473124e-05,
+      "loss": 0.2558,
+      "step": 13247
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3405940522350003e-05,
+      "loss": 0.3041,
+      "step": 13248
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3405003399392021e-05,
+      "loss": 0.2709,
+      "step": 13249
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3404066242608486e-05,
+      "loss": 0.2698,
+      "step": 13250
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3403129052008717e-05,
+      "loss": 0.1686,
+      "step": 13251
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3402191827602009e-05,
+      "loss": 0.2411,
+      "step": 13252
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3401254569397688e-05,
+      "loss": 0.2718,
+      "step": 13253
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3400317277405053e-05,
+      "loss": 0.2343,
+      "step": 13254
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.339937995163342e-05,
+      "loss": 0.2958,
+      "step": 13255
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.33984425920921e-05,
+      "loss": 0.2515,
+      "step": 13256
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3397505198790403e-05,
+      "loss": 0.2231,
+      "step": 13257
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3396567771737645e-05,
+      "loss": 0.3236,
+      "step": 13258
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3395630310943135e-05,
+      "loss": 0.198,
+      "step": 13259
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3394692816416193e-05,
+      "loss": 0.2513,
+      "step": 13260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3393755288166117e-05,
+      "loss": 0.2027,
+      "step": 13261
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3392817726202233e-05,
+      "loss": 0.2599,
+      "step": 13262
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3391880130533857e-05,
+      "loss": 0.2496,
+      "step": 13263
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.339094250117029e-05,
+      "loss": 0.2294,
+      "step": 13264
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3390004838120861e-05,
+      "loss": 0.2708,
+      "step": 13265
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3389067141394874e-05,
+      "loss": 0.2305,
+      "step": 13266
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3388129411001647e-05,
+      "loss": 0.2304,
+      "step": 13267
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3387191646950499e-05,
+      "loss": 0.2241,
+      "step": 13268
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3386253849250743e-05,
+      "loss": 0.2197,
+      "step": 13269
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3385316017911697e-05,
+      "loss": 0.2243,
+      "step": 13270
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3384378152942673e-05,
+      "loss": 0.2697,
+      "step": 13271
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3383440254352994e-05,
+      "loss": 0.2685,
+      "step": 13272
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.338250232215197e-05,
+      "loss": 0.206,
+      "step": 13273
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3381564356348923e-05,
+      "loss": 0.2291,
+      "step": 13274
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3380626356953175e-05,
+      "loss": 0.2146,
+      "step": 13275
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3379688323974036e-05,
+      "loss": 0.2176,
+      "step": 13276
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3378750257420829e-05,
+      "loss": 0.1804,
+      "step": 13277
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3377812157302872e-05,
+      "loss": 0.2387,
+      "step": 13278
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3376874023629483e-05,
+      "loss": 0.2301,
+      "step": 13279
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3375935856409982e-05,
+      "loss": 0.2765,
+      "step": 13280
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3374997655653691e-05,
+      "loss": 0.2167,
+      "step": 13281
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3374059421369927e-05,
+      "loss": 0.2242,
+      "step": 13282
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3373121153568013e-05,
+      "loss": 0.1952,
+      "step": 13283
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3372182852257272e-05,
+      "loss": 0.2551,
+      "step": 13284
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3371244517447017e-05,
+      "loss": 0.2349,
+      "step": 13285
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3370306149146572e-05,
+      "loss": 0.23,
+      "step": 13286
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3369367747365268e-05,
+      "loss": 0.2455,
+      "step": 13287
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3368429312112417e-05,
+      "loss": 0.2294,
+      "step": 13288
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3367490843397344e-05,
+      "loss": 0.2327,
+      "step": 13289
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3366552341229375e-05,
+      "loss": 0.2801,
+      "step": 13290
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3365613805617827e-05,
+      "loss": 0.225,
+      "step": 13291
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.336467523657203e-05,
+      "loss": 0.2297,
+      "step": 13292
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3363736634101304e-05,
+      "loss": 0.2746,
+      "step": 13293
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3362797998214978e-05,
+      "loss": 0.2106,
+      "step": 13294
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3361859328922368e-05,
+      "loss": 0.284,
+      "step": 13295
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3360920626232807e-05,
+      "loss": 0.219,
+      "step": 13296
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3359981890155615e-05,
+      "loss": 0.2144,
+      "step": 13297
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3359043120700118e-05,
+      "loss": 0.2118,
+      "step": 13298
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3358104317875644e-05,
+      "loss": 0.2026,
+      "step": 13299
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3357165481691519e-05,
+      "loss": 0.1604,
+      "step": 13300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3356226612157068e-05,
+      "loss": 0.1995,
+      "step": 13301
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3355287709281617e-05,
+      "loss": 0.2499,
+      "step": 13302
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3354348773074497e-05,
+      "loss": 0.263,
+      "step": 13303
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335340980354503e-05,
+      "loss": 0.262,
+      "step": 13304
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3352470800702548e-05,
+      "loss": 0.2723,
+      "step": 13305
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335153176455638e-05,
+      "loss": 0.2436,
+      "step": 13306
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3350592695115847e-05,
+      "loss": 0.2315,
+      "step": 13307
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3349653592390288e-05,
+      "loss": 0.232,
+      "step": 13308
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3348714456389028e-05,
+      "loss": 0.2266,
+      "step": 13309
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.334777528712139e-05,
+      "loss": 0.1779,
+      "step": 13310
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3346836084596715e-05,
+      "loss": 0.2441,
+      "step": 13311
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3345896848824325e-05,
+      "loss": 0.237,
+      "step": 13312
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3344957579813552e-05,
+      "loss": 0.29,
+      "step": 13313
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3344018277573729e-05,
+      "loss": 0.2676,
+      "step": 13314
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3343078942114185e-05,
+      "loss": 0.2474,
+      "step": 13315
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3342139573444254e-05,
+      "loss": 0.3019,
+      "step": 13316
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3341200171573263e-05,
+      "loss": 0.1789,
+      "step": 13317
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.334026073651055e-05,
+      "loss": 0.1927,
+      "step": 13318
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3339321268265443e-05,
+      "loss": 0.2994,
+      "step": 13319
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3338381766847275e-05,
+      "loss": 0.2104,
+      "step": 13320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3337442232265386e-05,
+      "loss": 0.2128,
+      "step": 13321
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3336502664529098e-05,
+      "loss": 0.2315,
+      "step": 13322
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3335563063647752e-05,
+      "loss": 0.2625,
+      "step": 13323
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3334623429630682e-05,
+      "loss": 0.285,
+      "step": 13324
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333368376248722e-05,
+      "loss": 0.2257,
+      "step": 13325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3332744062226702e-05,
+      "loss": 0.2504,
+      "step": 13326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3331804328858459e-05,
+      "loss": 0.2786,
+      "step": 13327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3330864562391836e-05,
+      "loss": 0.2395,
+      "step": 13328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3329924762836158e-05,
+      "loss": 0.2115,
+      "step": 13329
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.332898493020077e-05,
+      "loss": 0.2437,
+      "step": 13330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3328045064495001e-05,
+      "loss": 0.2519,
+      "step": 13331
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.332710516572819e-05,
+      "loss": 0.2986,
+      "step": 13332
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3326165233909679e-05,
+      "loss": 0.2429,
+      "step": 13333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3325225269048798e-05,
+      "loss": 0.2307,
+      "step": 13334
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3324285271154889e-05,
+      "loss": 0.2679,
+      "step": 13335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3323345240237288e-05,
+      "loss": 0.2815,
+      "step": 13336
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3322405176305337e-05,
+      "loss": 0.2497,
+      "step": 13337
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.332146507936837e-05,
+      "loss": 0.2442,
+      "step": 13338
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3320524949435727e-05,
+      "loss": 0.2067,
+      "step": 13339
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3319584786516751e-05,
+      "loss": 0.2868,
+      "step": 13340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3318644590620774e-05,
+      "loss": 0.2544,
+      "step": 13341
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3317704361757147e-05,
+      "loss": 0.2057,
+      "step": 13342
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.33167640999352e-05,
+      "loss": 0.2947,
+      "step": 13343
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3315823805164276e-05,
+      "loss": 0.2671,
+      "step": 13344
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3314883477453725e-05,
+      "loss": 0.2321,
+      "step": 13345
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3313943116812874e-05,
+      "loss": 0.1918,
+      "step": 13346
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3313002723251076e-05,
+      "loss": 0.2355,
+      "step": 13347
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3312062296777666e-05,
+      "loss": 0.224,
+      "step": 13348
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331112183740199e-05,
+      "loss": 0.2726,
+      "step": 13349
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3310181345133388e-05,
+      "loss": 0.2567,
+      "step": 13350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3309240819981206e-05,
+      "loss": 0.2828,
+      "step": 13351
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3308300261954788e-05,
+      "loss": 0.1996,
+      "step": 13352
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3307359671063469e-05,
+      "loss": 0.2838,
+      "step": 13353
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3306419047316606e-05,
+      "loss": 0.2317,
+      "step": 13354
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3305478390723532e-05,
+      "loss": 0.2403,
+      "step": 13355
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3304537701293596e-05,
+      "loss": 0.2411,
+      "step": 13356
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3303596979036144e-05,
+      "loss": 0.288,
+      "step": 13357
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3302656223960519e-05,
+      "loss": 0.2841,
+      "step": 13358
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3301715436076066e-05,
+      "loss": 0.2744,
+      "step": 13359
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3300774615392135e-05,
+      "loss": 0.2169,
+      "step": 13360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3299833761918069e-05,
+      "loss": 0.2295,
+      "step": 13361
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3298892875663214e-05,
+      "loss": 0.2572,
+      "step": 13362
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3297951956636918e-05,
+      "loss": 0.2923,
+      "step": 13363
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.329701100484853e-05,
+      "loss": 0.2714,
+      "step": 13364
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3296070020307394e-05,
+      "loss": 0.2682,
+      "step": 13365
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3295129003022859e-05,
+      "loss": 0.2841,
+      "step": 13366
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3294187953004274e-05,
+      "loss": 0.2296,
+      "step": 13367
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3293246870260985e-05,
+      "loss": 0.2958,
+      "step": 13368
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3292305754802345e-05,
+      "loss": 0.2021,
+      "step": 13369
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.32913646066377e-05,
+      "loss": 0.2599,
+      "step": 13370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.32904234257764e-05,
+      "loss": 0.2396,
+      "step": 13371
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3289482212227796e-05,
+      "loss": 0.2171,
+      "step": 13372
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3288540966001235e-05,
+      "loss": 0.288,
+      "step": 13373
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3287599687106071e-05,
+      "loss": 0.2526,
+      "step": 13374
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3286658375551654e-05,
+      "loss": 0.2321,
+      "step": 13375
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3285717031347335e-05,
+      "loss": 0.2182,
+      "step": 13376
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3284775654502462e-05,
+      "loss": 0.2426,
+      "step": 13377
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3283834245026385e-05,
+      "loss": 0.2689,
+      "step": 13378
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.328289280292847e-05,
+      "loss": 0.2243,
+      "step": 13379
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3281951328218054e-05,
+      "loss": 0.2602,
+      "step": 13380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3281009820904497e-05,
+      "loss": 0.2348,
+      "step": 13381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3280068280997149e-05,
+      "loss": 0.1859,
+      "step": 13382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3279126708505368e-05,
+      "loss": 0.2533,
+      "step": 13383
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.32781851034385e-05,
+      "loss": 0.3029,
+      "step": 13384
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3277243465805905e-05,
+      "loss": 0.2653,
+      "step": 13385
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3276301795616937e-05,
+      "loss": 0.2461,
+      "step": 13386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3275360092880946e-05,
+      "loss": 0.2427,
+      "step": 13387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3274418357607295e-05,
+      "loss": 0.2046,
+      "step": 13388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3273476589805332e-05,
+      "loss": 0.2009,
+      "step": 13389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3272534789484412e-05,
+      "loss": 0.2647,
+      "step": 13390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.32715929566539e-05,
+      "loss": 0.2776,
+      "step": 13391
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3270651091323143e-05,
+      "loss": 0.1754,
+      "step": 13392
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3269709193501501e-05,
+      "loss": 0.249,
+      "step": 13393
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326876726319833e-05,
+      "loss": 0.2455,
+      "step": 13394
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3267825300422991e-05,
+      "loss": 0.2551,
+      "step": 13395
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3266883305184833e-05,
+      "loss": 0.2284,
+      "step": 13396
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3265941277493221e-05,
+      "loss": 0.2343,
+      "step": 13397
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3264999217357514e-05,
+      "loss": 0.2655,
+      "step": 13398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3264057124787069e-05,
+      "loss": 0.2024,
+      "step": 13399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326311499979124e-05,
+      "loss": 0.2319,
+      "step": 13400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326217284237939e-05,
+      "loss": 0.2398,
+      "step": 13401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326123065256088e-05,
+      "loss": 0.203,
+      "step": 13402
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3260288430345068e-05,
+      "loss": 0.2274,
+      "step": 13403
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3259346175741313e-05,
+      "loss": 0.2772,
+      "step": 13404
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3258403888758978e-05,
+      "loss": 0.2421,
+      "step": 13405
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3257461569407419e-05,
+      "loss": 0.1788,
+      "step": 13406
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3256519217696005e-05,
+      "loss": 0.2275,
+      "step": 13407
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3255576833634092e-05,
+      "loss": 0.2301,
+      "step": 13408
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.325463441723104e-05,
+      "loss": 0.1974,
+      "step": 13409
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3253691968496217e-05,
+      "loss": 0.2588,
+      "step": 13410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.325274948743898e-05,
+      "loss": 0.2521,
+      "step": 13411
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3251806974068693e-05,
+      "loss": 0.2256,
+      "step": 13412
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3250864428394721e-05,
+      "loss": 0.2225,
+      "step": 13413
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3249921850426425e-05,
+      "loss": 0.2674,
+      "step": 13414
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3248979240173167e-05,
+      "loss": 0.267,
+      "step": 13415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3248036597644317e-05,
+      "loss": 0.2169,
+      "step": 13416
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3247093922849236e-05,
+      "loss": 0.2474,
+      "step": 13417
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3246151215797288e-05,
+      "loss": 0.2348,
+      "step": 13418
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3245208476497836e-05,
+      "loss": 0.1957,
+      "step": 13419
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.324426570496025e-05,
+      "loss": 0.3019,
+      "step": 13420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3243322901193892e-05,
+      "loss": 0.2622,
+      "step": 13421
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.324238006520813e-05,
+      "loss": 0.2115,
+      "step": 13422
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3241437197012328e-05,
+      "loss": 0.238,
+      "step": 13423
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3240494296615854e-05,
+      "loss": 0.2061,
+      "step": 13424
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3239551364028076e-05,
+      "loss": 0.2436,
+      "step": 13425
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3238608399258357e-05,
+      "loss": 0.2595,
+      "step": 13426
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3237665402316069e-05,
+      "loss": 0.2496,
+      "step": 13427
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3236722373210577e-05,
+      "loss": 0.2308,
+      "step": 13428
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3235779311951254e-05,
+      "loss": 0.2115,
+      "step": 13429
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3234836218547458e-05,
+      "loss": 0.2569,
+      "step": 13430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3233893093008568e-05,
+      "loss": 0.2177,
+      "step": 13431
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.323294993534395e-05,
+      "loss": 0.2516,
+      "step": 13432
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.323200674556297e-05,
+      "loss": 0.2114,
+      "step": 13433
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3231063523675006e-05,
+      "loss": 0.2714,
+      "step": 13434
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3230120269689417e-05,
+      "loss": 0.237,
+      "step": 13435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3229176983615582e-05,
+      "loss": 0.2644,
+      "step": 13436
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3228233665462867e-05,
+      "loss": 0.2009,
+      "step": 13437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3227290315240644e-05,
+      "loss": 0.2484,
+      "step": 13438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3226346932958285e-05,
+      "loss": 0.2826,
+      "step": 13439
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3225403518625162e-05,
+      "loss": 0.2587,
+      "step": 13440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3224460072250649e-05,
+      "loss": 0.2915,
+      "step": 13441
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3223516593844112e-05,
+      "loss": 0.2219,
+      "step": 13442
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.322257308341493e-05,
+      "loss": 0.1671,
+      "step": 13443
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3221629540972472e-05,
+      "loss": 0.2298,
+      "step": 13444
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3220685966526112e-05,
+      "loss": 0.263,
+      "step": 13445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3219742360085224e-05,
+      "loss": 0.2287,
+      "step": 13446
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3218798721659183e-05,
+      "loss": 0.2191,
+      "step": 13447
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.321785505125736e-05,
+      "loss": 0.2132,
+      "step": 13448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3216911348889133e-05,
+      "loss": 0.2209,
+      "step": 13449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3215967614563874e-05,
+      "loss": 0.1899,
+      "step": 13450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.321502384829096e-05,
+      "loss": 0.2848,
+      "step": 13451
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3214080050079767e-05,
+      "loss": 0.2318,
+      "step": 13452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3213136219939671e-05,
+      "loss": 0.2233,
+      "step": 13453
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3212192357880043e-05,
+      "loss": 0.2317,
+      "step": 13454
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3211248463910263e-05,
+      "loss": 0.2507,
+      "step": 13455
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3210304538039711e-05,
+      "loss": 0.222,
+      "step": 13456
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3209360580277757e-05,
+      "loss": 0.2673,
+      "step": 13457
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3208416590633786e-05,
+      "loss": 0.2174,
+      "step": 13458
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3207472569117168e-05,
+      "loss": 0.242,
+      "step": 13459
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3206528515737287e-05,
+      "loss": 0.2526,
+      "step": 13460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.320558443050352e-05,
+      "loss": 0.2239,
+      "step": 13461
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3204640313425243e-05,
+      "loss": 0.2344,
+      "step": 13462
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.320369616451184e-05,
+      "loss": 0.2503,
+      "step": 13463
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3202751983772682e-05,
+      "loss": 0.2942,
+      "step": 13464
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3201807771217156e-05,
+      "loss": 0.2603,
+      "step": 13465
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3200863526854639e-05,
+      "loss": 0.2792,
+      "step": 13466
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319991925069451e-05,
+      "loss": 0.2224,
+      "step": 13467
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3198974942746151e-05,
+      "loss": 0.2012,
+      "step": 13468
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3198030603018945e-05,
+      "loss": 0.2181,
+      "step": 13469
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3197086231522271e-05,
+      "loss": 0.2242,
+      "step": 13470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319614182826551e-05,
+      "loss": 0.2925,
+      "step": 13471
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3195197393258043e-05,
+      "loss": 0.313,
+      "step": 13472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3194252926509253e-05,
+      "loss": 0.2533,
+      "step": 13473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3193308428028525e-05,
+      "loss": 0.2465,
+      "step": 13474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319236389782524e-05,
+      "loss": 0.2234,
+      "step": 13475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3191419335908775e-05,
+      "loss": 0.2722,
+      "step": 13476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3190474742288522e-05,
+      "loss": 0.2845,
+      "step": 13477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3189530116973863e-05,
+      "loss": 0.2818,
+      "step": 13478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3188585459974181e-05,
+      "loss": 0.3014,
+      "step": 13479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3187640771298857e-05,
+      "loss": 0.2237,
+      "step": 13480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.318669605095728e-05,
+      "loss": 0.2053,
+      "step": 13481
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3185751298958831e-05,
+      "loss": 0.2632,
+      "step": 13482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.31848065153129e-05,
+      "loss": 0.3064,
+      "step": 13483
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3183861700028867e-05,
+      "loss": 0.2503,
+      "step": 13484
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3182916853116124e-05,
+      "loss": 0.279,
+      "step": 13485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3181971974584053e-05,
+      "loss": 0.2273,
+      "step": 13486
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3181027064442043e-05,
+      "loss": 0.3568,
+      "step": 13487
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3180082122699479e-05,
+      "loss": 0.2362,
+      "step": 13488
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3179137149365744e-05,
+      "loss": 0.2067,
+      "step": 13489
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3178192144450234e-05,
+      "loss": 0.2298,
+      "step": 13490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3177247107962332e-05,
+      "loss": 0.2625,
+      "step": 13491
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3176302039911427e-05,
+      "loss": 0.2392,
+      "step": 13492
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3175356940306906e-05,
+      "loss": 0.2499,
+      "step": 13493
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.317441180915816e-05,
+      "loss": 0.2201,
+      "step": 13494
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3173466646474575e-05,
+      "loss": 0.2444,
+      "step": 13495
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3172521452265543e-05,
+      "loss": 0.2258,
+      "step": 13496
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3171576226540455e-05,
+      "loss": 0.2454,
+      "step": 13497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3170630969308695e-05,
+      "loss": 0.2416,
+      "step": 13498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316968568057966e-05,
+      "loss": 0.2625,
+      "step": 13499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3168740360362735e-05,
+      "loss": 0.241,
+      "step": 13500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3167795008667313e-05,
+      "loss": 0.2677,
+      "step": 13501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316684962550279e-05,
+      "loss": 0.21,
+      "step": 13502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316590421087855e-05,
+      "loss": 0.2688,
+      "step": 13503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3164958764803989e-05,
+      "loss": 0.2554,
+      "step": 13504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3164013287288495e-05,
+      "loss": 0.2741,
+      "step": 13505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3163067778341468e-05,
+      "loss": 0.2141,
+      "step": 13506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316212223797229e-05,
+      "loss": 0.2166,
+      "step": 13507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3161176666190366e-05,
+      "loss": 0.2443,
+      "step": 13508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3160231063005082e-05,
+      "loss": 0.224,
+      "step": 13509
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.315928542842583e-05,
+      "loss": 0.2626,
+      "step": 13510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3158339762462014e-05,
+      "loss": 0.277,
+      "step": 13511
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3157394065123018e-05,
+      "loss": 0.2753,
+      "step": 13512
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3156448336418237e-05,
+      "loss": 0.2589,
+      "step": 13513
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3155502576357076e-05,
+      "loss": 0.2199,
+      "step": 13514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3154556784948918e-05,
+      "loss": 0.272,
+      "step": 13515
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3153610962203165e-05,
+      "loss": 0.2521,
+      "step": 13516
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3152665108129212e-05,
+      "loss": 0.2129,
+      "step": 13517
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3151719222736456e-05,
+      "loss": 0.283,
+      "step": 13518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3150773306034292e-05,
+      "loss": 0.275,
+      "step": 13519
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3149827358032119e-05,
+      "loss": 0.2506,
+      "step": 13520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3148881378739334e-05,
+      "loss": 0.2424,
+      "step": 13521
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3147935368165328e-05,
+      "loss": 0.3026,
+      "step": 13522
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3146989326319509e-05,
+      "loss": 0.1973,
+      "step": 13523
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3146043253211265e-05,
+      "loss": 0.2282,
+      "step": 13524
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3145097148850002e-05,
+      "loss": 0.3,
+      "step": 13525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3144151013245115e-05,
+      "loss": 0.2574,
+      "step": 13526
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3143204846406001e-05,
+      "loss": 0.2532,
+      "step": 13527
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3142258648342065e-05,
+      "loss": 0.2329,
+      "step": 13528
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3141312419062703e-05,
+      "loss": 0.2701,
+      "step": 13529
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3140366158577316e-05,
+      "loss": 0.286,
+      "step": 13530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3139419866895303e-05,
+      "loss": 0.259,
+      "step": 13531
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3138473544026064e-05,
+      "loss": 0.2645,
+      "step": 13532
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3137527189979005e-05,
+      "loss": 0.2417,
+      "step": 13533
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313658080476352e-05,
+      "loss": 0.2568,
+      "step": 13534
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3135634388389016e-05,
+      "loss": 0.2433,
+      "step": 13535
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313468794086489e-05,
+      "loss": 0.2201,
+      "step": 13536
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3133741462200546e-05,
+      "loss": 0.2209,
+      "step": 13537
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313279495240539e-05,
+      "loss": 0.2319,
+      "step": 13538
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3131848411488819e-05,
+      "loss": 0.2116,
+      "step": 13539
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3130901839460242e-05,
+      "loss": 0.2492,
+      "step": 13540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3129955236329055e-05,
+      "loss": 0.2539,
+      "step": 13541
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.312900860210467e-05,
+      "loss": 0.2252,
+      "step": 13542
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3128061936796483e-05,
+      "loss": 0.2591,
+      "step": 13543
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3127115240413901e-05,
+      "loss": 0.2621,
+      "step": 13544
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3126168512966333e-05,
+      "loss": 0.2465,
+      "step": 13545
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.312522175446318e-05,
+      "loss": 0.2304,
+      "step": 13546
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3124274964913841e-05,
+      "loss": 0.2736,
+      "step": 13547
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3123328144327736e-05,
+      "loss": 0.1899,
+      "step": 13548
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3122381292714259e-05,
+      "loss": 0.1869,
+      "step": 13549
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3121434410082818e-05,
+      "loss": 0.2934,
+      "step": 13550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3120487496442824e-05,
+      "loss": 0.2774,
+      "step": 13551
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3119540551803682e-05,
+      "loss": 0.2483,
+      "step": 13552
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3118593576174793e-05,
+      "loss": 0.2243,
+      "step": 13553
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3117646569565573e-05,
+      "loss": 0.3019,
+      "step": 13554
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3116699531985426e-05,
+      "loss": 0.2441,
+      "step": 13555
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3115752463443756e-05,
+      "loss": 0.2339,
+      "step": 13556
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3114805363949978e-05,
+      "loss": 0.1762,
+      "step": 13557
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3113858233513499e-05,
+      "loss": 0.2938,
+      "step": 13558
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3112911072143723e-05,
+      "loss": 0.226,
+      "step": 13559
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3111963879850063e-05,
+      "loss": 0.2947,
+      "step": 13560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3111016656641929e-05,
+      "loss": 0.2008,
+      "step": 13561
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3110069402528729e-05,
+      "loss": 0.2457,
+      "step": 13562
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3109122117519875e-05,
+      "loss": 0.2677,
+      "step": 13563
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3108174801624778e-05,
+      "loss": 0.1936,
+      "step": 13564
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3107227454852841e-05,
+      "loss": 0.2447,
+      "step": 13565
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3106280077213486e-05,
+      "loss": 0.2499,
+      "step": 13566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3105332668716119e-05,
+      "loss": 0.2664,
+      "step": 13567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3104385229370154e-05,
+      "loss": 0.209,
+      "step": 13568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3103437759184995e-05,
+      "loss": 0.2304,
+      "step": 13569
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3102490258170063e-05,
+      "loss": 0.1996,
+      "step": 13570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.310154272633477e-05,
+      "loss": 0.2569,
+      "step": 13571
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3100595163688525e-05,
+      "loss": 0.2966,
+      "step": 13572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3099647570240741e-05,
+      "loss": 0.2239,
+      "step": 13573
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3098699946000835e-05,
+      "loss": 0.2565,
+      "step": 13574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3097752290978219e-05,
+      "loss": 0.2579,
+      "step": 13575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3096804605182307e-05,
+      "loss": 0.2483,
+      "step": 13576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3095856888622513e-05,
+      "loss": 0.2235,
+      "step": 13577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309490914130825e-05,
+      "loss": 0.1803,
+      "step": 13578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3093961363248936e-05,
+      "loss": 0.2552,
+      "step": 13579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3093013554453988e-05,
+      "loss": 0.2498,
+      "step": 13580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3092065714932818e-05,
+      "loss": 0.317,
+      "step": 13581
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309111784469484e-05,
+      "loss": 0.2584,
+      "step": 13582
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.2843,
+      "step": 13583
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3089222012106137e-05,
+      "loss": 0.1988,
+      "step": 13584
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3088274049774245e-05,
+      "loss": 0.2612,
+      "step": 13585
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3087326056763217e-05,
+      "loss": 0.2975,
+      "step": 13586
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3086378033082461e-05,
+      "loss": 0.2831,
+      "step": 13587
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3085429978741405e-05,
+      "loss": 0.2167,
+      "step": 13588
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3084481893749465e-05,
+      "loss": 0.2316,
+      "step": 13589
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3083533778116056e-05,
+      "loss": 0.2178,
+      "step": 13590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3082585631850601e-05,
+      "loss": 0.2653,
+      "step": 13591
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3081637454962517e-05,
+      "loss": 0.2546,
+      "step": 13592
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3080689247461221e-05,
+      "loss": 0.2297,
+      "step": 13593
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3079741009356134e-05,
+      "loss": 0.254,
+      "step": 13594
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3078792740656679e-05,
+      "loss": 0.2436,
+      "step": 13595
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307784444137227e-05,
+      "loss": 0.262,
+      "step": 13596
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3076896111512334e-05,
+      "loss": 0.2093,
+      "step": 13597
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307594775108629e-05,
+      "loss": 0.2748,
+      "step": 13598
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307499936010355e-05,
+      "loss": 0.2349,
+      "step": 13599
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307405093857355e-05,
+      "loss": 0.2439,
+      "step": 13600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3073102486505707e-05,
+      "loss": 0.2283,
+      "step": 13601
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3072154003909434e-05,
+      "loss": 0.2749,
+      "step": 13602
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3071205490794168e-05,
+      "loss": 0.2316,
+      "step": 13603
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307025694716932e-05,
+      "loss": 0.2207,
+      "step": 13604
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3069308373044315e-05,
+      "loss": 0.2753,
+      "step": 13605
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.306835976842858e-05,
+      "loss": 0.2162,
+      "step": 13606
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3067411133331536e-05,
+      "loss": 0.2536,
+      "step": 13607
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3066462467762605e-05,
+      "loss": 0.2721,
+      "step": 13608
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3065513771731218e-05,
+      "loss": 0.2431,
+      "step": 13609
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3064565045246794e-05,
+      "loss": 0.3092,
+      "step": 13610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3063616288318755e-05,
+      "loss": 0.2654,
+      "step": 13611
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.306266750095653e-05,
+      "loss": 0.2532,
+      "step": 13612
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3061718683169548e-05,
+      "loss": 0.237,
+      "step": 13613
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3060769834967228e-05,
+      "loss": 0.2778,
+      "step": 13614
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3059820956358998e-05,
+      "loss": 0.2561,
+      "step": 13615
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3058872047354285e-05,
+      "loss": 0.2579,
+      "step": 13616
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3057923107962513e-05,
+      "loss": 0.227,
+      "step": 13617
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3056974138193113e-05,
+      "loss": 0.2634,
+      "step": 13618
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.305602513805551e-05,
+      "loss": 0.2682,
+      "step": 13619
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3055076107559134e-05,
+      "loss": 0.3272,
+      "step": 13620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3054127046713404e-05,
+      "loss": 0.2783,
+      "step": 13621
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3053177955527762e-05,
+      "loss": 0.2573,
+      "step": 13622
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3052228834011625e-05,
+      "loss": 0.224,
+      "step": 13623
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3051279682174422e-05,
+      "loss": 0.2227,
+      "step": 13624
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3050330500025592e-05,
+      "loss": 0.2638,
+      "step": 13625
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3049381287574554e-05,
+      "loss": 0.2766,
+      "step": 13626
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3048432044830741e-05,
+      "loss": 0.2753,
+      "step": 13627
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3047482771803581e-05,
+      "loss": 0.2177,
+      "step": 13628
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3046533468502508e-05,
+      "loss": 0.2564,
+      "step": 13629
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.304558413493695e-05,
+      "loss": 0.25,
+      "step": 13630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.304463477111634e-05,
+      "loss": 0.2076,
+      "step": 13631
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3043685377050109e-05,
+      "loss": 0.2313,
+      "step": 13632
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.304273595274768e-05,
+      "loss": 0.2348,
+      "step": 13633
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3041786498218499e-05,
+      "loss": 0.2616,
+      "step": 13634
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3040837013471985e-05,
+      "loss": 0.1912,
+      "step": 13635
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3039887498517573e-05,
+      "loss": 0.2519,
+      "step": 13636
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3038937953364706e-05,
+      "loss": 0.2195,
+      "step": 13637
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3037988378022803e-05,
+      "loss": 0.175,
+      "step": 13638
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3037038772501305e-05,
+      "loss": 0.2021,
+      "step": 13639
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3036089136809645e-05,
+      "loss": 0.3178,
+      "step": 13640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3035139470957252e-05,
+      "loss": 0.2593,
+      "step": 13641
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3034189774953565e-05,
+      "loss": 0.2897,
+      "step": 13642
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3033240048808016e-05,
+      "loss": 0.2696,
+      "step": 13643
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3032290292530043e-05,
+      "loss": 0.2511,
+      "step": 13644
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3031340506129074e-05,
+      "loss": 0.2139,
+      "step": 13645
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.303039068961455e-05,
+      "loss": 0.2244,
+      "step": 13646
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3029440842995906e-05,
+      "loss": 0.2162,
+      "step": 13647
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3028490966282577e-05,
+      "loss": 0.2375,
+      "step": 13648
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3027541059483999e-05,
+      "loss": 0.3149,
+      "step": 13649
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3026591122609607e-05,
+      "loss": 0.2643,
+      "step": 13650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302564115566884e-05,
+      "loss": 0.2496,
+      "step": 13651
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3024691158671134e-05,
+      "loss": 0.2487,
+      "step": 13652
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3023741131625926e-05,
+      "loss": 0.2179,
+      "step": 13653
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3022791074542656e-05,
+      "loss": 0.2534,
+      "step": 13654
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3021840987430761e-05,
+      "loss": 0.2927,
+      "step": 13655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3020890870299675e-05,
+      "loss": 0.1789,
+      "step": 13656
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3019940723158844e-05,
+      "loss": 0.221,
+      "step": 13657
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3018990546017698e-05,
+      "loss": 0.292,
+      "step": 13658
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3018040338885684e-05,
+      "loss": 0.304,
+      "step": 13659
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3017090101772239e-05,
+      "loss": 0.2531,
+      "step": 13660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.30161398346868e-05,
+      "loss": 0.2571,
+      "step": 13661
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.301518953763881e-05,
+      "loss": 0.214,
+      "step": 13662
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3014239210637707e-05,
+      "loss": 0.2989,
+      "step": 13663
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3013288853692934e-05,
+      "loss": 0.2637,
+      "step": 13664
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.301233846681393e-05,
+      "loss": 0.2824,
+      "step": 13665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.301138805001014e-05,
+      "loss": 0.278,
+      "step": 13666
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3010437603291e-05,
+      "loss": 0.2393,
+      "step": 13667
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3009487126665957e-05,
+      "loss": 0.2132,
+      "step": 13668
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3008536620144448e-05,
+      "loss": 0.255,
+      "step": 13669
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3007586083735918e-05,
+      "loss": 0.2695,
+      "step": 13670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3006635517449811e-05,
+      "loss": 0.2131,
+      "step": 13671
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3005684921295568e-05,
+      "loss": 0.2673,
+      "step": 13672
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3004734295282633e-05,
+      "loss": 0.286,
+      "step": 13673
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3003783639420448e-05,
+      "loss": 0.2528,
+      "step": 13674
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3002832953718462e-05,
+      "loss": 0.2421,
+      "step": 13675
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3001882238186113e-05,
+      "loss": 0.1783,
+      "step": 13676
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3000931492832847e-05,
+      "loss": 0.2233,
+      "step": 13677
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2999980717668112e-05,
+      "loss": 0.2311,
+      "step": 13678
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2999029912701348e-05,
+      "loss": 0.2931,
+      "step": 13679
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2998079077942008e-05,
+      "loss": 0.2264,
+      "step": 13680
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.299712821339953e-05,
+      "loss": 0.1655,
+      "step": 13681
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2996177319083362e-05,
+      "loss": 0.2303,
+      "step": 13682
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2995226395002953e-05,
+      "loss": 0.2808,
+      "step": 13683
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2994275441167748e-05,
+      "loss": 0.2271,
+      "step": 13684
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2993324457587194e-05,
+      "loss": 0.2711,
+      "step": 13685
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2992373444270733e-05,
+      "loss": 0.2486,
+      "step": 13686
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2991422401227822e-05,
+      "loss": 0.2712,
+      "step": 13687
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2990471328467902e-05,
+      "loss": 0.2417,
+      "step": 13688
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2989520226000423e-05,
+      "loss": 0.217,
+      "step": 13689
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2988569093834835e-05,
+      "loss": 0.2481,
+      "step": 13690
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2987617931980581e-05,
+      "loss": 0.2614,
+      "step": 13691
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2986666740447118e-05,
+      "loss": 0.2419,
+      "step": 13692
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.298571551924389e-05,
+      "loss": 0.2488,
+      "step": 13693
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2984764268380343e-05,
+      "loss": 0.2023,
+      "step": 13694
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2983812987865936e-05,
+      "loss": 0.2939,
+      "step": 13695
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2982861677710113e-05,
+      "loss": 0.1996,
+      "step": 13696
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2981910337922323e-05,
+      "loss": 0.2881,
+      "step": 13697
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2980958968512022e-05,
+      "loss": 0.2935,
+      "step": 13698
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.298000756948866e-05,
+      "loss": 0.213,
+      "step": 13699
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2979056140861686e-05,
+      "loss": 0.2617,
+      "step": 13700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2978104682640548e-05,
+      "loss": 0.2582,
+      "step": 13701
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2977153194834706e-05,
+      "loss": 0.2904,
+      "step": 13702
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2976201677453608e-05,
+      "loss": 0.241,
+      "step": 13703
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2975250130506706e-05,
+      "loss": 0.2278,
+      "step": 13704
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2974298554003451e-05,
+      "loss": 0.2283,
+      "step": 13705
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.29733469479533e-05,
+      "loss": 0.2119,
+      "step": 13706
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2972395312365706e-05,
+      "loss": 0.2049,
+      "step": 13707
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.297144364725012e-05,
+      "loss": 0.2274,
+      "step": 13708
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2970491952616e-05,
+      "loss": 0.2782,
+      "step": 13709
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2969540228472792e-05,
+      "loss": 0.1962,
+      "step": 13710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296858847482996e-05,
+      "loss": 0.305,
+      "step": 13711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2967636691696953e-05,
+      "loss": 0.3002,
+      "step": 13712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296668487908323e-05,
+      "loss": 0.2154,
+      "step": 13713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2965733036998242e-05,
+      "loss": 0.2477,
+      "step": 13714
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2964781165451448e-05,
+      "loss": 0.2282,
+      "step": 13715
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.29638292644523e-05,
+      "loss": 0.2415,
+      "step": 13716
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2962877334010263e-05,
+      "loss": 0.2626,
+      "step": 13717
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2961925374134784e-05,
+      "loss": 0.2989,
+      "step": 13718
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2960973384835321e-05,
+      "loss": 0.2267,
+      "step": 13719
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2960021366121335e-05,
+      "loss": 0.2375,
+      "step": 13720
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2959069318002287e-05,
+      "loss": 0.2265,
+      "step": 13721
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2958117240487623e-05,
+      "loss": 0.2171,
+      "step": 13722
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.295716513358681e-05,
+      "loss": 0.2827,
+      "step": 13723
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2956212997309305e-05,
+      "loss": 0.2154,
+      "step": 13724
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2955260831664564e-05,
+      "loss": 0.2302,
+      "step": 13725
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.295430863666205e-05,
+      "loss": 0.2758,
+      "step": 13726
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2953356412311217e-05,
+      "loss": 0.2026,
+      "step": 13727
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.295240415862153e-05,
+      "loss": 0.2209,
+      "step": 13728
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2951451875602446e-05,
+      "loss": 0.2469,
+      "step": 13729
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2950499563263421e-05,
+      "loss": 0.2541,
+      "step": 13730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2949547221613923e-05,
+      "loss": 0.2755,
+      "step": 13731
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2948594850663405e-05,
+      "loss": 0.2117,
+      "step": 13732
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2947642450421336e-05,
+      "loss": 0.2807,
+      "step": 13733
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294669002089717e-05,
+      "loss": 0.2465,
+      "step": 13734
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2945737562100373e-05,
+      "loss": 0.2672,
+      "step": 13735
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2944785074040408e-05,
+      "loss": 0.2502,
+      "step": 13736
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294383255672673e-05,
+      "loss": 0.1997,
+      "step": 13737
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2942880010168808e-05,
+      "loss": 0.2061,
+      "step": 13738
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2941927434376101e-05,
+      "loss": 0.2654,
+      "step": 13739
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2940974829358074e-05,
+      "loss": 0.2176,
+      "step": 13740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2940022195124189e-05,
+      "loss": 0.2399,
+      "step": 13741
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2939069531683912e-05,
+      "loss": 0.2211,
+      "step": 13742
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2938116839046703e-05,
+      "loss": 0.2667,
+      "step": 13743
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.293716411722203e-05,
+      "loss": 0.2958,
+      "step": 13744
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2936211366219359e-05,
+      "loss": 0.2134,
+      "step": 13745
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2935258586048145e-05,
+      "loss": 0.2108,
+      "step": 13746
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2934305776717858e-05,
+      "loss": 0.2158,
+      "step": 13747
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2933352938237973e-05,
+      "loss": 0.2025,
+      "step": 13748
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2932400070617942e-05,
+      "loss": 0.2514,
+      "step": 13749
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2931447173867236e-05,
+      "loss": 0.2655,
+      "step": 13750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.293049424799532e-05,
+      "loss": 0.2818,
+      "step": 13751
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2929541293011662e-05,
+      "loss": 0.2018,
+      "step": 13752
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2928588308925729e-05,
+      "loss": 0.2487,
+      "step": 13753
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2927635295746987e-05,
+      "loss": 0.2549,
+      "step": 13754
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2926682253484906e-05,
+      "loss": 0.216,
+      "step": 13755
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2925729182148946e-05,
+      "loss": 0.2206,
+      "step": 13756
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2924776081748584e-05,
+      "loss": 0.2391,
+      "step": 13757
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2923822952293286e-05,
+      "loss": 0.242,
+      "step": 13758
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2922869793792512e-05,
+      "loss": 0.2631,
+      "step": 13759
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2921916606255742e-05,
+      "loss": 0.2938,
+      "step": 13760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.292096338969244e-05,
+      "loss": 0.2546,
+      "step": 13761
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2920010144112076e-05,
+      "loss": 0.261,
+      "step": 13762
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2919056869524118e-05,
+      "loss": 0.2136,
+      "step": 13763
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2918103565938037e-05,
+      "loss": 0.2084,
+      "step": 13764
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2917150233363307e-05,
+      "loss": 0.2094,
+      "step": 13765
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2916196871809392e-05,
+      "loss": 0.2363,
+      "step": 13766
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2915243481285768e-05,
+      "loss": 0.2412,
+      "step": 13767
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.29142900618019e-05,
+      "loss": 0.2568,
+      "step": 13768
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2913336613367264e-05,
+      "loss": 0.2733,
+      "step": 13769
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2912383135991333e-05,
+      "loss": 0.228,
+      "step": 13770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2911429629683574e-05,
+      "loss": 0.2313,
+      "step": 13771
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2910476094453466e-05,
+      "loss": 0.2352,
+      "step": 13772
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2909522530310475e-05,
+      "loss": 0.19,
+      "step": 13773
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2908568937264075e-05,
+      "loss": 0.2088,
+      "step": 13774
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.290761531532374e-05,
+      "loss": 0.2461,
+      "step": 13775
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2906661664498946e-05,
+      "loss": 0.1939,
+      "step": 13776
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2905707984799161e-05,
+      "loss": 0.2446,
+      "step": 13777
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2904754276233864e-05,
+      "loss": 0.2537,
+      "step": 13778
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2903800538812528e-05,
+      "loss": 0.2724,
+      "step": 13779
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2902846772544625e-05,
+      "loss": 0.2239,
+      "step": 13780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2901892977439632e-05,
+      "loss": 0.2243,
+      "step": 13781
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2900939153507026e-05,
+      "loss": 0.2207,
+      "step": 13782
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2899985300756277e-05,
+      "loss": 0.2776,
+      "step": 13783
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2899031419196864e-05,
+      "loss": 0.2507,
+      "step": 13784
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2898077508838263e-05,
+      "loss": 0.2194,
+      "step": 13785
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2897123569689949e-05,
+      "loss": 0.2473,
+      "step": 13786
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2896169601761402e-05,
+      "loss": 0.232,
+      "step": 13787
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2895215605062093e-05,
+      "loss": 0.2147,
+      "step": 13788
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2894261579601506e-05,
+      "loss": 0.224,
+      "step": 13789
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2893307525389109e-05,
+      "loss": 0.2176,
+      "step": 13790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2892353442434389e-05,
+      "loss": 0.2249,
+      "step": 13791
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2891399330746819e-05,
+      "loss": 0.2162,
+      "step": 13792
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2890445190335877e-05,
+      "loss": 0.245,
+      "step": 13793
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2889491021211045e-05,
+      "loss": 0.285,
+      "step": 13794
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2888536823381799e-05,
+      "loss": 0.2308,
+      "step": 13795
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2887582596857617e-05,
+      "loss": 0.3906,
+      "step": 13796
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2886628341647983e-05,
+      "loss": 0.2427,
+      "step": 13797
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288567405776237e-05,
+      "loss": 0.2563,
+      "step": 13798
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2884719745210264e-05,
+      "loss": 0.2083,
+      "step": 13799
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288376540400114e-05,
+      "loss": 0.2523,
+      "step": 13800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2882811034144487e-05,
+      "loss": 0.2691,
+      "step": 13801
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2881856635649774e-05,
+      "loss": 0.2528,
+      "step": 13802
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288090220852649e-05,
+      "loss": 0.2038,
+      "step": 13803
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2879947752784117e-05,
+      "loss": 0.2995,
+      "step": 13804
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.287899326843213e-05,
+      "loss": 0.213,
+      "step": 13805
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.287803875548002e-05,
+      "loss": 0.2579,
+      "step": 13806
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2877084213937262e-05,
+      "loss": 0.2084,
+      "step": 13807
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2876129643813337e-05,
+      "loss": 0.2251,
+      "step": 13808
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2875175045117737e-05,
+      "loss": 0.2069,
+      "step": 13809
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2874220417859934e-05,
+      "loss": 0.2368,
+      "step": 13810
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.287326576204942e-05,
+      "loss": 0.2356,
+      "step": 13811
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2872311077695676e-05,
+      "loss": 0.2759,
+      "step": 13812
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2871356364808184e-05,
+      "loss": 0.2216,
+      "step": 13813
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.287040162339643e-05,
+      "loss": 0.2295,
+      "step": 13814
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2869446853469898e-05,
+      "loss": 0.2003,
+      "step": 13815
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2868492055038075e-05,
+      "loss": 0.1927,
+      "step": 13816
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2867537228110441e-05,
+      "loss": 0.241,
+      "step": 13817
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2866582372696486e-05,
+      "loss": 0.2663,
+      "step": 13818
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2865627488805692e-05,
+      "loss": 0.2462,
+      "step": 13819
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2864672576447546e-05,
+      "loss": 0.2215,
+      "step": 13820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2863717635631537e-05,
+      "loss": 0.257,
+      "step": 13821
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2862762666367148e-05,
+      "loss": 0.2146,
+      "step": 13822
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2861807668663869e-05,
+      "loss": 0.1999,
+      "step": 13823
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.286085264253118e-05,
+      "loss": 0.2315,
+      "step": 13824
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.285989758797858e-05,
+      "loss": 0.1544,
+      "step": 13825
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2858942505015546e-05,
+      "loss": 0.2247,
+      "step": 13826
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2857987393651568e-05,
+      "loss": 0.2914,
+      "step": 13827
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.285703225389614e-05,
+      "loss": 0.3325,
+      "step": 13828
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2856077085758744e-05,
+      "loss": 0.2063,
+      "step": 13829
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2855121889248872e-05,
+      "loss": 0.1382,
+      "step": 13830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2854166664376011e-05,
+      "loss": 0.2259,
+      "step": 13831
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2853211411149651e-05,
+      "loss": 0.2521,
+      "step": 13832
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2852256129579282e-05,
+      "loss": 0.2444,
+      "step": 13833
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2851300819674395e-05,
+      "loss": 0.2128,
+      "step": 13834
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.285034548144448e-05,
+      "loss": 0.2484,
+      "step": 13835
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284939011489902e-05,
+      "loss": 0.2868,
+      "step": 13836
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284843472004752e-05,
+      "loss": 0.2461,
+      "step": 13837
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2847479296899459e-05,
+      "loss": 0.2547,
+      "step": 13838
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284652384546433e-05,
+      "loss": 0.2596,
+      "step": 13839
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2845568365751629e-05,
+      "loss": 0.2612,
+      "step": 13840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2844612857770847e-05,
+      "loss": 0.2119,
+      "step": 13841
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284365732153147e-05,
+      "loss": 0.2096,
+      "step": 13842
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2842701757042999e-05,
+      "loss": 0.2718,
+      "step": 13843
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284174616431492e-05,
+      "loss": 0.2185,
+      "step": 13844
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.284079054335673e-05,
+      "loss": 0.1969,
+      "step": 13845
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2839834894177918e-05,
+      "loss": 0.2526,
+      "step": 13846
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2838879216787985e-05,
+      "loss": 0.1955,
+      "step": 13847
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2837923511196416e-05,
+      "loss": 0.2383,
+      "step": 13848
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.283696777741271e-05,
+      "loss": 0.218,
+      "step": 13849
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2836012015446359e-05,
+      "loss": 0.2059,
+      "step": 13850
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2835056225306859e-05,
+      "loss": 0.2614,
+      "step": 13851
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2834100407003707e-05,
+      "loss": 0.1774,
+      "step": 13852
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2833144560546395e-05,
+      "loss": 0.308,
+      "step": 13853
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.283218868594442e-05,
+      "loss": 0.228,
+      "step": 13854
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2831232783207278e-05,
+      "loss": 0.2209,
+      "step": 13855
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2830276852344458e-05,
+      "loss": 0.212,
+      "step": 13856
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2829320893365469e-05,
+      "loss": 0.2528,
+      "step": 13857
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2828364906279797e-05,
+      "loss": 0.1871,
+      "step": 13858
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2827408891096948e-05,
+      "loss": 0.2176,
+      "step": 13859
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2826452847826407e-05,
+      "loss": 0.3015,
+      "step": 13860
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2825496776477686e-05,
+      "loss": 0.2288,
+      "step": 13861
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.282454067706027e-05,
+      "loss": 0.1791,
+      "step": 13862
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2823584549583663e-05,
+      "loss": 0.2189,
+      "step": 13863
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2822628394057362e-05,
+      "loss": 0.2581,
+      "step": 13864
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2821672210490865e-05,
+      "loss": 0.2067,
+      "step": 13865
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2820715998893673e-05,
+      "loss": 0.2165,
+      "step": 13866
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2819759759275284e-05,
+      "loss": 0.1828,
+      "step": 13867
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2818803491645195e-05,
+      "loss": 0.223,
+      "step": 13868
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2817847196012912e-05,
+      "loss": 0.2093,
+      "step": 13869
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2816890872387923e-05,
+      "loss": 0.1857,
+      "step": 13870
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2815934520779742e-05,
+      "loss": 0.2338,
+      "step": 13871
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2814978141197863e-05,
+      "loss": 0.2407,
+      "step": 13872
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2814021733651782e-05,
+      "loss": 0.2198,
+      "step": 13873
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.281306529815101e-05,
+      "loss": 0.2511,
+      "step": 13874
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.281210883470504e-05,
+      "loss": 0.2304,
+      "step": 13875
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.281115234332338e-05,
+      "loss": 0.2534,
+      "step": 13876
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2810195824015526e-05,
+      "loss": 0.2253,
+      "step": 13877
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2809239276790983e-05,
+      "loss": 0.2531,
+      "step": 13878
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2808282701659254e-05,
+      "loss": 0.2238,
+      "step": 13879
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2807326098629839e-05,
+      "loss": 0.2651,
+      "step": 13880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2806369467712247e-05,
+      "loss": 0.2175,
+      "step": 13881
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280541280891597e-05,
+      "loss": 0.2344,
+      "step": 13882
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2804456122250528e-05,
+      "loss": 0.2181,
+      "step": 13883
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280349940772541e-05,
+      "loss": 0.2786,
+      "step": 13884
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2802542665350124e-05,
+      "loss": 0.2773,
+      "step": 13885
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2801585895134178e-05,
+      "loss": 0.2257,
+      "step": 13886
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2800629097087076e-05,
+      "loss": 0.2732,
+      "step": 13887
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2799672271218319e-05,
+      "loss": 0.2458,
+      "step": 13888
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2798715417537415e-05,
+      "loss": 0.2197,
+      "step": 13889
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2797758536053872e-05,
+      "loss": 0.2991,
+      "step": 13890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2796801626777188e-05,
+      "loss": 0.2579,
+      "step": 13891
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2795844689716876e-05,
+      "loss": 0.3363,
+      "step": 13892
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.279488772488244e-05,
+      "loss": 0.2541,
+      "step": 13893
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2793930732283385e-05,
+      "loss": 0.2468,
+      "step": 13894
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2792973711929222e-05,
+      "loss": 0.2738,
+      "step": 13895
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2792016663829454e-05,
+      "loss": 0.2478,
+      "step": 13896
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.279105958799359e-05,
+      "loss": 0.1645,
+      "step": 13897
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2790102484431138e-05,
+      "loss": 0.2466,
+      "step": 13898
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2789145353151606e-05,
+      "loss": 0.2288,
+      "step": 13899
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2788188194164501e-05,
+      "loss": 0.2302,
+      "step": 13900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.278723100747933e-05,
+      "loss": 0.2498,
+      "step": 13901
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.278627379310561e-05,
+      "loss": 0.2994,
+      "step": 13902
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2785316551052837e-05,
+      "loss": 0.2569,
+      "step": 13903
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2784359281330532e-05,
+      "loss": 0.2182,
+      "step": 13904
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2783401983948198e-05,
+      "loss": 0.2276,
+      "step": 13905
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2782444658915348e-05,
+      "loss": 0.1753,
+      "step": 13906
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.278148730624149e-05,
+      "loss": 0.2234,
+      "step": 13907
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2780529925936133e-05,
+      "loss": 0.2381,
+      "step": 13908
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2779572518008793e-05,
+      "loss": 0.2407,
+      "step": 13909
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2778615082468977e-05,
+      "loss": 0.212,
+      "step": 13910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2777657619326198e-05,
+      "loss": 0.187,
+      "step": 13911
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2776700128589965e-05,
+      "loss": 0.2619,
+      "step": 13912
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2775742610269792e-05,
+      "loss": 0.3086,
+      "step": 13913
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2774785064375194e-05,
+      "loss": 0.2432,
+      "step": 13914
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2773827490915678e-05,
+      "loss": 0.2275,
+      "step": 13915
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2772869889900754e-05,
+      "loss": 0.2307,
+      "step": 13916
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2771912261339946e-05,
+      "loss": 0.2337,
+      "step": 13917
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2770954605242755e-05,
+      "loss": 0.1808,
+      "step": 13918
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2769996921618702e-05,
+      "loss": 0.2287,
+      "step": 13919
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2769039210477299e-05,
+      "loss": 0.2271,
+      "step": 13920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.276808147182806e-05,
+      "loss": 0.2513,
+      "step": 13921
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2767123705680497e-05,
+      "loss": 0.2542,
+      "step": 13922
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2766165912044128e-05,
+      "loss": 0.2045,
+      "step": 13923
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2765208090928466e-05,
+      "loss": 0.295,
+      "step": 13924
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2764250242343025e-05,
+      "loss": 0.2293,
+      "step": 13925
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2763292366297321e-05,
+      "loss": 0.2128,
+      "step": 13926
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2762334462800877e-05,
+      "loss": 0.214,
+      "step": 13927
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2761376531863193e-05,
+      "loss": 0.2578,
+      "step": 13928
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2760418573493798e-05,
+      "loss": 0.2927,
+      "step": 13929
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2759460587702205e-05,
+      "loss": 0.205,
+      "step": 13930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.275850257449793e-05,
+      "loss": 0.2379,
+      "step": 13931
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.275754453389049e-05,
+      "loss": 0.2959,
+      "step": 13932
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2756586465889403e-05,
+      "loss": 0.1936,
+      "step": 13933
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2755628370504185e-05,
+      "loss": 0.2598,
+      "step": 13934
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2754670247744353e-05,
+      "loss": 0.2325,
+      "step": 13935
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2753712097619433e-05,
+      "loss": 0.2444,
+      "step": 13936
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2752753920138932e-05,
+      "loss": 0.1863,
+      "step": 13937
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2751795715312377e-05,
+      "loss": 0.1892,
+      "step": 13938
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2750837483149283e-05,
+      "loss": 0.262,
+      "step": 13939
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2749879223659166e-05,
+      "loss": 0.1851,
+      "step": 13940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2748920936851554e-05,
+      "loss": 0.2476,
+      "step": 13941
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2747962622735961e-05,
+      "loss": 0.2278,
+      "step": 13942
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2747004281321906e-05,
+      "loss": 0.2284,
+      "step": 13943
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2746045912618914e-05,
+      "loss": 0.2553,
+      "step": 13944
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2745087516636501e-05,
+      "loss": 0.2708,
+      "step": 13945
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2744129093384188e-05,
+      "loss": 0.231,
+      "step": 13946
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2743170642871498e-05,
+      "loss": 0.2738,
+      "step": 13947
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2742212165107957e-05,
+      "loss": 0.2962,
+      "step": 13948
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2741253660103077e-05,
+      "loss": 0.2122,
+      "step": 13949
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2740295127866383e-05,
+      "loss": 0.2266,
+      "step": 13950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.27393365684074e-05,
+      "loss": 0.2454,
+      "step": 13951
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.273837798173565e-05,
+      "loss": 0.2216,
+      "step": 13952
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2737419367860652e-05,
+      "loss": 0.2139,
+      "step": 13953
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2736460726791933e-05,
+      "loss": 0.3187,
+      "step": 13954
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2735502058539014e-05,
+      "loss": 0.3178,
+      "step": 13955
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2734543363111418e-05,
+      "loss": 0.1991,
+      "step": 13956
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2733584640518672e-05,
+      "loss": 0.2351,
+      "step": 13957
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2732625890770299e-05,
+      "loss": 0.2426,
+      "step": 13958
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.273166711387582e-05,
+      "loss": 0.1948,
+      "step": 13959
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.273070830984476e-05,
+      "loss": 0.3119,
+      "step": 13960
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272974947868665e-05,
+      "loss": 0.226,
+      "step": 13961
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2728790620411005e-05,
+      "loss": 0.2639,
+      "step": 13962
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272783173502736e-05,
+      "loss": 0.2201,
+      "step": 13963
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2726872822545237e-05,
+      "loss": 0.2518,
+      "step": 13964
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272591388297416e-05,
+      "loss": 0.2362,
+      "step": 13965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2724954916323658e-05,
+      "loss": 0.2294,
+      "step": 13966
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2723995922603253e-05,
+      "loss": 0.2472,
+      "step": 13967
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2723036901822476e-05,
+      "loss": 0.294,
+      "step": 13968
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2722077853990855e-05,
+      "loss": 0.2395,
+      "step": 13969
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2721118779117915e-05,
+      "loss": 0.2331,
+      "step": 13970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2720159677213182e-05,
+      "loss": 0.2516,
+      "step": 13971
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2719200548286189e-05,
+      "loss": 0.2223,
+      "step": 13972
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2718241392346456e-05,
+      "loss": 0.2275,
+      "step": 13973
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2717282209403514e-05,
+      "loss": 0.2195,
+      "step": 13974
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.27163229994669e-05,
+      "loss": 0.2007,
+      "step": 13975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2715363762546131e-05,
+      "loss": 0.3187,
+      "step": 13976
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2714404498650743e-05,
+      "loss": 0.266,
+      "step": 13977
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2713445207790264e-05,
+      "loss": 0.2292,
+      "step": 13978
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2712485889974224e-05,
+      "loss": 0.2422,
+      "step": 13979
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2711526545212151e-05,
+      "loss": 0.1897,
+      "step": 13980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2710567173513577e-05,
+      "loss": 0.2182,
+      "step": 13981
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2709607774888036e-05,
+      "loss": 0.3152,
+      "step": 13982
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.270864834934505e-05,
+      "loss": 0.2915,
+      "step": 13983
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2707688896894154e-05,
+      "loss": 0.2757,
+      "step": 13984
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2706729417544885e-05,
+      "loss": 0.2591,
+      "step": 13985
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2705769911306768e-05,
+      "loss": 0.2302,
+      "step": 13986
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2704810378189333e-05,
+      "loss": 0.3232,
+      "step": 13987
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2703850818202116e-05,
+      "loss": 0.2647,
+      "step": 13988
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2702891231354653e-05,
+      "loss": 0.2567,
+      "step": 13989
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2701931617656468e-05,
+      "loss": 0.2175,
+      "step": 13990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2700971977117101e-05,
+      "loss": 0.2518,
+      "step": 13991
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2700012309746081e-05,
+      "loss": 0.2094,
+      "step": 13992
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2699052615552945e-05,
+      "loss": 0.2421,
+      "step": 13993
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2698092894547224e-05,
+      "loss": 0.3052,
+      "step": 13994
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2697133146738452e-05,
+      "loss": 0.2605,
+      "step": 13995
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.269617337213616e-05,
+      "loss": 0.1834,
+      "step": 13996
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2695213570749891e-05,
+      "loss": 0.2178,
+      "step": 13997
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2694253742589173e-05,
+      "loss": 0.3208,
+      "step": 13998
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2693293887663542e-05,
+      "loss": 0.2926,
+      "step": 13999
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2692334005982535e-05,
+      "loss": 0.2375,
+      "step": 14000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2691374097555687e-05,
+      "loss": 0.2307,
+      "step": 14001
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2690414162392533e-05,
+      "loss": 0.216,
+      "step": 14002
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2689454200502609e-05,
+      "loss": 0.2432,
+      "step": 14003
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2688494211895455e-05,
+      "loss": 0.2847,
+      "step": 14004
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2687534196580597e-05,
+      "loss": 0.2563,
+      "step": 14005
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2686574154567587e-05,
+      "loss": 0.2786,
+      "step": 14006
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2685614085865952e-05,
+      "loss": 0.2812,
+      "step": 14007
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.268465399048523e-05,
+      "loss": 0.2333,
+      "step": 14008
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2683693868434964e-05,
+      "loss": 0.2647,
+      "step": 14009
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2682733719724685e-05,
+      "loss": 0.2231,
+      "step": 14010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2681773544363935e-05,
+      "loss": 0.276,
+      "step": 14011
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2680813342362255e-05,
+      "loss": 0.1544,
+      "step": 14012
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267985311372918e-05,
+      "loss": 0.2227,
+      "step": 14013
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2678892858474246e-05,
+      "loss": 0.255,
+      "step": 14014
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2677932576606998e-05,
+      "loss": 0.2421,
+      "step": 14015
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2676972268136976e-05,
+      "loss": 0.2803,
+      "step": 14016
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2676011933073712e-05,
+      "loss": 0.2604,
+      "step": 14017
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2675051571426757e-05,
+      "loss": 0.2524,
+      "step": 14018
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2674091183205644e-05,
+      "loss": 0.2322,
+      "step": 14019
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2673130768419912e-05,
+      "loss": 0.2229,
+      "step": 14020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2672170327079108e-05,
+      "loss": 0.2309,
+      "step": 14021
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267120985919277e-05,
+      "loss": 0.256,
+      "step": 14022
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2670249364770438e-05,
+      "loss": 0.2049,
+      "step": 14023
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2669288843821657e-05,
+      "loss": 0.2172,
+      "step": 14024
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2668328296355964e-05,
+      "loss": 0.2611,
+      "step": 14025
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2667367722382906e-05,
+      "loss": 0.2865,
+      "step": 14026
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2666407121912023e-05,
+      "loss": 0.2906,
+      "step": 14027
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2665446494952862e-05,
+      "loss": 0.2149,
+      "step": 14028
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2664485841514959e-05,
+      "loss": 0.2547,
+      "step": 14029
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.266352516160786e-05,
+      "loss": 0.2804,
+      "step": 14030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2662564455241108e-05,
+      "loss": 0.2654,
+      "step": 14031
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2661603722424249e-05,
+      "loss": 0.225,
+      "step": 14032
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2660642963166824e-05,
+      "loss": 0.2522,
+      "step": 14033
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265968217747838e-05,
+      "loss": 0.1984,
+      "step": 14034
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2658721365368462e-05,
+      "loss": 0.2292,
+      "step": 14035
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2657760526846609e-05,
+      "loss": 0.2216,
+      "step": 14036
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2656799661922375e-05,
+      "loss": 0.2274,
+      "step": 14037
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2655838770605297e-05,
+      "loss": 0.2072,
+      "step": 14038
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2654877852904924e-05,
+      "loss": 0.319,
+      "step": 14039
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2653916908830804e-05,
+      "loss": 0.2276,
+      "step": 14040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265295593839248e-05,
+      "loss": 0.2452,
+      "step": 14041
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2651994941599499e-05,
+      "loss": 0.1639,
+      "step": 14042
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2651033918461406e-05,
+      "loss": 0.2542,
+      "step": 14043
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2650072868987753e-05,
+      "loss": 0.1723,
+      "step": 14044
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.264911179318808e-05,
+      "loss": 0.2773,
+      "step": 14045
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.264815069107194e-05,
+      "loss": 0.2388,
+      "step": 14046
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2647189562648882e-05,
+      "loss": 0.2062,
+      "step": 14047
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2646228407928444e-05,
+      "loss": 0.2323,
+      "step": 14048
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2645267226920188e-05,
+      "loss": 0.1834,
+      "step": 14049
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2644306019633654e-05,
+      "loss": 0.2601,
+      "step": 14050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2643344786078386e-05,
+      "loss": 0.2443,
+      "step": 14051
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2642383526263948e-05,
+      "loss": 0.241,
+      "step": 14052
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2641422240199876e-05,
+      "loss": 0.2435,
+      "step": 14053
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2640460927895721e-05,
+      "loss": 0.2099,
+      "step": 14054
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2639499589361041e-05,
+      "loss": 0.2166,
+      "step": 14055
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2638538224605377e-05,
+      "loss": 0.1929,
+      "step": 14056
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2637576833638285e-05,
+      "loss": 0.2434,
+      "step": 14057
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2636615416469313e-05,
+      "loss": 0.2742,
+      "step": 14058
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2635653973108013e-05,
+      "loss": 0.3186,
+      "step": 14059
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2634692503563932e-05,
+      "loss": 0.2339,
+      "step": 14060
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2633731007846626e-05,
+      "loss": 0.2415,
+      "step": 14061
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2632769485965645e-05,
+      "loss": 0.1992,
+      "step": 14062
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.263180793793054e-05,
+      "loss": 0.1792,
+      "step": 14063
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2630846363750868e-05,
+      "loss": 0.2619,
+      "step": 14064
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2629884763436176e-05,
+      "loss": 0.3144,
+      "step": 14065
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2628923136996014e-05,
+      "loss": 0.227,
+      "step": 14066
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2627961484439944e-05,
+      "loss": 0.2567,
+      "step": 14067
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.262699980577751e-05,
+      "loss": 0.1905,
+      "step": 14068
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.262603810101827e-05,
+      "loss": 0.2661,
+      "step": 14069
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2625076370171778e-05,
+      "loss": 0.2503,
+      "step": 14070
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2624114613247588e-05,
+      "loss": 0.2117,
+      "step": 14071
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2623152830255249e-05,
+      "loss": 0.2444,
+      "step": 14072
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2622191021204323e-05,
+      "loss": 0.222,
+      "step": 14073
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2621229186104363e-05,
+      "loss": 0.2554,
+      "step": 14074
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2620267324964918e-05,
+      "loss": 0.2595,
+      "step": 14075
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2619305437795548e-05,
+      "loss": 0.2578,
+      "step": 14076
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2618343524605809e-05,
+      "loss": 0.2638,
+      "step": 14077
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2617381585405255e-05,
+      "loss": 0.2137,
+      "step": 14078
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2616419620203442e-05,
+      "loss": 0.2777,
+      "step": 14079
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2615457629009926e-05,
+      "loss": 0.2837,
+      "step": 14080
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2614495611834265e-05,
+      "loss": 0.2263,
+      "step": 14081
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2613533568686016e-05,
+      "loss": 0.278,
+      "step": 14082
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2612571499574736e-05,
+      "loss": 0.271,
+      "step": 14083
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2611609404509979e-05,
+      "loss": 0.2263,
+      "step": 14084
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2610647283501303e-05,
+      "loss": 0.2156,
+      "step": 14085
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.260968513655827e-05,
+      "loss": 0.2545,
+      "step": 14086
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2608722963690437e-05,
+      "loss": 0.2753,
+      "step": 14087
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2607760764907357e-05,
+      "loss": 0.19,
+      "step": 14088
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2606798540218595e-05,
+      "loss": 0.2232,
+      "step": 14089
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2605836289633705e-05,
+      "loss": 0.2565,
+      "step": 14090
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2604874013162249e-05,
+      "loss": 0.2224,
+      "step": 14091
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2603911710813788e-05,
+      "loss": 0.2339,
+      "step": 14092
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2602949382597877e-05,
+      "loss": 0.2172,
+      "step": 14093
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2601987028524077e-05,
+      "loss": 0.2698,
+      "step": 14094
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.260102464860195e-05,
+      "loss": 0.2365,
+      "step": 14095
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2600062242841059e-05,
+      "loss": 0.2366,
+      "step": 14096
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2599099811250955e-05,
+      "loss": 0.2529,
+      "step": 14097
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2598137353841209e-05,
+      "loss": 0.2491,
+      "step": 14098
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2597174870621376e-05,
+      "loss": 0.2234,
+      "step": 14099
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2596212361601019e-05,
+      "loss": 0.1781,
+      "step": 14100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2595249826789702e-05,
+      "loss": 0.2557,
+      "step": 14101
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2594287266196985e-05,
+      "loss": 0.3323,
+      "step": 14102
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2593324679832429e-05,
+      "loss": 0.2484,
+      "step": 14103
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2592362067705598e-05,
+      "loss": 0.2325,
+      "step": 14104
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2591399429826054e-05,
+      "loss": 0.2588,
+      "step": 14105
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2590436766203359e-05,
+      "loss": 0.2462,
+      "step": 14106
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2589474076847077e-05,
+      "loss": 0.1608,
+      "step": 14107
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2588511361766776e-05,
+      "loss": 0.2491,
+      "step": 14108
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.258754862097201e-05,
+      "loss": 0.2277,
+      "step": 14109
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2586585854472351e-05,
+      "loss": 0.2848,
+      "step": 14110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2585623062277359e-05,
+      "loss": 0.2305,
+      "step": 14111
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.25846602443966e-05,
+      "loss": 0.2028,
+      "step": 14112
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2583697400839641e-05,
+      "loss": 0.2392,
+      "step": 14113
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2582734531616044e-05,
+      "loss": 0.2461,
+      "step": 14114
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.258177163673537e-05,
+      "loss": 0.193,
+      "step": 14115
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2580808716207194e-05,
+      "loss": 0.2336,
+      "step": 14116
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2579845770041077e-05,
+      "loss": 0.2056,
+      "step": 14117
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2578882798246585e-05,
+      "loss": 0.2881,
+      "step": 14118
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257791980083328e-05,
+      "loss": 0.2456,
+      "step": 14119
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2576956777810738e-05,
+      "loss": 0.2309,
+      "step": 14120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2575993729188518e-05,
+      "loss": 0.2149,
+      "step": 14121
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257503065497619e-05,
+      "loss": 0.2909,
+      "step": 14122
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257406755518332e-05,
+      "loss": 0.2301,
+      "step": 14123
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2573104429819475e-05,
+      "loss": 0.2307,
+      "step": 14124
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2572141278894226e-05,
+      "loss": 0.21,
+      "step": 14125
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2571178102417138e-05,
+      "loss": 0.2812,
+      "step": 14126
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2570214900397783e-05,
+      "loss": 0.2021,
+      "step": 14127
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2569251672845722e-05,
+      "loss": 0.1977,
+      "step": 14128
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2568288419770532e-05,
+      "loss": 0.2837,
+      "step": 14129
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2567325141181777e-05,
+      "loss": 0.2738,
+      "step": 14130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2566361837089027e-05,
+      "loss": 0.2457,
+      "step": 14131
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2565398507501853e-05,
+      "loss": 0.2492,
+      "step": 14132
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2564435152429825e-05,
+      "loss": 0.1897,
+      "step": 14133
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2563471771882512e-05,
+      "loss": 0.2759,
+      "step": 14134
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2562508365869484e-05,
+      "loss": 0.2345,
+      "step": 14135
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2561544934400312e-05,
+      "loss": 0.203,
+      "step": 14136
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2560581477484566e-05,
+      "loss": 0.2047,
+      "step": 14137
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2559617995131819e-05,
+      "loss": 0.274,
+      "step": 14138
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2558654487351643e-05,
+      "loss": 0.2626,
+      "step": 14139
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.25576909541536e-05,
+      "loss": 0.2173,
+      "step": 14140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255672739554728e-05,
+      "loss": 0.194,
+      "step": 14141
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255576381154224e-05,
+      "loss": 0.1857,
+      "step": 14142
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2554800202148052e-05,
+      "loss": 0.2353,
+      "step": 14143
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.25538365673743e-05,
+      "loss": 0.3055,
+      "step": 14144
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2552872907230547e-05,
+      "loss": 0.2079,
+      "step": 14145
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255190922172637e-05,
+      "loss": 0.2525,
+      "step": 14146
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2550945510871342e-05,
+      "loss": 0.2346,
+      "step": 14147
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2549981774675036e-05,
+      "loss": 0.2352,
+      "step": 14148
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2549018013147026e-05,
+      "loss": 0.2345,
+      "step": 14149
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2548054226296887e-05,
+      "loss": 0.2302,
+      "step": 14150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2547090414134193e-05,
+      "loss": 0.2503,
+      "step": 14151
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2546126576668516e-05,
+      "loss": 0.2707,
+      "step": 14152
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2545162713909433e-05,
+      "loss": 0.2308,
+      "step": 14153
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2544198825866521e-05,
+      "loss": 0.231,
+      "step": 14154
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2543234912549352e-05,
+      "loss": 0.2377,
+      "step": 14155
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2542270973967504e-05,
+      "loss": 0.2452,
+      "step": 14156
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2541307010130548e-05,
+      "loss": 0.2918,
+      "step": 14157
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2540343021048068e-05,
+      "loss": 0.2187,
+      "step": 14158
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2539379006729633e-05,
+      "loss": 0.2223,
+      "step": 14159
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2538414967184825e-05,
+      "loss": 0.2568,
+      "step": 14160
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2537450902423216e-05,
+      "loss": 0.2433,
+      "step": 14161
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2536486812454387e-05,
+      "loss": 0.2616,
+      "step": 14162
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2535522697287916e-05,
+      "loss": 0.2289,
+      "step": 14163
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2534558556933379e-05,
+      "loss": 0.2591,
+      "step": 14164
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2533594391400347e-05,
+      "loss": 0.2498,
+      "step": 14165
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.253263020069841e-05,
+      "loss": 0.2453,
+      "step": 14166
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.253166598483714e-05,
+      "loss": 0.2336,
+      "step": 14167
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2530701743826117e-05,
+      "loss": 0.1806,
+      "step": 14168
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252973747767492e-05,
+      "loss": 0.2424,
+      "step": 14169
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2528773186393126e-05,
+      "loss": 0.3203,
+      "step": 14170
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2527808869990312e-05,
+      "loss": 0.232,
+      "step": 14171
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2526844528476065e-05,
+      "loss": 0.2223,
+      "step": 14172
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2525880161859965e-05,
+      "loss": 0.2236,
+      "step": 14173
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252491577015158e-05,
+      "loss": 0.2929,
+      "step": 14174
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2523951353360504e-05,
+      "loss": 0.2435,
+      "step": 14175
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252298691149631e-05,
+      "loss": 0.2439,
+      "step": 14176
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252202244456858e-05,
+      "loss": 0.2047,
+      "step": 14177
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2521057952586898e-05,
+      "loss": 0.2122,
+      "step": 14178
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2520093435560845e-05,
+      "loss": 0.2898,
+      "step": 14179
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2519128893499999e-05,
+      "loss": 0.2903,
+      "step": 14180
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2518164326413944e-05,
+      "loss": 0.2347,
+      "step": 14181
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2517199734312264e-05,
+      "loss": 0.2309,
+      "step": 14182
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2516235117204536e-05,
+      "loss": 0.2232,
+      "step": 14183
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2515270475100345e-05,
+      "loss": 0.2484,
+      "step": 14184
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2514305808009279e-05,
+      "loss": 0.2363,
+      "step": 14185
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.251334111594091e-05,
+      "loss": 0.1883,
+      "step": 14186
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2512376398904835e-05,
+      "loss": 0.245,
+      "step": 14187
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2511411656910627e-05,
+      "loss": 0.2457,
+      "step": 14188
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2510446889967876e-05,
+      "loss": 0.2657,
+      "step": 14189
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2509482098086161e-05,
+      "loss": 0.252,
+      "step": 14190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.250851728127507e-05,
+      "loss": 0.2511,
+      "step": 14191
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2507552439544186e-05,
+      "loss": 0.2608,
+      "step": 14192
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2506587572903093e-05,
+      "loss": 0.2809,
+      "step": 14193
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2505622681361378e-05,
+      "loss": 0.2317,
+      "step": 14194
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2504657764928626e-05,
+      "loss": 0.1682,
+      "step": 14195
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.250369282361442e-05,
+      "loss": 0.2133,
+      "step": 14196
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2502727857428352e-05,
+      "loss": 0.2023,
+      "step": 14197
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2501762866380004e-05,
+      "loss": 0.2286,
+      "step": 14198
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2500797850478956e-05,
+      "loss": 0.2812,
+      "step": 14199
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2499832809734805e-05,
+      "loss": 0.2974,
+      "step": 14200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2498867744157133e-05,
+      "loss": 0.2206,
+      "step": 14201
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2497902653755528e-05,
+      "loss": 0.2165,
+      "step": 14202
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2496937538539575e-05,
+      "loss": 0.2793,
+      "step": 14203
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2495972398518865e-05,
+      "loss": 0.2236,
+      "step": 14204
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2495007233702982e-05,
+      "loss": 0.2171,
+      "step": 14205
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2494042044101518e-05,
+      "loss": 0.2747,
+      "step": 14206
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2493076829724063e-05,
+      "loss": 0.2786,
+      "step": 14207
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2492111590580194e-05,
+      "loss": 0.233,
+      "step": 14208
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2491146326679514e-05,
+      "loss": 0.2924,
+      "step": 14209
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2490181038031604e-05,
+      "loss": 0.2308,
+      "step": 14210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2489215724646053e-05,
+      "loss": 0.1749,
+      "step": 14211
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2488250386532452e-05,
+      "loss": 0.2704,
+      "step": 14212
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2487285023700392e-05,
+      "loss": 0.3046,
+      "step": 14213
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2486319636159465e-05,
+      "loss": 0.2288,
+      "step": 14214
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2485354223919255e-05,
+      "loss": 0.2772,
+      "step": 14215
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2484388786989359e-05,
+      "loss": 0.2391,
+      "step": 14216
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.248342332537936e-05,
+      "loss": 0.2871,
+      "step": 14217
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2482457839098856e-05,
+      "loss": 0.2804,
+      "step": 14218
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2481492328157439e-05,
+      "loss": 0.2558,
+      "step": 14219
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2480526792564691e-05,
+      "loss": 0.2661,
+      "step": 14220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2479561232330216e-05,
+      "loss": 0.2298,
+      "step": 14221
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2478595647463597e-05,
+      "loss": 0.2326,
+      "step": 14222
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2477630037974424e-05,
+      "loss": 0.2162,
+      "step": 14223
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2476664403872303e-05,
+      "loss": 0.2008,
+      "step": 14224
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2475698745166813e-05,
+      "loss": 0.2456,
+      "step": 14225
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2474733061867554e-05,
+      "loss": 0.2821,
+      "step": 14226
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2473767353984117e-05,
+      "loss": 0.2275,
+      "step": 14227
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2472801621526096e-05,
+      "loss": 0.2324,
+      "step": 14228
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2471835864503082e-05,
+      "loss": 0.2293,
+      "step": 14229
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2470870082924674e-05,
+      "loss": 0.2262,
+      "step": 14230
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2469904276800462e-05,
+      "loss": 0.2357,
+      "step": 14231
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.246893844614004e-05,
+      "loss": 0.2276,
+      "step": 14232
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2467972590953008e-05,
+      "loss": 0.2131,
+      "step": 14233
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2467006711248955e-05,
+      "loss": 0.2437,
+      "step": 14234
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2466040807037478e-05,
+      "loss": 0.2072,
+      "step": 14235
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2465074878328174e-05,
+      "loss": 0.2625,
+      "step": 14236
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2464108925130635e-05,
+      "loss": 0.2219,
+      "step": 14237
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2463142947454462e-05,
+      "loss": 0.2402,
+      "step": 14238
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2462176945309247e-05,
+      "loss": 0.2545,
+      "step": 14239
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2461210918704588e-05,
+      "loss": 0.2741,
+      "step": 14240
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.246024486765008e-05,
+      "loss": 0.27,
+      "step": 14241
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2459278792155323e-05,
+      "loss": 0.2298,
+      "step": 14242
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2458312692229912e-05,
+      "loss": 0.241,
+      "step": 14243
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2457346567883444e-05,
+      "loss": 0.2583,
+      "step": 14244
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2456380419125516e-05,
+      "loss": 0.2201,
+      "step": 14245
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2455414245965726e-05,
+      "loss": 0.305,
+      "step": 14246
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2454448048413674e-05,
+      "loss": 0.2722,
+      "step": 14247
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2453481826478958e-05,
+      "loss": 0.2137,
+      "step": 14248
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2452515580171171e-05,
+      "loss": 0.3167,
+      "step": 14249
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2451549309499922e-05,
+      "loss": 0.222,
+      "step": 14250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.24505830144748e-05,
+      "loss": 0.2418,
+      "step": 14251
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2449616695105414e-05,
+      "loss": 0.264,
+      "step": 14252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2448650351401352e-05,
+      "loss": 0.2479,
+      "step": 14253
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.244768398337222e-05,
+      "loss": 0.2471,
+      "step": 14254
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2446717591027624e-05,
+      "loss": 0.2265,
+      "step": 14255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2445751174377151e-05,
+      "loss": 0.2399,
+      "step": 14256
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2444784733430412e-05,
+      "loss": 0.2714,
+      "step": 14257
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2443818268197003e-05,
+      "loss": 0.2156,
+      "step": 14258
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2442851778686526e-05,
+      "loss": 0.2212,
+      "step": 14259
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2441885264908584e-05,
+      "loss": 0.2303,
+      "step": 14260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2440918726872774e-05,
+      "loss": 0.2756,
+      "step": 14261
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2439952164588705e-05,
+      "loss": 0.2245,
+      "step": 14262
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2438985578065968e-05,
+      "loss": 0.2812,
+      "step": 14263
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2438018967314173e-05,
+      "loss": 0.2228,
+      "step": 14264
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2437052332342923e-05,
+      "loss": 0.2884,
+      "step": 14265
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2436085673161815e-05,
+      "loss": 0.2662,
+      "step": 14266
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2435118989780459e-05,
+      "loss": 0.2252,
+      "step": 14267
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2434152282208451e-05,
+      "loss": 0.2237,
+      "step": 14268
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.24331855504554e-05,
+      "loss": 0.2215,
+      "step": 14269
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2432218794530906e-05,
+      "loss": 0.2101,
+      "step": 14270
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2431252014444574e-05,
+      "loss": 0.1971,
+      "step": 14271
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2430285210206011e-05,
+      "loss": 0.2882,
+      "step": 14272
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2429318381824813e-05,
+      "loss": 0.2511,
+      "step": 14273
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2428351529310596e-05,
+      "loss": 0.2248,
+      "step": 14274
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2427384652672954e-05,
+      "loss": 0.2662,
+      "step": 14275
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2426417751921496e-05,
+      "loss": 0.276,
+      "step": 14276
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2425450827065833e-05,
+      "loss": 0.2225,
+      "step": 14277
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2424483878115563e-05,
+      "loss": 0.2424,
+      "step": 14278
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2423516905080292e-05,
+      "loss": 0.2608,
+      "step": 14279
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2422549907969632e-05,
+      "loss": 0.2376,
+      "step": 14280
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2421582886793184e-05,
+      "loss": 0.2931,
+      "step": 14281
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2420615841560556e-05,
+      "loss": 0.2057,
+      "step": 14282
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2419648772281355e-05,
+      "loss": 0.2278,
+      "step": 14283
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2418681678965185e-05,
+      "loss": 0.3,
+      "step": 14284
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2417714561621656e-05,
+      "loss": 0.2501,
+      "step": 14285
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.241674742026038e-05,
+      "loss": 0.2403,
+      "step": 14286
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2415780254890956e-05,
+      "loss": 0.2493,
+      "step": 14287
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2414813065522995e-05,
+      "loss": 0.2451,
+      "step": 14288
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2413845852166109e-05,
+      "loss": 0.2318,
+      "step": 14289
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2412878614829899e-05,
+      "loss": 0.2317,
+      "step": 14290
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.241191135352398e-05,
+      "loss": 0.2467,
+      "step": 14291
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2410944068257959e-05,
+      "loss": 0.2263,
+      "step": 14292
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2409976759041444e-05,
+      "loss": 0.3087,
+      "step": 14293
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2409009425884046e-05,
+      "loss": 0.2395,
+      "step": 14294
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2408042068795373e-05,
+      "loss": 0.2479,
+      "step": 14295
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2407074687785036e-05,
+      "loss": 0.2854,
+      "step": 14296
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240610728286264e-05,
+      "loss": 0.2501,
+      "step": 14297
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2405139854037807e-05,
+      "loss": 0.2971,
+      "step": 14298
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2404172401320135e-05,
+      "loss": 0.198,
+      "step": 14299
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240320492471924e-05,
+      "loss": 0.1865,
+      "step": 14300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2402237424244736e-05,
+      "loss": 0.3082,
+      "step": 14301
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2401269899906229e-05,
+      "loss": 0.2662,
+      "step": 14302
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2400302351713336e-05,
+      "loss": 0.2446,
+      "step": 14303
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.239933477967566e-05,
+      "loss": 0.245,
+      "step": 14304
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2398367183802822e-05,
+      "loss": 0.1782,
+      "step": 14305
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2397399564104429e-05,
+      "loss": 0.2471,
+      "step": 14306
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2396431920590095e-05,
+      "loss": 0.2198,
+      "step": 14307
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2395464253269434e-05,
+      "loss": 0.2097,
+      "step": 14308
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2394496562152054e-05,
+      "loss": 0.1691,
+      "step": 14309
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2393528847247573e-05,
+      "loss": 0.2417,
+      "step": 14310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2392561108565604e-05,
+      "loss": 0.2122,
+      "step": 14311
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2391593346115756e-05,
+      "loss": 0.2872,
+      "step": 14312
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.239062555990765e-05,
+      "loss": 0.2317,
+      "step": 14313
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2389657749950895e-05,
+      "loss": 0.3015,
+      "step": 14314
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2388689916255106e-05,
+      "loss": 0.2169,
+      "step": 14315
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.23877220588299e-05,
+      "loss": 0.2084,
+      "step": 14316
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2386754177684885e-05,
+      "loss": 0.1909,
+      "step": 14317
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2385786272829686e-05,
+      "loss": 0.2792,
+      "step": 14318
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.238481834427391e-05,
+      "loss": 0.2643,
+      "step": 14319
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2383850392027178e-05,
+      "loss": 0.2342,
+      "step": 14320
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.23828824160991e-05,
+      "loss": 0.2247,
+      "step": 14321
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.23819144164993e-05,
+      "loss": 0.2487,
+      "step": 14322
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2380946393237385e-05,
+      "loss": 0.2314,
+      "step": 14323
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2379978346322978e-05,
+      "loss": 0.2638,
+      "step": 14324
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2379010275765693e-05,
+      "loss": 0.2069,
+      "step": 14325
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2378042181575145e-05,
+      "loss": 0.237,
+      "step": 14326
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2377074063760957e-05,
+      "loss": 0.2476,
+      "step": 14327
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.237610592233274e-05,
+      "loss": 0.2533,
+      "step": 14328
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2375137757300116e-05,
+      "loss": 0.2561,
+      "step": 14329
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2374169568672702e-05,
+      "loss": 0.1829,
+      "step": 14330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2373201356460112e-05,
+      "loss": 0.2155,
+      "step": 14331
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2372233120671973e-05,
+      "loss": 0.2527,
+      "step": 14332
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2371264861317893e-05,
+      "loss": 0.2483,
+      "step": 14333
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2370296578407495e-05,
+      "loss": 0.2367,
+      "step": 14334
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2369328271950404e-05,
+      "loss": 0.1646,
+      "step": 14335
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2368359941956231e-05,
+      "loss": 0.2759,
+      "step": 14336
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2367391588434598e-05,
+      "loss": 0.2424,
+      "step": 14337
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2366423211395126e-05,
+      "loss": 0.1949,
+      "step": 14338
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2365454810847437e-05,
+      "loss": 0.2309,
+      "step": 14339
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2364486386801144e-05,
+      "loss": 0.2939,
+      "step": 14340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2363517939265873e-05,
+      "loss": 0.2989,
+      "step": 14341
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2362549468251246e-05,
+      "loss": 0.1731,
+      "step": 14342
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2361580973766876e-05,
+      "loss": 0.2189,
+      "step": 14343
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2360612455822394e-05,
+      "loss": 0.2135,
+      "step": 14344
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2359643914427415e-05,
+      "loss": 0.2388,
+      "step": 14345
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2358675349591559e-05,
+      "loss": 0.2427,
+      "step": 14346
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2357706761324455e-05,
+      "loss": 0.1793,
+      "step": 14347
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2356738149635721e-05,
+      "loss": 0.262,
+      "step": 14348
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2355769514534976e-05,
+      "loss": 0.2467,
+      "step": 14349
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2354800856031848e-05,
+      "loss": 0.2374,
+      "step": 14350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2353832174135958e-05,
+      "loss": 0.2233,
+      "step": 14351
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2352863468856926e-05,
+      "loss": 0.2436,
+      "step": 14352
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2351894740204379e-05,
+      "loss": 0.2515,
+      "step": 14353
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2350925988187941e-05,
+      "loss": 0.2293,
+      "step": 14354
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2349957212817229e-05,
+      "loss": 0.232,
+      "step": 14355
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2348988414101877e-05,
+      "loss": 0.2771,
+      "step": 14356
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2348019592051501e-05,
+      "loss": 0.2914,
+      "step": 14357
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2347050746675729e-05,
+      "loss": 0.2991,
+      "step": 14358
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2346081877984183e-05,
+      "loss": 0.2631,
+      "step": 14359
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2345112985986489e-05,
+      "loss": 0.2792,
+      "step": 14360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2344144070692275e-05,
+      "loss": 0.2403,
+      "step": 14361
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.234317513211116e-05,
+      "loss": 0.2171,
+      "step": 14362
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2342206170252777e-05,
+      "loss": 0.1951,
+      "step": 14363
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2341237185126746e-05,
+      "loss": 0.2621,
+      "step": 14364
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2340268176742696e-05,
+      "loss": 0.2574,
+      "step": 14365
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2339299145110254e-05,
+      "loss": 0.1855,
+      "step": 14366
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2338330090239042e-05,
+      "loss": 0.2598,
+      "step": 14367
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2337361012138688e-05,
+      "loss": 0.2311,
+      "step": 14368
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.233639191081882e-05,
+      "loss": 0.2943,
+      "step": 14369
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2335422786289068e-05,
+      "loss": 0.2787,
+      "step": 14370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2334453638559057e-05,
+      "loss": 0.3023,
+      "step": 14371
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.233348446763841e-05,
+      "loss": 0.2593,
+      "step": 14372
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2332515273536763e-05,
+      "loss": 0.2456,
+      "step": 14373
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.233154605626374e-05,
+      "loss": 0.2932,
+      "step": 14374
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2330576815828964e-05,
+      "loss": 0.2186,
+      "step": 14375
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2329607552242075e-05,
+      "loss": 0.1998,
+      "step": 14376
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2328638265512694e-05,
+      "loss": 0.2504,
+      "step": 14377
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2327668955650452e-05,
+      "loss": 0.2419,
+      "step": 14378
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2326699622664977e-05,
+      "loss": 0.2107,
+      "step": 14379
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2325730266565898e-05,
+      "loss": 0.2442,
+      "step": 14380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2324760887362845e-05,
+      "loss": 0.2326,
+      "step": 14381
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2323791485065451e-05,
+      "loss": 0.2176,
+      "step": 14382
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2322822059683344e-05,
+      "loss": 0.2672,
+      "step": 14383
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2321852611226153e-05,
+      "loss": 0.2496,
+      "step": 14384
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2320883139703512e-05,
+      "loss": 0.2184,
+      "step": 14385
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2319913645125044e-05,
+      "loss": 0.1852,
+      "step": 14386
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.231894412750039e-05,
+      "loss": 0.2195,
+      "step": 14387
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2317974586839176e-05,
+      "loss": 0.2143,
+      "step": 14388
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2317005023151031e-05,
+      "loss": 0.2454,
+      "step": 14389
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2316035436445597e-05,
+      "loss": 0.2628,
+      "step": 14390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2315065826732496e-05,
+      "loss": 0.1888,
+      "step": 14391
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2314096194021359e-05,
+      "loss": 0.2578,
+      "step": 14392
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2313126538321827e-05,
+      "loss": 0.1968,
+      "step": 14393
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2312156859643526e-05,
+      "loss": 0.2579,
+      "step": 14394
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2311187157996091e-05,
+      "loss": 0.2954,
+      "step": 14395
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2310217433389155e-05,
+      "loss": 0.3012,
+      "step": 14396
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2309247685832355e-05,
+      "loss": 0.2657,
+      "step": 14397
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2308277915335317e-05,
+      "loss": 0.243,
+      "step": 14398
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2307308121907679e-05,
+      "loss": 0.2155,
+      "step": 14399
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2306338305559079e-05,
+      "loss": 0.2436,
+      "step": 14400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2305368466299142e-05,
+      "loss": 0.267,
+      "step": 14401
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2304398604137511e-05,
+      "loss": 0.2341,
+      "step": 14402
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2303428719083814e-05,
+      "loss": 0.2116,
+      "step": 14403
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2302458811147688e-05,
+      "loss": 0.2063,
+      "step": 14404
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2301488880338772e-05,
+      "loss": 0.2711,
+      "step": 14405
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2300518926666698e-05,
+      "loss": 0.2472,
+      "step": 14406
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2299548950141102e-05,
+      "loss": 0.2399,
+      "step": 14407
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2298578950771618e-05,
+      "loss": 0.2584,
+      "step": 14408
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2297608928567887e-05,
+      "loss": 0.1951,
+      "step": 14409
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2296638883539543e-05,
+      "loss": 0.257,
+      "step": 14410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2295668815696215e-05,
+      "loss": 0.2073,
+      "step": 14411
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2294698725047556e-05,
+      "loss": 0.1658,
+      "step": 14412
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2293728611603186e-05,
+      "loss": 0.186,
+      "step": 14413
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2292758475372753e-05,
+      "loss": 0.2577,
+      "step": 14414
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229178831636589e-05,
+      "loss": 0.2191,
+      "step": 14415
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2290818134592234e-05,
+      "loss": 0.2673,
+      "step": 14416
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2289847930061426e-05,
+      "loss": 0.2319,
+      "step": 14417
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2288877702783105e-05,
+      "loss": 0.2189,
+      "step": 14418
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2287907452766905e-05,
+      "loss": 0.2618,
+      "step": 14419
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2286937180022462e-05,
+      "loss": 0.2207,
+      "step": 14420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2285966884559427e-05,
+      "loss": 0.2538,
+      "step": 14421
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2284996566387426e-05,
+      "loss": 0.2233,
+      "step": 14422
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22840262255161e-05,
+      "loss": 0.1969,
+      "step": 14423
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22830558619551e-05,
+      "loss": 0.2001,
+      "step": 14424
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2282085475714052e-05,
+      "loss": 0.2677,
+      "step": 14425
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2281115066802603e-05,
+      "loss": 0.2325,
+      "step": 14426
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.228014463523039e-05,
+      "loss": 0.2267,
+      "step": 14427
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2279174181007056e-05,
+      "loss": 0.272,
+      "step": 14428
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2278203704142242e-05,
+      "loss": 0.2535,
+      "step": 14429
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2277233204645584e-05,
+      "loss": 0.235,
+      "step": 14430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2276262682526729e-05,
+      "loss": 0.2535,
+      "step": 14431
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2275292137795314e-05,
+      "loss": 0.2258,
+      "step": 14432
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.227432157046098e-05,
+      "loss": 0.2038,
+      "step": 14433
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2273350980533375e-05,
+      "loss": 0.2833,
+      "step": 14434
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2272380368022133e-05,
+      "loss": 0.2063,
+      "step": 14435
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2271409732936902e-05,
+      "loss": 0.2712,
+      "step": 14436
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.227043907528732e-05,
+      "loss": 0.2737,
+      "step": 14437
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226946839508303e-05,
+      "loss": 0.2308,
+      "step": 14438
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2268497692333678e-05,
+      "loss": 0.2416,
+      "step": 14439
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2267526967048907e-05,
+      "loss": 0.2321,
+      "step": 14440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2266556219238354e-05,
+      "loss": 0.2464,
+      "step": 14441
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2265585448911672e-05,
+      "loss": 0.1707,
+      "step": 14442
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22646146560785e-05,
+      "loss": 0.2497,
+      "step": 14443
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2263643840748476e-05,
+      "loss": 0.2494,
+      "step": 14444
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2262673002931254e-05,
+      "loss": 0.1768,
+      "step": 14445
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2261702142636476e-05,
+      "loss": 0.1999,
+      "step": 14446
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2260731259873782e-05,
+      "loss": 0.3097,
+      "step": 14447
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2259760354652818e-05,
+      "loss": 0.2113,
+      "step": 14448
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2258789426983232e-05,
+      "loss": 0.2065,
+      "step": 14449
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.225781847687467e-05,
+      "loss": 0.2502,
+      "step": 14450
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2256847504336776e-05,
+      "loss": 0.2365,
+      "step": 14451
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2255876509379192e-05,
+      "loss": 0.1804,
+      "step": 14452
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2254905492011568e-05,
+      "loss": 0.2255,
+      "step": 14453
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.225393445224355e-05,
+      "loss": 0.2371,
+      "step": 14454
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2252963390084784e-05,
+      "loss": 0.1928,
+      "step": 14455
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2251992305544915e-05,
+      "loss": 0.1871,
+      "step": 14456
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2251021198633591e-05,
+      "loss": 0.2546,
+      "step": 14457
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2250050069360462e-05,
+      "loss": 0.2626,
+      "step": 14458
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.224907891773517e-05,
+      "loss": 0.2722,
+      "step": 14459
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2248107743767364e-05,
+      "loss": 0.2541,
+      "step": 14460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2247136547466693e-05,
+      "loss": 0.2752,
+      "step": 14461
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2246165328842806e-05,
+      "loss": 0.2449,
+      "step": 14462
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2245194087905348e-05,
+      "loss": 0.2508,
+      "step": 14463
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224422282466397e-05,
+      "loss": 0.256,
+      "step": 14464
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224325153912832e-05,
+      "loss": 0.1876,
+      "step": 14465
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2242280231308043e-05,
+      "loss": 0.2417,
+      "step": 14466
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2241308901212798e-05,
+      "loss": 0.2703,
+      "step": 14467
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2240337548852223e-05,
+      "loss": 0.1612,
+      "step": 14468
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2239366174235972e-05,
+      "loss": 0.2242,
+      "step": 14469
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.22383947773737e-05,
+      "loss": 0.1881,
+      "step": 14470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2237423358275047e-05,
+      "loss": 0.2351,
+      "step": 14471
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2236451916949667e-05,
+      "loss": 0.2867,
+      "step": 14472
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2235480453407216e-05,
+      "loss": 0.223,
+      "step": 14473
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2234508967657337e-05,
+      "loss": 0.1983,
+      "step": 14474
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2233537459709683e-05,
+      "loss": 0.196,
+      "step": 14475
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2232565929573907e-05,
+      "loss": 0.2409,
+      "step": 14476
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.223159437725966e-05,
+      "loss": 0.2216,
+      "step": 14477
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.223062280277659e-05,
+      "loss": 0.2733,
+      "step": 14478
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2229651206134357e-05,
+      "loss": 0.2251,
+      "step": 14479
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2228679587342603e-05,
+      "loss": 0.2524,
+      "step": 14480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2227707946410983e-05,
+      "loss": 0.2528,
+      "step": 14481
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2226736283349153e-05,
+      "loss": 0.2441,
+      "step": 14482
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2225764598166763e-05,
+      "loss": 0.2401,
+      "step": 14483
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2224792890873465e-05,
+      "loss": 0.2487,
+      "step": 14484
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2223821161478912e-05,
+      "loss": 0.2463,
+      "step": 14485
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.222284940999276e-05,
+      "loss": 0.265,
+      "step": 14486
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2221877636424662e-05,
+      "loss": 0.1893,
+      "step": 14487
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2220905840784267e-05,
+      "loss": 0.206,
+      "step": 14488
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2219934023081237e-05,
+      "loss": 0.253,
+      "step": 14489
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2218962183325217e-05,
+      "loss": 0.2554,
+      "step": 14490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2217990321525868e-05,
+      "loss": 0.1825,
+      "step": 14491
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.221701843769284e-05,
+      "loss": 0.2555,
+      "step": 14492
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2216046531835793e-05,
+      "loss": 0.2561,
+      "step": 14493
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2215074603964376e-05,
+      "loss": 0.243,
+      "step": 14494
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2214102654088248e-05,
+      "loss": 0.2493,
+      "step": 14495
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2213130682217065e-05,
+      "loss": 0.2859,
+      "step": 14496
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2212158688360481e-05,
+      "loss": 0.1976,
+      "step": 14497
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.221118667252815e-05,
+      "loss": 0.2414,
+      "step": 14498
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2210214634729733e-05,
+      "loss": 0.2308,
+      "step": 14499
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.220924257497488e-05,
+      "loss": 0.2936,
+      "step": 14500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2208270493273255e-05,
+      "loss": 0.2518,
+      "step": 14501
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2207298389634507e-05,
+      "loss": 0.2581,
+      "step": 14502
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2206326264068298e-05,
+      "loss": 0.2279,
+      "step": 14503
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2205354116584284e-05,
+      "loss": 0.2475,
+      "step": 14504
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2204381947192123e-05,
+      "loss": 0.2264,
+      "step": 14505
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2203409755901472e-05,
+      "loss": 0.2312,
+      "step": 14506
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2202437542721988e-05,
+      "loss": 0.252,
+      "step": 14507
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2201465307663331e-05,
+      "loss": 0.2408,
+      "step": 14508
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2200493050735155e-05,
+      "loss": 0.3039,
+      "step": 14509
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2199520771947122e-05,
+      "loss": 0.2229,
+      "step": 14510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2198548471308893e-05,
+      "loss": 0.21,
+      "step": 14511
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2197576148830119e-05,
+      "loss": 0.216,
+      "step": 14512
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2196603804520468e-05,
+      "loss": 0.2111,
+      "step": 14513
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2195631438389595e-05,
+      "loss": 0.2095,
+      "step": 14514
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2194659050447158e-05,
+      "loss": 0.211,
+      "step": 14515
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2193686640702823e-05,
+      "loss": 0.2036,
+      "step": 14516
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.219271420916624e-05,
+      "loss": 0.2539,
+      "step": 14517
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.219174175584708e-05,
+      "loss": 0.2627,
+      "step": 14518
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2190769280754995e-05,
+      "loss": 0.3109,
+      "step": 14519
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2189796783899652e-05,
+      "loss": 0.2715,
+      "step": 14520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2188824265290707e-05,
+      "loss": 0.2903,
+      "step": 14521
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.218785172493782e-05,
+      "loss": 0.2565,
+      "step": 14522
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2186879162850662e-05,
+      "loss": 0.2212,
+      "step": 14523
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2185906579038881e-05,
+      "loss": 0.2678,
+      "step": 14524
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2184933973512152e-05,
+      "loss": 0.1941,
+      "step": 14525
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2183961346280127e-05,
+      "loss": 0.2901,
+      "step": 14526
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.218298869735247e-05,
+      "loss": 0.2168,
+      "step": 14527
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2182016026738845e-05,
+      "loss": 0.2843,
+      "step": 14528
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2181043334448916e-05,
+      "loss": 0.2062,
+      "step": 14529
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2180070620492346e-05,
+      "loss": 0.2489,
+      "step": 14530
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2179097884878793e-05,
+      "loss": 0.2603,
+      "step": 14531
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2178125127617925e-05,
+      "loss": 0.2843,
+      "step": 14532
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2177152348719403e-05,
+      "loss": 0.2936,
+      "step": 14533
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2176179548192893e-05,
+      "loss": 0.217,
+      "step": 14534
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2175206726048058e-05,
+      "loss": 0.2314,
+      "step": 14535
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2174233882294561e-05,
+      "loss": 0.1826,
+      "step": 14536
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2173261016942066e-05,
+      "loss": 0.2162,
+      "step": 14537
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.217228813000024e-05,
+      "loss": 0.2537,
+      "step": 14538
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2171315221478745e-05,
+      "loss": 0.1661,
+      "step": 14539
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2170342291387248e-05,
+      "loss": 0.2499,
+      "step": 14540
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.216936933973541e-05,
+      "loss": 0.1854,
+      "step": 14541
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2168396366532902e-05,
+      "loss": 0.2792,
+      "step": 14542
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2167423371789387e-05,
+      "loss": 0.1853,
+      "step": 14543
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2166450355514532e-05,
+      "loss": 0.2109,
+      "step": 14544
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2165477317718002e-05,
+      "loss": 0.2911,
+      "step": 14545
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.216450425840946e-05,
+      "loss": 0.2659,
+      "step": 14546
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.216353117759858e-05,
+      "loss": 0.2341,
+      "step": 14547
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2162558075295021e-05,
+      "loss": 0.2329,
+      "step": 14548
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2161584951508454e-05,
+      "loss": 0.2183,
+      "step": 14549
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2160611806248545e-05,
+      "loss": 0.2457,
+      "step": 14550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2159638639524964e-05,
+      "loss": 0.2404,
+      "step": 14551
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2158665451347373e-05,
+      "loss": 0.1642,
+      "step": 14552
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2157692241725445e-05,
+      "loss": 0.264,
+      "step": 14553
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2156719010668848e-05,
+      "loss": 0.2119,
+      "step": 14554
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.215574575818724e-05,
+      "loss": 0.2553,
+      "step": 14555
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2154772484290303e-05,
+      "loss": 0.253,
+      "step": 14556
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.21537991889877e-05,
+      "loss": 0.23,
+      "step": 14557
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2152825872289096e-05,
+      "loss": 0.267,
+      "step": 14558
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2151852534204168e-05,
+      "loss": 0.2116,
+      "step": 14559
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2150879174742579e-05,
+      "loss": 0.2648,
+      "step": 14560
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2149905793914e-05,
+      "loss": 0.2821,
+      "step": 14561
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.21489323917281e-05,
+      "loss": 0.3074,
+      "step": 14562
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.214795896819455e-05,
+      "loss": 0.2466,
+      "step": 14563
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2146985523323021e-05,
+      "loss": 0.2805,
+      "step": 14564
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.214601205712318e-05,
+      "loss": 0.2315,
+      "step": 14565
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2145038569604704e-05,
+      "loss": 0.2552,
+      "step": 14566
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2144065060777255e-05,
+      "loss": 0.2,
+      "step": 14567
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2143091530650508e-05,
+      "loss": 0.2866,
+      "step": 14568
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2142117979234138e-05,
+      "loss": 0.3062,
+      "step": 14569
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2141144406537808e-05,
+      "loss": 0.2858,
+      "step": 14570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2140170812571197e-05,
+      "loss": 0.2463,
+      "step": 14571
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2139197197343973e-05,
+      "loss": 0.2511,
+      "step": 14572
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.213822356086581e-05,
+      "loss": 0.2552,
+      "step": 14573
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2137249903146378e-05,
+      "loss": 0.2494,
+      "step": 14574
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2136276224195349e-05,
+      "loss": 0.247,
+      "step": 14575
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2135302524022399e-05,
+      "loss": 0.208,
+      "step": 14576
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.21343288026372e-05,
+      "loss": 0.1762,
+      "step": 14577
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2133355060049422e-05,
+      "loss": 0.2382,
+      "step": 14578
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2132381296268739e-05,
+      "loss": 0.2677,
+      "step": 14579
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2131407511304825e-05,
+      "loss": 0.2186,
+      "step": 14580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2130433705167356e-05,
+      "loss": 0.21,
+      "step": 14581
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2129459877866006e-05,
+      "loss": 0.2328,
+      "step": 14582
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2128486029410445e-05,
+      "loss": 0.2304,
+      "step": 14583
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2127512159810348e-05,
+      "loss": 0.236,
+      "step": 14584
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2126538269075392e-05,
+      "loss": 0.2593,
+      "step": 14585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2125564357215249e-05,
+      "loss": 0.273,
+      "step": 14586
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2124590424239598e-05,
+      "loss": 0.2086,
+      "step": 14587
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2123616470158113e-05,
+      "loss": 0.288,
+      "step": 14588
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2122642494980463e-05,
+      "loss": 0.2575,
+      "step": 14589
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2121668498716332e-05,
+      "loss": 0.2176,
+      "step": 14590
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2120694481375393e-05,
+      "loss": 0.2297,
+      "step": 14591
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2119720442967315e-05,
+      "loss": 0.2384,
+      "step": 14592
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2118746383501787e-05,
+      "loss": 0.2385,
+      "step": 14593
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2117772302988477e-05,
+      "loss": 0.2627,
+      "step": 14594
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2116798201437061e-05,
+      "loss": 0.2236,
+      "step": 14595
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2115824078857221e-05,
+      "loss": 0.2559,
+      "step": 14596
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.211484993525863e-05,
+      "loss": 0.2337,
+      "step": 14597
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2113875770650964e-05,
+      "loss": 0.2415,
+      "step": 14598
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2112901585043907e-05,
+      "loss": 0.1963,
+      "step": 14599
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2111927378447132e-05,
+      "loss": 0.2328,
+      "step": 14600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2110953150870313e-05,
+      "loss": 0.2142,
+      "step": 14601
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2109978902323134e-05,
+      "loss": 0.257,
+      "step": 14602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2109004632815275e-05,
+      "loss": 0.2181,
+      "step": 14603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2108030342356404e-05,
+      "loss": 0.1653,
+      "step": 14604
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2107056030956214e-05,
+      "loss": 0.2212,
+      "step": 14605
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2106081698624375e-05,
+      "loss": 0.2529,
+      "step": 14606
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2105107345370564e-05,
+      "loss": 0.1976,
+      "step": 14607
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2104132971204465e-05,
+      "loss": 0.2643,
+      "step": 14608
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2103158576135759e-05,
+      "loss": 0.2735,
+      "step": 14609
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2102184160174122e-05,
+      "loss": 0.1916,
+      "step": 14610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2101209723329233e-05,
+      "loss": 0.2677,
+      "step": 14611
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2100235265610778e-05,
+      "loss": 0.1988,
+      "step": 14612
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2099260787028429e-05,
+      "loss": 0.2571,
+      "step": 14613
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2098286287591874e-05,
+      "loss": 0.285,
+      "step": 14614
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2097311767310791e-05,
+      "loss": 0.1879,
+      "step": 14615
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2096337226194859e-05,
+      "loss": 0.2257,
+      "step": 14616
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2095362664253763e-05,
+      "loss": 0.2555,
+      "step": 14617
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.209438808149718e-05,
+      "loss": 0.1894,
+      "step": 14618
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2093413477934794e-05,
+      "loss": 0.1984,
+      "step": 14619
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2092438853576287e-05,
+      "loss": 0.2509,
+      "step": 14620
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2091464208431341e-05,
+      "loss": 0.3066,
+      "step": 14621
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2090489542509638e-05,
+      "loss": 0.2433,
+      "step": 14622
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2089514855820862e-05,
+      "loss": 0.301,
+      "step": 14623
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2088540148374693e-05,
+      "loss": 0.2313,
+      "step": 14624
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.208756542018081e-05,
+      "loss": 0.2545,
+      "step": 14625
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2086590671248904e-05,
+      "loss": 0.254,
+      "step": 14626
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2085615901588656e-05,
+      "loss": 0.2544,
+      "step": 14627
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2084641111209747e-05,
+      "loss": 0.2438,
+      "step": 14628
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2083666300121864e-05,
+      "loss": 0.2592,
+      "step": 14629
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2082691468334687e-05,
+      "loss": 0.2958,
+      "step": 14630
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2081716615857902e-05,
+      "loss": 0.2114,
+      "step": 14631
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2080741742701192e-05,
+      "loss": 0.2007,
+      "step": 14632
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2079766848874244e-05,
+      "loss": 0.2311,
+      "step": 14633
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2078791934386743e-05,
+      "loss": 0.3129,
+      "step": 14634
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2077816999248366e-05,
+      "loss": 0.238,
+      "step": 14635
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.207684204346881e-05,
+      "loss": 0.1976,
+      "step": 14636
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2075867067057754e-05,
+      "loss": 0.2459,
+      "step": 14637
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.207489207002488e-05,
+      "loss": 0.2728,
+      "step": 14638
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2073917052379882e-05,
+      "loss": 0.2866,
+      "step": 14639
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.207294201413244e-05,
+      "loss": 0.2249,
+      "step": 14640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2071966955292243e-05,
+      "loss": 0.2069,
+      "step": 14641
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2070991875868972e-05,
+      "loss": 0.2361,
+      "step": 14642
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.207001677587232e-05,
+      "loss": 0.2878,
+      "step": 14643
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.206904165531197e-05,
+      "loss": 0.2384,
+      "step": 14644
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2068066514197611e-05,
+      "loss": 0.2392,
+      "step": 14645
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2067091352538933e-05,
+      "loss": 0.2191,
+      "step": 14646
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2066116170345614e-05,
+      "loss": 0.2233,
+      "step": 14647
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2065140967627349e-05,
+      "loss": 0.2385,
+      "step": 14648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2064165744393826e-05,
+      "loss": 0.2593,
+      "step": 14649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2063190500654725e-05,
+      "loss": 0.2876,
+      "step": 14650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2062215236419746e-05,
+      "loss": 0.2744,
+      "step": 14651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2061239951698566e-05,
+      "loss": 0.2001,
+      "step": 14652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2060264646500884e-05,
+      "loss": 0.2689,
+      "step": 14653
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.205928932083638e-05,
+      "loss": 0.2749,
+      "step": 14654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2058313974714746e-05,
+      "loss": 0.24,
+      "step": 14655
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2057338608145673e-05,
+      "loss": 0.2972,
+      "step": 14656
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205636322113885e-05,
+      "loss": 0.2697,
+      "step": 14657
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2055387813703965e-05,
+      "loss": 0.2539,
+      "step": 14658
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205441238585071e-05,
+      "loss": 0.2411,
+      "step": 14659
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205343693758877e-05,
+      "loss": 0.2267,
+      "step": 14660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2052461468927838e-05,
+      "loss": 0.2793,
+      "step": 14661
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2051485979877606e-05,
+      "loss": 0.2531,
+      "step": 14662
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2050510470447765e-05,
+      "loss": 0.2798,
+      "step": 14663
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2049534940648004e-05,
+      "loss": 0.2817,
+      "step": 14664
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2048559390488013e-05,
+      "loss": 0.2422,
+      "step": 14665
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2047583819977485e-05,
+      "loss": 0.2226,
+      "step": 14666
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2046608229126111e-05,
+      "loss": 0.251,
+      "step": 14667
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2045632617943584e-05,
+      "loss": 0.2692,
+      "step": 14668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.204465698643959e-05,
+      "loss": 0.2713,
+      "step": 14669
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2043681334623826e-05,
+      "loss": 0.2676,
+      "step": 14670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2042705662505987e-05,
+      "loss": 0.2135,
+      "step": 14671
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2041729970095757e-05,
+      "loss": 0.2353,
+      "step": 14672
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2040754257402837e-05,
+      "loss": 0.259,
+      "step": 14673
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2039778524436916e-05,
+      "loss": 0.2334,
+      "step": 14674
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2038802771207686e-05,
+      "loss": 0.2698,
+      "step": 14675
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.203782699772484e-05,
+      "loss": 0.264,
+      "step": 14676
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2036851203998076e-05,
+      "loss": 0.2185,
+      "step": 14677
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2035875390037082e-05,
+      "loss": 0.238,
+      "step": 14678
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2034899555851554e-05,
+      "loss": 0.2774,
+      "step": 14679
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2033923701451189e-05,
+      "loss": 0.2031,
+      "step": 14680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2032947826845674e-05,
+      "loss": 0.2262,
+      "step": 14681
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2031971932044711e-05,
+      "loss": 0.2546,
+      "step": 14682
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.203099601705799e-05,
+      "loss": 0.2324,
+      "step": 14683
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2030020081895206e-05,
+      "loss": 0.286,
+      "step": 14684
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2029044126566058e-05,
+      "loss": 0.3059,
+      "step": 14685
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2028068151080239e-05,
+      "loss": 0.2484,
+      "step": 14686
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2027092155447441e-05,
+      "loss": 0.214,
+      "step": 14687
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2026116139677365e-05,
+      "loss": 0.2776,
+      "step": 14688
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2025140103779704e-05,
+      "loss": 0.2749,
+      "step": 14689
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2024164047764151e-05,
+      "loss": 0.2684,
+      "step": 14690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2023187971640406e-05,
+      "loss": 0.1948,
+      "step": 14691
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.202221187541817e-05,
+      "loss": 0.2423,
+      "step": 14692
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2021235759107128e-05,
+      "loss": 0.2104,
+      "step": 14693
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2020259622716988e-05,
+      "loss": 0.3185,
+      "step": 14694
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2019283466257441e-05,
+      "loss": 0.2839,
+      "step": 14695
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2018307289738184e-05,
+      "loss": 0.3231,
+      "step": 14696
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2017331093168917e-05,
+      "loss": 0.2372,
+      "step": 14697
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.201635487655934e-05,
+      "loss": 0.2431,
+      "step": 14698
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2015378639919141e-05,
+      "loss": 0.2449,
+      "step": 14699
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.201440238325803e-05,
+      "loss": 0.1984,
+      "step": 14700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2013426106585697e-05,
+      "loss": 0.2209,
+      "step": 14701
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2012449809911841e-05,
+      "loss": 0.2716,
+      "step": 14702
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2011473493246167e-05,
+      "loss": 0.2824,
+      "step": 14703
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2010497156598369e-05,
+      "loss": 0.266,
+      "step": 14704
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2009520799978143e-05,
+      "loss": 0.2499,
+      "step": 14705
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2008544423395194e-05,
+      "loss": 0.264,
+      "step": 14706
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2007568026859217e-05,
+      "loss": 0.2502,
+      "step": 14707
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2006591610379914e-05,
+      "loss": 0.2192,
+      "step": 14708
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2005615173966984e-05,
+      "loss": 0.2575,
+      "step": 14709
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.200463871763013e-05,
+      "loss": 0.205,
+      "step": 14710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2003662241379048e-05,
+      "loss": 0.2735,
+      "step": 14711
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.200268574522344e-05,
+      "loss": 0.2211,
+      "step": 14712
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2001709229173006e-05,
+      "loss": 0.2207,
+      "step": 14713
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2000732693237449e-05,
+      "loss": 0.2443,
+      "step": 14714
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1999756137426465e-05,
+      "loss": 0.2288,
+      "step": 14715
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1998779561749762e-05,
+      "loss": 0.2356,
+      "step": 14716
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1997802966217037e-05,
+      "loss": 0.322,
+      "step": 14717
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1996826350837992e-05,
+      "loss": 0.2692,
+      "step": 14718
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1995849715622328e-05,
+      "loss": 0.2382,
+      "step": 14719
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1994873060579748e-05,
+      "loss": 0.2696,
+      "step": 14720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1993896385719954e-05,
+      "loss": 0.2243,
+      "step": 14721
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1992919691052652e-05,
+      "loss": 0.2369,
+      "step": 14722
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.199194297658754e-05,
+      "loss": 0.2188,
+      "step": 14723
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1990966242334318e-05,
+      "loss": 0.2667,
+      "step": 14724
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1989989488302695e-05,
+      "loss": 0.2745,
+      "step": 14725
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1989012714502377e-05,
+      "loss": 0.2309,
+      "step": 14726
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1988035920943055e-05,
+      "loss": 0.1925,
+      "step": 14727
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1987059107634447e-05,
+      "loss": 0.2437,
+      "step": 14728
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1986082274586244e-05,
+      "loss": 0.207,
+      "step": 14729
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1985105421808159e-05,
+      "loss": 0.2479,
+      "step": 14730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.198412854930989e-05,
+      "loss": 0.1782,
+      "step": 14731
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1983151657101144e-05,
+      "loss": 0.3406,
+      "step": 14732
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1982174745191629e-05,
+      "loss": 0.2547,
+      "step": 14733
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1981197813591043e-05,
+      "loss": 0.2648,
+      "step": 14734
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1980220862309097e-05,
+      "loss": 0.2206,
+      "step": 14735
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.197924389135549e-05,
+      "loss": 0.2337,
+      "step": 14736
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1978266900739932e-05,
+      "loss": 0.2234,
+      "step": 14737
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1977289890472129e-05,
+      "loss": 0.2704,
+      "step": 14738
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1976312860561783e-05,
+      "loss": 0.1859,
+      "step": 14739
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.19753358110186e-05,
+      "loss": 0.2188,
+      "step": 14740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.197435874185229e-05,
+      "loss": 0.2458,
+      "step": 14741
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1973381653072558e-05,
+      "loss": 0.2907,
+      "step": 14742
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1972404544689107e-05,
+      "loss": 0.2666,
+      "step": 14743
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1971427416711646e-05,
+      "loss": 0.2913,
+      "step": 14744
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1970450269149882e-05,
+      "loss": 0.3142,
+      "step": 14745
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1969473102013525e-05,
+      "loss": 0.2527,
+      "step": 14746
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1968495915312279e-05,
+      "loss": 0.2145,
+      "step": 14747
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.196751870905585e-05,
+      "loss": 0.2574,
+      "step": 14748
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1966541483253947e-05,
+      "loss": 0.2294,
+      "step": 14749
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1965564237916279e-05,
+      "loss": 0.2457,
+      "step": 14750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1964586973052551e-05,
+      "loss": 0.2759,
+      "step": 14751
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1963609688672477e-05,
+      "loss": 0.259,
+      "step": 14752
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1962632384785762e-05,
+      "loss": 0.2413,
+      "step": 14753
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.196165506140211e-05,
+      "loss": 0.2383,
+      "step": 14754
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.196067771853124e-05,
+      "loss": 0.2741,
+      "step": 14755
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.195970035618285e-05,
+      "loss": 0.2598,
+      "step": 14756
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1958722974366661e-05,
+      "loss": 0.2289,
+      "step": 14757
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.195774557309237e-05,
+      "loss": 0.281,
+      "step": 14758
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1956768152369695e-05,
+      "loss": 0.1989,
+      "step": 14759
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1955790712208344e-05,
+      "loss": 0.2439,
+      "step": 14760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.195481325261802e-05,
+      "loss": 0.2995,
+      "step": 14761
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1953835773608448e-05,
+      "loss": 0.2508,
+      "step": 14762
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1952858275189324e-05,
+      "loss": 0.2758,
+      "step": 14763
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1951880757370366e-05,
+      "loss": 0.2154,
+      "step": 14764
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.2576,
+      "step": 14765
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1949925663571786e-05,
+      "loss": 0.2666,
+      "step": 14766
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1948948087611587e-05,
+      "loss": 0.2161,
+      "step": 14767
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1947970492290395e-05,
+      "loss": 0.2972,
+      "step": 14768
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1946992877617924e-05,
+      "loss": 0.2787,
+      "step": 14769
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1946015243603882e-05,
+      "loss": 0.2682,
+      "step": 14770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1945037590257984e-05,
+      "loss": 0.2501,
+      "step": 14771
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1944059917589946e-05,
+      "loss": 0.2285,
+      "step": 14772
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1943082225609468e-05,
+      "loss": 0.2324,
+      "step": 14773
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1942104514326276e-05,
+      "loss": 0.2511,
+      "step": 14774
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1941126783750075e-05,
+      "loss": 0.1757,
+      "step": 14775
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.194014903389058e-05,
+      "loss": 0.1989,
+      "step": 14776
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1939171264757503e-05,
+      "loss": 0.2673,
+      "step": 14777
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1938193476360558e-05,
+      "loss": 0.2911,
+      "step": 14778
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1937215668709458e-05,
+      "loss": 0.2498,
+      "step": 14779
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.193623784181392e-05,
+      "loss": 0.2729,
+      "step": 14780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1935259995683655e-05,
+      "loss": 0.2586,
+      "step": 14781
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1934282130328371e-05,
+      "loss": 0.2122,
+      "step": 14782
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1933304245757791e-05,
+      "loss": 0.2186,
+      "step": 14783
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1932326341981628e-05,
+      "loss": 0.2435,
+      "step": 14784
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1931348419009594e-05,
+      "loss": 0.2536,
+      "step": 14785
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1930370476851406e-05,
+      "loss": 0.2284,
+      "step": 14786
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1929392515516775e-05,
+      "loss": 0.2123,
+      "step": 14787
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.192841453501542e-05,
+      "loss": 0.2141,
+      "step": 14788
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1927436535357056e-05,
+      "loss": 0.2521,
+      "step": 14789
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1926458516551396e-05,
+      "loss": 0.2614,
+      "step": 14790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1925480478608155e-05,
+      "loss": 0.2836,
+      "step": 14791
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1924502421537057e-05,
+      "loss": 0.2306,
+      "step": 14792
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.192352434534781e-05,
+      "loss": 0.2526,
+      "step": 14793
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1922546250050134e-05,
+      "loss": 0.2502,
+      "step": 14794
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1921568135653738e-05,
+      "loss": 0.2626,
+      "step": 14795
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.192059000216835e-05,
+      "loss": 0.2374,
+      "step": 14796
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.191961184960368e-05,
+      "loss": 0.2881,
+      "step": 14797
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1918633677969447e-05,
+      "loss": 0.2528,
+      "step": 14798
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1917655487275367e-05,
+      "loss": 0.2528,
+      "step": 14799
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1916677277531161e-05,
+      "loss": 0.203,
+      "step": 14800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.191569904874654e-05,
+      "loss": 0.2454,
+      "step": 14801
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.191472080093123e-05,
+      "loss": 0.1931,
+      "step": 14802
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1913742534094943e-05,
+      "loss": 0.2862,
+      "step": 14803
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1912764248247397e-05,
+      "loss": 0.2749,
+      "step": 14804
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1911785943398316e-05,
+      "loss": 0.2693,
+      "step": 14805
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1910807619557411e-05,
+      "loss": 0.2489,
+      "step": 14806
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1909829276734404e-05,
+      "loss": 0.2344,
+      "step": 14807
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1908850914939017e-05,
+      "loss": 0.2335,
+      "step": 14808
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1907872534180966e-05,
+      "loss": 0.173,
+      "step": 14809
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1906894134469968e-05,
+      "loss": 0.2926,
+      "step": 14810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.190591571581575e-05,
+      "loss": 0.2433,
+      "step": 14811
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1904937278228024e-05,
+      "loss": 0.2317,
+      "step": 14812
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1903958821716513e-05,
+      "loss": 0.2062,
+      "step": 14813
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1902980346290938e-05,
+      "loss": 0.2622,
+      "step": 14814
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1902001851961019e-05,
+      "loss": 0.2372,
+      "step": 14815
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1901023338736473e-05,
+      "loss": 0.2209,
+      "step": 14816
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1900044806627024e-05,
+      "loss": 0.2719,
+      "step": 14817
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1899066255642393e-05,
+      "loss": 0.2504,
+      "step": 14818
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1898087685792299e-05,
+      "loss": 0.2559,
+      "step": 14819
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1897109097086465e-05,
+      "loss": 0.2136,
+      "step": 14820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1896130489534613e-05,
+      "loss": 0.2025,
+      "step": 14821
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1895151863146462e-05,
+      "loss": 0.2656,
+      "step": 14822
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1894173217931733e-05,
+      "loss": 0.2414,
+      "step": 14823
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1893194553900152e-05,
+      "loss": 0.1835,
+      "step": 14824
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1892215871061436e-05,
+      "loss": 0.2578,
+      "step": 14825
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1891237169425314e-05,
+      "loss": 0.2302,
+      "step": 14826
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1890258449001503e-05,
+      "loss": 0.2462,
+      "step": 14827
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1889279709799728e-05,
+      "loss": 0.1928,
+      "step": 14828
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1888300951829713e-05,
+      "loss": 0.2764,
+      "step": 14829
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1887322175101174e-05,
+      "loss": 0.2499,
+      "step": 14830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1886343379623844e-05,
+      "loss": 0.2271,
+      "step": 14831
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.188536456540744e-05,
+      "loss": 0.2731,
+      "step": 14832
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.188438573246169e-05,
+      "loss": 0.2637,
+      "step": 14833
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1883406880796313e-05,
+      "loss": 0.283,
+      "step": 14834
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1882428010421036e-05,
+      "loss": 0.2438,
+      "step": 14835
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1881449121345583e-05,
+      "loss": 0.2798,
+      "step": 14836
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.188047021357968e-05,
+      "loss": 0.2287,
+      "step": 14837
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1879491287133045e-05,
+      "loss": 0.2586,
+      "step": 14838
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1878512342015412e-05,
+      "loss": 0.2353,
+      "step": 14839
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.18775333782365e-05,
+      "loss": 0.2497,
+      "step": 14840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.187655439580603e-05,
+      "loss": 0.2359,
+      "step": 14841
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1875575394733739e-05,
+      "loss": 0.285,
+      "step": 14842
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1874596375029342e-05,
+      "loss": 0.2355,
+      "step": 14843
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.187361733670257e-05,
+      "loss": 0.2861,
+      "step": 14844
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1872638279763147e-05,
+      "loss": 0.2208,
+      "step": 14845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1871659204220802e-05,
+      "loss": 0.1993,
+      "step": 14846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1870680110085254e-05,
+      "loss": 0.2461,
+      "step": 14847
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1869700997366236e-05,
+      "loss": 0.2269,
+      "step": 14848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1868721866073477e-05,
+      "loss": 0.2502,
+      "step": 14849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1867742716216692e-05,
+      "loss": 0.1947,
+      "step": 14850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1866763547805623e-05,
+      "loss": 0.1969,
+      "step": 14851
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1865784360849983e-05,
+      "loss": 0.2664,
+      "step": 14852
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1864805155359508e-05,
+      "loss": 0.273,
+      "step": 14853
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1863825931343923e-05,
+      "loss": 0.2482,
+      "step": 14854
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1862846688812956e-05,
+      "loss": 0.2347,
+      "step": 14855
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1861867427776337e-05,
+      "loss": 0.2292,
+      "step": 14856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.186088814824379e-05,
+      "loss": 0.3068,
+      "step": 14857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1859908850225045e-05,
+      "loss": 0.2208,
+      "step": 14858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1858929533729831e-05,
+      "loss": 0.2541,
+      "step": 14859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1857950198767876e-05,
+      "loss": 0.2173,
+      "step": 14860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185697084534891e-05,
+      "loss": 0.2332,
+      "step": 14861
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1855991473482658e-05,
+      "loss": 0.3157,
+      "step": 14862
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1855012083178856e-05,
+      "loss": 0.2703,
+      "step": 14863
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1854032674447228e-05,
+      "loss": 0.3524,
+      "step": 14864
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1853053247297503e-05,
+      "loss": 0.2476,
+      "step": 14865
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1852073801739415e-05,
+      "loss": 0.214,
+      "step": 14866
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185109433778269e-05,
+      "loss": 0.2349,
+      "step": 14867
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1850114855437058e-05,
+      "loss": 0.2854,
+      "step": 14868
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1849135354712254e-05,
+      "loss": 0.2114,
+      "step": 14869
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1848155835618002e-05,
+      "loss": 0.2144,
+      "step": 14870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1847176298164037e-05,
+      "loss": 0.2175,
+      "step": 14871
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.184619674236009e-05,
+      "loss": 0.2485,
+      "step": 14872
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1845217168215893e-05,
+      "loss": 0.249,
+      "step": 14873
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1844237575741171e-05,
+      "loss": 0.2582,
+      "step": 14874
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.184325796494566e-05,
+      "loss": 0.2409,
+      "step": 14875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.184227833583909e-05,
+      "loss": 0.2208,
+      "step": 14876
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1841298688431194e-05,
+      "loss": 0.2156,
+      "step": 14877
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1840319022731704e-05,
+      "loss": 0.2315,
+      "step": 14878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1839339338750348e-05,
+      "loss": 0.2127,
+      "step": 14879
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1838359636496868e-05,
+      "loss": 0.2104,
+      "step": 14880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1837379915980982e-05,
+      "loss": 0.2122,
+      "step": 14881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1836400177212434e-05,
+      "loss": 0.239,
+      "step": 14882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1835420420200958e-05,
+      "loss": 0.2545,
+      "step": 14883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1834440644956276e-05,
+      "loss": 0.2086,
+      "step": 14884
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1833460851488132e-05,
+      "loss": 0.2347,
+      "step": 14885
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1832481039806251e-05,
+      "loss": 0.2376,
+      "step": 14886
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1831501209920373e-05,
+      "loss": 0.2289,
+      "step": 14887
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1830521361840227e-05,
+      "loss": 0.1742,
+      "step": 14888
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1829541495575552e-05,
+      "loss": 0.23,
+      "step": 14889
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1828561611136078e-05,
+      "loss": 0.243,
+      "step": 14890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1827581708531539e-05,
+      "loss": 0.2765,
+      "step": 14891
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1826601787771672e-05,
+      "loss": 0.2727,
+      "step": 14892
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182562184886621e-05,
+      "loss": 0.2741,
+      "step": 14893
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1824641891824885e-05,
+      "loss": 0.1757,
+      "step": 14894
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1823661916657441e-05,
+      "loss": 0.2456,
+      "step": 14895
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1822681923373602e-05,
+      "loss": 0.2218,
+      "step": 14896
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1821701911983112e-05,
+      "loss": 0.2572,
+      "step": 14897
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1820721882495702e-05,
+      "loss": 0.1918,
+      "step": 14898
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1819741834921108e-05,
+      "loss": 0.1795,
+      "step": 14899
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1818761769269068e-05,
+      "loss": 0.261,
+      "step": 14900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1817781685549315e-05,
+      "loss": 0.2858,
+      "step": 14901
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1816801583771586e-05,
+      "loss": 0.1981,
+      "step": 14902
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1815821463945621e-05,
+      "loss": 0.2478,
+      "step": 14903
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1814841326081155e-05,
+      "loss": 0.2047,
+      "step": 14904
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1813861170187919e-05,
+      "loss": 0.2625,
+      "step": 14905
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1812880996275655e-05,
+      "loss": 0.2071,
+      "step": 14906
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1811900804354104e-05,
+      "loss": 0.24,
+      "step": 14907
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1810920594432998e-05,
+      "loss": 0.2042,
+      "step": 14908
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1809940366522072e-05,
+      "loss": 0.2174,
+      "step": 14909
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1808960120631068e-05,
+      "loss": 0.2881,
+      "step": 14910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1807979856769725e-05,
+      "loss": 0.2765,
+      "step": 14911
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1806999574947777e-05,
+      "loss": 0.2831,
+      "step": 14912
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1806019275174966e-05,
+      "loss": 0.2445,
+      "step": 14913
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1805038957461026e-05,
+      "loss": 0.2491,
+      "step": 14914
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1804058621815699e-05,
+      "loss": 0.1899,
+      "step": 14915
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1803078268248725e-05,
+      "loss": 0.2707,
+      "step": 14916
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.180209789676984e-05,
+      "loss": 0.174,
+      "step": 14917
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.180111750738878e-05,
+      "loss": 0.2108,
+      "step": 14918
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1800137100115292e-05,
+      "loss": 0.2434,
+      "step": 14919
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1799156674959109e-05,
+      "loss": 0.2849,
+      "step": 14920
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1798176231929973e-05,
+      "loss": 0.2648,
+      "step": 14921
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1797195771037626e-05,
+      "loss": 0.2208,
+      "step": 14922
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1796215292291807e-05,
+      "loss": 0.2554,
+      "step": 14923
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.179523479570225e-05,
+      "loss": 0.2705,
+      "step": 14924
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1794254281278704e-05,
+      "loss": 0.2463,
+      "step": 14925
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1793273749030907e-05,
+      "loss": 0.2447,
+      "step": 14926
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1792293198968595e-05,
+      "loss": 0.2865,
+      "step": 14927
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1791312631101514e-05,
+      "loss": 0.2044,
+      "step": 14928
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1790332045439403e-05,
+      "loss": 0.2261,
+      "step": 14929
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1789351441992e-05,
+      "loss": 0.3296,
+      "step": 14930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1788370820769058e-05,
+      "loss": 0.2624,
+      "step": 14931
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1787390181780305e-05,
+      "loss": 0.1875,
+      "step": 14932
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1786409525035489e-05,
+      "loss": 0.2497,
+      "step": 14933
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1785428850544352e-05,
+      "loss": 0.2542,
+      "step": 14934
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1784448158316633e-05,
+      "loss": 0.241,
+      "step": 14935
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1783467448362077e-05,
+      "loss": 0.2253,
+      "step": 14936
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1782486720690429e-05,
+      "loss": 0.2511,
+      "step": 14937
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.178150597531143e-05,
+      "loss": 0.2209,
+      "step": 14938
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1780525212234815e-05,
+      "loss": 0.283,
+      "step": 14939
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1779544431470335e-05,
+      "loss": 0.2218,
+      "step": 14940
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1778563633027735e-05,
+      "loss": 0.2854,
+      "step": 14941
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1777582816916753e-05,
+      "loss": 0.1877,
+      "step": 14942
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1776601983147135e-05,
+      "loss": 0.2291,
+      "step": 14943
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1775621131728623e-05,
+      "loss": 0.2748,
+      "step": 14944
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1774640262670962e-05,
+      "loss": 0.1989,
+      "step": 14945
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1773659375983897e-05,
+      "loss": 0.1966,
+      "step": 14946
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1772678471677168e-05,
+      "loss": 0.2371,
+      "step": 14947
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1771697549760525e-05,
+      "loss": 0.2558,
+      "step": 14948
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177071661024371e-05,
+      "loss": 0.2048,
+      "step": 14949
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.176973565313647e-05,
+      "loss": 0.2159,
+      "step": 14950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1768754678448541e-05,
+      "loss": 0.1924,
+      "step": 14951
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1767773686189678e-05,
+      "loss": 0.2612,
+      "step": 14952
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1766792676369627e-05,
+      "loss": 0.2542,
+      "step": 14953
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1765811648998127e-05,
+      "loss": 0.2495,
+      "step": 14954
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1764830604084923e-05,
+      "loss": 0.2518,
+      "step": 14955
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1763849541639765e-05,
+      "loss": 0.2057,
+      "step": 14956
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1762868461672399e-05,
+      "loss": 0.2941,
+      "step": 14957
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.176188736419257e-05,
+      "loss": 0.2577,
+      "step": 14958
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1760906249210023e-05,
+      "loss": 0.2759,
+      "step": 14959
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1759925116734507e-05,
+      "loss": 0.2784,
+      "step": 14960
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1758943966775765e-05,
+      "loss": 0.2296,
+      "step": 14961
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1757962799343548e-05,
+      "loss": 0.2335,
+      "step": 14962
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1756981614447601e-05,
+      "loss": 0.2662,
+      "step": 14963
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1756000412097665e-05,
+      "loss": 0.2933,
+      "step": 14964
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1755019192303503e-05,
+      "loss": 0.2342,
+      "step": 14965
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1754037955074848e-05,
+      "loss": 0.1651,
+      "step": 14966
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1753056700421452e-05,
+      "loss": 0.1962,
+      "step": 14967
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1752075428353064e-05,
+      "loss": 0.2532,
+      "step": 14968
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1751094138879436e-05,
+      "loss": 0.2803,
+      "step": 14969
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1750112832010307e-05,
+      "loss": 0.2454,
+      "step": 14970
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.174913150775543e-05,
+      "loss": 0.2475,
+      "step": 14971
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1748150166124559e-05,
+      "loss": 0.222,
+      "step": 14972
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1747168807127428e-05,
+      "loss": 0.3232,
+      "step": 14973
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1746187430773805e-05,
+      "loss": 0.2472,
+      "step": 14974
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1745206037073426e-05,
+      "loss": 0.2862,
+      "step": 14975
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.174422462603604e-05,
+      "loss": 0.214,
+      "step": 14976
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1743243197671405e-05,
+      "loss": 0.2298,
+      "step": 14977
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1742261751989263e-05,
+      "loss": 0.2513,
+      "step": 14978
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1741280288999369e-05,
+      "loss": 0.2278,
+      "step": 14979
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1740298808711465e-05,
+      "loss": 0.2466,
+      "step": 14980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1739317311135313e-05,
+      "loss": 0.3097,
+      "step": 14981
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173833579628065e-05,
+      "loss": 0.2297,
+      "step": 14982
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1737354264157236e-05,
+      "loss": 0.1997,
+      "step": 14983
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1736372714774822e-05,
+      "loss": 0.1983,
+      "step": 14984
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1735391148143149e-05,
+      "loss": 0.2167,
+      "step": 14985
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1734409564271982e-05,
+      "loss": 0.2863,
+      "step": 14986
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173342796317106e-05,
+      "loss": 0.2302,
+      "step": 14987
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1732446344850139e-05,
+      "loss": 0.2139,
+      "step": 14988
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1731464709318969e-05,
+      "loss": 0.2449,
+      "step": 14989
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1730483056587306e-05,
+      "loss": 0.2452,
+      "step": 14990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1729501386664897e-05,
+      "loss": 0.2917,
+      "step": 14991
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1728519699561496e-05,
+      "loss": 0.2547,
+      "step": 14992
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1727537995286854e-05,
+      "loss": 0.2334,
+      "step": 14993
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1726556273850726e-05,
+      "loss": 0.2374,
+      "step": 14994
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1725574535262863e-05,
+      "loss": 0.2751,
+      "step": 14995
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1724592779533018e-05,
+      "loss": 0.3101,
+      "step": 14996
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1723611006670941e-05,
+      "loss": 0.2819,
+      "step": 14997
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1722629216686388e-05,
+      "loss": 0.2059,
+      "step": 14998
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1721647409589109e-05,
+      "loss": 0.1948,
+      "step": 14999
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1720665585388861e-05,
+      "loss": 0.2808,
+      "step": 15000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1719683744095399e-05,
+      "loss": 0.2343,
+      "step": 15001
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1718701885718471e-05,
+      "loss": 0.2641,
+      "step": 15002
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1717720010267834e-05,
+      "loss": 0.2376,
+      "step": 15003
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1716738117753243e-05,
+      "loss": 0.271,
+      "step": 15004
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.171575620818445e-05,
+      "loss": 0.2157,
+      "step": 15005
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1714774281571213e-05,
+      "loss": 0.2721,
+      "step": 15006
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.171379233792328e-05,
+      "loss": 0.2895,
+      "step": 15007
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1712810377250414e-05,
+      "loss": 0.3054,
+      "step": 15008
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1711828399562362e-05,
+      "loss": 0.2399,
+      "step": 15009
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1710846404868881e-05,
+      "loss": 0.2156,
+      "step": 15010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1709864393179732e-05,
+      "loss": 0.3096,
+      "step": 15011
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1708882364504664e-05,
+      "loss": 0.2759,
+      "step": 15012
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1707900318853436e-05,
+      "loss": 0.2838,
+      "step": 15013
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.17069182562358e-05,
+      "loss": 0.2285,
+      "step": 15014
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1705936176661516e-05,
+      "loss": 0.238,
+      "step": 15015
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1704954080140336e-05,
+      "loss": 0.291,
+      "step": 15016
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.170397196668202e-05,
+      "loss": 0.2485,
+      "step": 15017
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1702989836296326e-05,
+      "loss": 0.2303,
+      "step": 15018
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1702007688993e-05,
+      "loss": 0.2389,
+      "step": 15019
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1701025524781812e-05,
+      "loss": 0.3556,
+      "step": 15020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1700043343672511e-05,
+      "loss": 0.2068,
+      "step": 15021
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1699061145674855e-05,
+      "loss": 0.1625,
+      "step": 15022
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1698078930798606e-05,
+      "loss": 0.2698,
+      "step": 15023
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1697096699053516e-05,
+      "loss": 0.2702,
+      "step": 15024
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1696114450449343e-05,
+      "loss": 0.2805,
+      "step": 15025
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1695132184995848e-05,
+      "loss": 0.254,
+      "step": 15026
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1694149902702788e-05,
+      "loss": 0.296,
+      "step": 15027
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1693167603579913e-05,
+      "loss": 0.2166,
+      "step": 15028
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1692185287636995e-05,
+      "loss": 0.2574,
+      "step": 15029
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1691202954883785e-05,
+      "loss": 0.2798,
+      "step": 15030
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1690220605330036e-05,
+      "loss": 0.194,
+      "step": 15031
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.168923823898552e-05,
+      "loss": 0.2091,
+      "step": 15032
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1688255855859987e-05,
+      "loss": 0.221,
+      "step": 15033
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1687273455963197e-05,
+      "loss": 0.1844,
+      "step": 15034
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.168629103930491e-05,
+      "loss": 0.2198,
+      "step": 15035
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1685308605894885e-05,
+      "loss": 0.1783,
+      "step": 15036
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1684326155742883e-05,
+      "loss": 0.2404,
+      "step": 15037
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1683343688858662e-05,
+      "loss": 0.2629,
+      "step": 15038
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1682361205251982e-05,
+      "loss": 0.2421,
+      "step": 15039
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1681378704932607e-05,
+      "loss": 0.2154,
+      "step": 15040
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.168039618791029e-05,
+      "loss": 0.2726,
+      "step": 15041
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1679413654194798e-05,
+      "loss": 0.2275,
+      "step": 15042
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1678431103795888e-05,
+      "loss": 0.2352,
+      "step": 15043
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1677448536723322e-05,
+      "loss": 0.1943,
+      "step": 15044
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.167646595298686e-05,
+      "loss": 0.2763,
+      "step": 15045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1675483352596264e-05,
+      "loss": 0.3033,
+      "step": 15046
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1674500735561294e-05,
+      "loss": 0.2515,
+      "step": 15047
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1673518101891713e-05,
+      "loss": 0.2425,
+      "step": 15048
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1672535451597284e-05,
+      "loss": 0.2782,
+      "step": 15049
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.167155278468776e-05,
+      "loss": 0.2056,
+      "step": 15050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1670570101172913e-05,
+      "loss": 0.2705,
+      "step": 15051
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1669587401062505e-05,
+      "loss": 0.2781,
+      "step": 15052
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1668604684366288e-05,
+      "loss": 0.241,
+      "step": 15053
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1667621951094037e-05,
+      "loss": 0.242,
+      "step": 15054
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1666639201255507e-05,
+      "loss": 0.2583,
+      "step": 15055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1665656434860461e-05,
+      "loss": 0.2149,
+      "step": 15056
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1664673651918664e-05,
+      "loss": 0.2463,
+      "step": 15057
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1663690852439878e-05,
+      "loss": 0.2291,
+      "step": 15058
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1662708036433865e-05,
+      "loss": 0.2048,
+      "step": 15059
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.166172520391039e-05,
+      "loss": 0.2555,
+      "step": 15060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.166074235487922e-05,
+      "loss": 0.2193,
+      "step": 15061
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1659759489350108e-05,
+      "loss": 0.2254,
+      "step": 15062
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1658776607332828e-05,
+      "loss": 0.2021,
+      "step": 15063
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1657793708837146e-05,
+      "loss": 0.2436,
+      "step": 15064
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1656810793872813e-05,
+      "loss": 0.2467,
+      "step": 15065
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1655827862449608e-05,
+      "loss": 0.2649,
+      "step": 15066
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1654844914577285e-05,
+      "loss": 0.3052,
+      "step": 15067
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1653861950265613e-05,
+      "loss": 0.2405,
+      "step": 15068
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1652878969524357e-05,
+      "loss": 0.283,
+      "step": 15069
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.165189597236328e-05,
+      "loss": 0.2298,
+      "step": 15070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1650912958792151e-05,
+      "loss": 0.271,
+      "step": 15071
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1649929928820732e-05,
+      "loss": 0.3344,
+      "step": 15072
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.164894688245879e-05,
+      "loss": 0.2233,
+      "step": 15073
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1647963819716085e-05,
+      "loss": 0.2559,
+      "step": 15074
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1646980740602394e-05,
+      "loss": 0.2297,
+      "step": 15075
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1645997645127476e-05,
+      "loss": 0.2313,
+      "step": 15076
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1645014533301098e-05,
+      "loss": 0.2612,
+      "step": 15077
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1644031405133025e-05,
+      "loss": 0.2352,
+      "step": 15078
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1643048260633024e-05,
+      "loss": 0.267,
+      "step": 15079
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1642065099810862e-05,
+      "loss": 0.2051,
+      "step": 15080
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1641081922676307e-05,
+      "loss": 0.2844,
+      "step": 15081
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1640098729239126e-05,
+      "loss": 0.2604,
+      "step": 15082
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1639115519509086e-05,
+      "loss": 0.2387,
+      "step": 15083
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1638132293495951e-05,
+      "loss": 0.2276,
+      "step": 15084
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1637149051209496e-05,
+      "loss": 0.2386,
+      "step": 15085
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1636165792659478e-05,
+      "loss": 0.2616,
+      "step": 15086
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.163518251785567e-05,
+      "loss": 0.2515,
+      "step": 15087
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1634199226807842e-05,
+      "loss": 0.2202,
+      "step": 15088
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1633215919525762e-05,
+      "loss": 0.2869,
+      "step": 15089
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1632232596019195e-05,
+      "loss": 0.2689,
+      "step": 15090
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1631249256297911e-05,
+      "loss": 0.3644,
+      "step": 15091
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1630265900371678e-05,
+      "loss": 0.2795,
+      "step": 15092
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1629282528250268e-05,
+      "loss": 0.2326,
+      "step": 15093
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1628299139943444e-05,
+      "loss": 0.217,
+      "step": 15094
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.162731573546098e-05,
+      "loss": 0.239,
+      "step": 15095
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1626332314812641e-05,
+      "loss": 0.2801,
+      "step": 15096
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1625348878008202e-05,
+      "loss": 0.2805,
+      "step": 15097
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1624365425057429e-05,
+      "loss": 0.2269,
+      "step": 15098
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1623381955970088e-05,
+      "loss": 0.2738,
+      "step": 15099
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1622398470755957e-05,
+      "loss": 0.2574,
+      "step": 15100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.16214149694248e-05,
+      "loss": 0.2498,
+      "step": 15101
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.162043145198639e-05,
+      "loss": 0.229,
+      "step": 15102
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1619447918450497e-05,
+      "loss": 0.2398,
+      "step": 15103
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1618464368826889e-05,
+      "loss": 0.2243,
+      "step": 15104
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1617480803125338e-05,
+      "loss": 0.2646,
+      "step": 15105
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1616497221355618e-05,
+      "loss": 0.2205,
+      "step": 15106
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1615513623527498e-05,
+      "loss": 0.2281,
+      "step": 15107
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1614530009650743e-05,
+      "loss": 0.2566,
+      "step": 15108
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1613546379735134e-05,
+      "loss": 0.2146,
+      "step": 15109
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.161256273379044e-05,
+      "loss": 0.2649,
+      "step": 15110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1611579071826427e-05,
+      "loss": 0.2585,
+      "step": 15111
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1610595393852873e-05,
+      "loss": 0.2285,
+      "step": 15112
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1609611699879546e-05,
+      "loss": 0.2319,
+      "step": 15113
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1608627989916221e-05,
+      "loss": 0.2065,
+      "step": 15114
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1607644263972669e-05,
+      "loss": 0.253,
+      "step": 15115
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.160666052205866e-05,
+      "loss": 0.224,
+      "step": 15116
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.160567676418397e-05,
+      "loss": 0.234,
+      "step": 15117
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.160469299035837e-05,
+      "loss": 0.2525,
+      "step": 15118
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1603709200591635e-05,
+      "loss": 0.2056,
+      "step": 15119
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1602725394893533e-05,
+      "loss": 0.276,
+      "step": 15120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1601741573273843e-05,
+      "loss": 0.2478,
+      "step": 15121
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1600757735742338e-05,
+      "loss": 0.2107,
+      "step": 15122
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1599773882308786e-05,
+      "loss": 0.2504,
+      "step": 15123
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1598790012982968e-05,
+      "loss": 0.2443,
+      "step": 15124
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1597806127774651e-05,
+      "loss": 0.2292,
+      "step": 15125
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1596822226693614e-05,
+      "loss": 0.2419,
+      "step": 15126
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1595838309749627e-05,
+      "loss": 0.2963,
+      "step": 15127
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1594854376952468e-05,
+      "loss": 0.2356,
+      "step": 15128
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.159387042831191e-05,
+      "loss": 0.2772,
+      "step": 15129
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1592886463837724e-05,
+      "loss": 0.2551,
+      "step": 15130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1591902483539694e-05,
+      "loss": 0.2622,
+      "step": 15131
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1590918487427585e-05,
+      "loss": 0.2202,
+      "step": 15132
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1589934475511175e-05,
+      "loss": 0.2699,
+      "step": 15133
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1588950447800246e-05,
+      "loss": 0.276,
+      "step": 15134
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1587966404304565e-05,
+      "loss": 0.23,
+      "step": 15135
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1586982345033911e-05,
+      "loss": 0.2415,
+      "step": 15136
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1585998269998057e-05,
+      "loss": 0.2388,
+      "step": 15137
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1585014179206788e-05,
+      "loss": 0.2621,
+      "step": 15138
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1584030072669865e-05,
+      "loss": 0.2343,
+      "step": 15139
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1583045950397076e-05,
+      "loss": 0.2375,
+      "step": 15140
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1582061812398196e-05,
+      "loss": 0.2214,
+      "step": 15141
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1581077658682993e-05,
+      "loss": 0.2207,
+      "step": 15142
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1580093489261255e-05,
+      "loss": 0.2715,
+      "step": 15143
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1579109304142754e-05,
+      "loss": 0.2377,
+      "step": 15144
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.157812510333726e-05,
+      "loss": 0.2749,
+      "step": 15145
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1577140886854564e-05,
+      "loss": 0.2848,
+      "step": 15146
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1576156654704434e-05,
+      "loss": 0.2596,
+      "step": 15147
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1575172406896646e-05,
+      "loss": 0.2573,
+      "step": 15148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1574188143440985e-05,
+      "loss": 0.3188,
+      "step": 15149
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1573203864347222e-05,
+      "loss": 0.2273,
+      "step": 15150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.157221956962514e-05,
+      "loss": 0.2333,
+      "step": 15151
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1571235259284515e-05,
+      "loss": 0.2804,
+      "step": 15152
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1570250933335125e-05,
+      "loss": 0.2935,
+      "step": 15153
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1569266591786744e-05,
+      "loss": 0.1798,
+      "step": 15154
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.156828223464916e-05,
+      "loss": 0.2313,
+      "step": 15155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1567297861932145e-05,
+      "loss": 0.2336,
+      "step": 15156
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1566313473645478e-05,
+      "loss": 0.1781,
+      "step": 15157
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.156532906979894e-05,
+      "loss": 0.2925,
+      "step": 15158
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.2964,
+      "step": 15159
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1563360215465367e-05,
+      "loss": 0.2064,
+      "step": 15160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1562375764997889e-05,
+      "loss": 0.278,
+      "step": 15161
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1561391299009657e-05,
+      "loss": 0.2434,
+      "step": 15162
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.156040681751045e-05,
+      "loss": 0.2102,
+      "step": 15163
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.155942232051005e-05,
+      "loss": 0.2424,
+      "step": 15164
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1558437808018238e-05,
+      "loss": 0.2153,
+      "step": 15165
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1557453280044788e-05,
+      "loss": 0.258,
+      "step": 15166
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1556468736599485e-05,
+      "loss": 0.2171,
+      "step": 15167
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1555484177692107e-05,
+      "loss": 0.2724,
+      "step": 15168
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1554499603332439e-05,
+      "loss": 0.248,
+      "step": 15169
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1553515013530257e-05,
+      "loss": 0.2229,
+      "step": 15170
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1552530408295344e-05,
+      "loss": 0.2267,
+      "step": 15171
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1551545787637483e-05,
+      "loss": 0.2802,
+      "step": 15172
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1550561151566453e-05,
+      "loss": 0.2103,
+      "step": 15173
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1549576500092035e-05,
+      "loss": 0.1758,
+      "step": 15174
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1548591833224015e-05,
+      "loss": 0.2648,
+      "step": 15175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1547607150972166e-05,
+      "loss": 0.2394,
+      "step": 15176
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1546622453346281e-05,
+      "loss": 0.2213,
+      "step": 15177
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1545637740356132e-05,
+      "loss": 0.2511,
+      "step": 15178
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1544653012011506e-05,
+      "loss": 0.3104,
+      "step": 15179
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1543668268322187e-05,
+      "loss": 0.2542,
+      "step": 15180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1542683509297956e-05,
+      "loss": 0.2399,
+      "step": 15181
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1541698734948592e-05,
+      "loss": 0.2083,
+      "step": 15182
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1540713945283883e-05,
+      "loss": 0.2242,
+      "step": 15183
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1539729140313611e-05,
+      "loss": 0.1589,
+      "step": 15184
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1538744320047554e-05,
+      "loss": 0.2208,
+      "step": 15185
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1537759484495503e-05,
+      "loss": 0.2857,
+      "step": 15186
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1536774633667238e-05,
+      "loss": 0.2083,
+      "step": 15187
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1535789767572538e-05,
+      "loss": 0.2058,
+      "step": 15188
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1534804886221195e-05,
+      "loss": 0.2704,
+      "step": 15189
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1533819989622987e-05,
+      "loss": 0.1833,
+      "step": 15190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1532835077787699e-05,
+      "loss": 0.2365,
+      "step": 15191
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.153185015072512e-05,
+      "loss": 0.2104,
+      "step": 15192
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1530865208445028e-05,
+      "loss": 0.2021,
+      "step": 15193
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152988025095721e-05,
+      "loss": 0.1963,
+      "step": 15194
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152889527827145e-05,
+      "loss": 0.2198,
+      "step": 15195
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1527910290397537e-05,
+      "loss": 0.2336,
+      "step": 15196
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1526925287345248e-05,
+      "loss": 0.3036,
+      "step": 15197
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1525940269124374e-05,
+      "loss": 0.2298,
+      "step": 15198
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.15249552357447e-05,
+      "loss": 0.2343,
+      "step": 15199
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1523970187216004e-05,
+      "loss": 0.204,
+      "step": 15200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1522985123548085e-05,
+      "loss": 0.239,
+      "step": 15201
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1522000044750718e-05,
+      "loss": 0.2061,
+      "step": 15202
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1521014950833693e-05,
+      "loss": 0.2518,
+      "step": 15203
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1520029841806795e-05,
+      "loss": 0.2831,
+      "step": 15204
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.151904471767981e-05,
+      "loss": 0.3072,
+      "step": 15205
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1518059578462524e-05,
+      "loss": 0.4136,
+      "step": 15206
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1517074424164724e-05,
+      "loss": 0.2539,
+      "step": 15207
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1516089254796199e-05,
+      "loss": 0.2194,
+      "step": 15208
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1515104070366732e-05,
+      "loss": 0.3065,
+      "step": 15209
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1514118870886108e-05,
+      "loss": 0.2558,
+      "step": 15210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1513133656364123e-05,
+      "loss": 0.2171,
+      "step": 15211
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1512148426810557e-05,
+      "loss": 0.2231,
+      "step": 15212
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1511163182235196e-05,
+      "loss": 0.2332,
+      "step": 15213
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1510177922647832e-05,
+      "loss": 0.2508,
+      "step": 15214
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.150919264805825e-05,
+      "loss": 0.2855,
+      "step": 15215
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1508207358476237e-05,
+      "loss": 0.2828,
+      "step": 15216
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1507222053911586e-05,
+      "loss": 0.2174,
+      "step": 15217
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1506236734374082e-05,
+      "loss": 0.181,
+      "step": 15218
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1505251399873511e-05,
+      "loss": 0.2306,
+      "step": 15219
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1504266050419663e-05,
+      "loss": 0.2274,
+      "step": 15220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.150328068602233e-05,
+      "loss": 0.2692,
+      "step": 15221
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1502295306691295e-05,
+      "loss": 0.2784,
+      "step": 15222
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1501309912436351e-05,
+      "loss": 0.217,
+      "step": 15223
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1500324503267287e-05,
+      "loss": 0.2032,
+      "step": 15224
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1499339079193886e-05,
+      "loss": 0.2053,
+      "step": 15225
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1498353640225943e-05,
+      "loss": 0.2682,
+      "step": 15226
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1497368186373247e-05,
+      "loss": 0.2731,
+      "step": 15227
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.149638271764559e-05,
+      "loss": 0.1773,
+      "step": 15228
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1495397234052752e-05,
+      "loss": 0.2435,
+      "step": 15229
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1494411735604536e-05,
+      "loss": 0.2592,
+      "step": 15230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.149342622231072e-05,
+      "loss": 0.2115,
+      "step": 15231
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1492440694181101e-05,
+      "loss": 0.238,
+      "step": 15232
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1491455151225473e-05,
+      "loss": 0.2141,
+      "step": 15233
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1490469593453612e-05,
+      "loss": 0.2034,
+      "step": 15234
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1489484020875325e-05,
+      "loss": 0.2343,
+      "step": 15235
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1488498433500395e-05,
+      "loss": 0.1907,
+      "step": 15236
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1487512831338612e-05,
+      "loss": 0.2284,
+      "step": 15237
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1486527214399768e-05,
+      "loss": 0.2261,
+      "step": 15238
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1485541582693655e-05,
+      "loss": 0.2735,
+      "step": 15239
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1484555936230064e-05,
+      "loss": 0.3003,
+      "step": 15240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1483570275018787e-05,
+      "loss": 0.2263,
+      "step": 15241
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1482584599069619e-05,
+      "loss": 0.2118,
+      "step": 15242
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1481598908392342e-05,
+      "loss": 0.2677,
+      "step": 15243
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1480613202996755e-05,
+      "loss": 0.2339,
+      "step": 15244
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1479627482892655e-05,
+      "loss": 0.2311,
+      "step": 15245
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1478641748089822e-05,
+      "loss": 0.2609,
+      "step": 15246
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1477655998598057e-05,
+      "loss": 0.2494,
+      "step": 15247
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.147667023442715e-05,
+      "loss": 0.163,
+      "step": 15248
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1475684455586893e-05,
+      "loss": 0.2394,
+      "step": 15249
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1474698662087081e-05,
+      "loss": 0.2764,
+      "step": 15250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1473712853937502e-05,
+      "loss": 0.2574,
+      "step": 15251
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1472727031147954e-05,
+      "loss": 0.2452,
+      "step": 15252
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1471741193728232e-05,
+      "loss": 0.2569,
+      "step": 15253
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1470755341688125e-05,
+      "loss": 0.1814,
+      "step": 15254
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1469769475037427e-05,
+      "loss": 0.262,
+      "step": 15255
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.146878359378593e-05,
+      "loss": 0.2225,
+      "step": 15256
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1467797697943435e-05,
+      "loss": 0.2539,
+      "step": 15257
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1466811787519726e-05,
+      "loss": 0.2272,
+      "step": 15258
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1465825862524607e-05,
+      "loss": 0.2213,
+      "step": 15259
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1464839922967866e-05,
+      "loss": 0.2539,
+      "step": 15260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1463853968859296e-05,
+      "loss": 0.1641,
+      "step": 15261
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1462868000208698e-05,
+      "loss": 0.3375,
+      "step": 15262
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.146188201702586e-05,
+      "loss": 0.2902,
+      "step": 15263
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1460896019320584e-05,
+      "loss": 0.2682,
+      "step": 15264
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1459910007102657e-05,
+      "loss": 0.239,
+      "step": 15265
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1458923980381882e-05,
+      "loss": 0.317,
+      "step": 15266
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1457937939168048e-05,
+      "loss": 0.2486,
+      "step": 15267
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1456951883470951e-05,
+      "loss": 0.2177,
+      "step": 15268
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1455965813300391e-05,
+      "loss": 0.2197,
+      "step": 15269
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.145497972866616e-05,
+      "loss": 0.2932,
+      "step": 15270
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1453993629578055e-05,
+      "loss": 0.3302,
+      "step": 15271
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1453007516045873e-05,
+      "loss": 0.2716,
+      "step": 15272
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.145202138807941e-05,
+      "loss": 0.2561,
+      "step": 15273
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1451035245688456e-05,
+      "loss": 0.228,
+      "step": 15274
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1450049088882816e-05,
+      "loss": 0.226,
+      "step": 15275
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1449062917672284e-05,
+      "loss": 0.2098,
+      "step": 15276
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1448076732066653e-05,
+      "loss": 0.2773,
+      "step": 15277
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1447090532075725e-05,
+      "loss": 0.2383,
+      "step": 15278
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1446104317709295e-05,
+      "loss": 0.1676,
+      "step": 15279
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1445118088977156e-05,
+      "loss": 0.3384,
+      "step": 15280
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1444131845889115e-05,
+      "loss": 0.1738,
+      "step": 15281
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1443145588454958e-05,
+      "loss": 0.2831,
+      "step": 15282
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1442159316684492e-05,
+      "loss": 0.2104,
+      "step": 15283
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1441173030587508e-05,
+      "loss": 0.277,
+      "step": 15284
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1440186730173808e-05,
+      "loss": 0.266,
+      "step": 15285
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1439200415453186e-05,
+      "loss": 0.2628,
+      "step": 15286
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1438214086435447e-05,
+      "loss": 0.2398,
+      "step": 15287
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1437227743130384e-05,
+      "loss": 0.249,
+      "step": 15288
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1436241385547794e-05,
+      "loss": 0.3009,
+      "step": 15289
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1435255013697477e-05,
+      "loss": 0.2412,
+      "step": 15290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1434268627589236e-05,
+      "loss": 0.2656,
+      "step": 15291
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1433282227232865e-05,
+      "loss": 0.2663,
+      "step": 15292
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1432295812638163e-05,
+      "loss": 0.2189,
+      "step": 15293
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1431309383814933e-05,
+      "loss": 0.2862,
+      "step": 15294
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.143032294077297e-05,
+      "loss": 0.1988,
+      "step": 15295
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1429336483522076e-05,
+      "loss": 0.2185,
+      "step": 15296
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.142835001207205e-05,
+      "loss": 0.2419,
+      "step": 15297
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1427363526432692e-05,
+      "loss": 0.2748,
+      "step": 15298
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1426377026613797e-05,
+      "loss": 0.2386,
+      "step": 15299
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1425390512625175e-05,
+      "loss": 0.2193,
+      "step": 15300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1424403984476615e-05,
+      "loss": 0.2169,
+      "step": 15301
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1423417442177922e-05,
+      "loss": 0.2016,
+      "step": 15302
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1422430885738902e-05,
+      "loss": 0.1763,
+      "step": 15303
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1421444315169348e-05,
+      "loss": 0.2439,
+      "step": 15304
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1420457730479061e-05,
+      "loss": 0.1801,
+      "step": 15305
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1419471131677847e-05,
+      "loss": 0.3039,
+      "step": 15306
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1418484518775505e-05,
+      "loss": 0.2112,
+      "step": 15307
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.141749789178183e-05,
+      "loss": 0.1502,
+      "step": 15308
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.141651125070663e-05,
+      "loss": 0.2352,
+      "step": 15309
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1415524595559709e-05,
+      "loss": 0.2585,
+      "step": 15310
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1414537926350854e-05,
+      "loss": 0.2097,
+      "step": 15311
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1413551243089887e-05,
+      "loss": 0.2498,
+      "step": 15312
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1412564545786596e-05,
+      "loss": 0.1957,
+      "step": 15313
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1411577834450782e-05,
+      "loss": 0.2365,
+      "step": 15314
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1410591109092257e-05,
+      "loss": 0.2536,
+      "step": 15315
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1409604369720814e-05,
+      "loss": 0.2361,
+      "step": 15316
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1408617616346261e-05,
+      "loss": 0.2108,
+      "step": 15317
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1407630848978397e-05,
+      "loss": 0.2303,
+      "step": 15318
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1406644067627026e-05,
+      "loss": 0.2523,
+      "step": 15319
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.140565727230195e-05,
+      "loss": 0.2747,
+      "step": 15320
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1404670463012975e-05,
+      "loss": 0.3033,
+      "step": 15321
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1403683639769903e-05,
+      "loss": 0.2472,
+      "step": 15322
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1402696802582531e-05,
+      "loss": 0.2526,
+      "step": 15323
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1401709951460674e-05,
+      "loss": 0.2246,
+      "step": 15324
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1400723086414122e-05,
+      "loss": 0.217,
+      "step": 15325
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.139973620745269e-05,
+      "loss": 0.2491,
+      "step": 15326
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1398749314586172e-05,
+      "loss": 0.2044,
+      "step": 15327
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1397762407824379e-05,
+      "loss": 0.1926,
+      "step": 15328
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1396775487177114e-05,
+      "loss": 0.2549,
+      "step": 15329
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1395788552654179e-05,
+      "loss": 0.2603,
+      "step": 15330
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.139480160426538e-05,
+      "loss": 0.2912,
+      "step": 15331
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1393814642020519e-05,
+      "loss": 0.2569,
+      "step": 15332
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1392827665929405e-05,
+      "loss": 0.1932,
+      "step": 15333
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.139184067600184e-05,
+      "loss": 0.1966,
+      "step": 15334
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1390853672247629e-05,
+      "loss": 0.3041,
+      "step": 15335
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1389866654676574e-05,
+      "loss": 0.2326,
+      "step": 15336
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1388879623298484e-05,
+      "loss": 0.2159,
+      "step": 15337
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1387892578123161e-05,
+      "loss": 0.2168,
+      "step": 15338
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1386905519160413e-05,
+      "loss": 0.2699,
+      "step": 15339
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1385918446420048e-05,
+      "loss": 0.2426,
+      "step": 15340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1384931359911867e-05,
+      "loss": 0.2628,
+      "step": 15341
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1383944259645676e-05,
+      "loss": 0.2582,
+      "step": 15342
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1382957145631284e-05,
+      "loss": 0.21,
+      "step": 15343
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1381970017878497e-05,
+      "loss": 0.1695,
+      "step": 15344
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1380982876397115e-05,
+      "loss": 0.2611,
+      "step": 15345
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1379995721196952e-05,
+      "loss": 0.226,
+      "step": 15346
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1379008552287812e-05,
+      "loss": 0.2357,
+      "step": 15347
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1378021369679498e-05,
+      "loss": 0.2597,
+      "step": 15348
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1377034173381823e-05,
+      "loss": 0.2245,
+      "step": 15349
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1376046963404588e-05,
+      "loss": 0.2029,
+      "step": 15350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1375059739757604e-05,
+      "loss": 0.2148,
+      "step": 15351
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1374072502450675e-05,
+      "loss": 0.2756,
+      "step": 15352
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1373085251493615e-05,
+      "loss": 0.2103,
+      "step": 15353
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.137209798689622e-05,
+      "loss": 0.2398,
+      "step": 15354
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1371110708668307e-05,
+      "loss": 0.2279,
+      "step": 15355
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1370123416819683e-05,
+      "loss": 0.2479,
+      "step": 15356
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1369136111360148e-05,
+      "loss": 0.2555,
+      "step": 15357
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1368148792299523e-05,
+      "loss": 0.2706,
+      "step": 15358
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1367161459647604e-05,
+      "loss": 0.2762,
+      "step": 15359
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1366174113414202e-05,
+      "loss": 0.2086,
+      "step": 15360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1365186753609132e-05,
+      "loss": 0.1976,
+      "step": 15361
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1364199380242193e-05,
+      "loss": 0.1931,
+      "step": 15362
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1363211993323203e-05,
+      "loss": 0.2428,
+      "step": 15363
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1362224592861963e-05,
+      "loss": 0.3287,
+      "step": 15364
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1361237178868288e-05,
+      "loss": 0.2069,
+      "step": 15365
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.136024975135198e-05,
+      "loss": 0.2579,
+      "step": 15366
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1359262310322854e-05,
+      "loss": 0.281,
+      "step": 15367
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.135827485579072e-05,
+      "loss": 0.2052,
+      "step": 15368
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1357287387765383e-05,
+      "loss": 0.2803,
+      "step": 15369
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1356299906256656e-05,
+      "loss": 0.2553,
+      "step": 15370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1355312411274344e-05,
+      "loss": 0.3009,
+      "step": 15371
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1354324902828262e-05,
+      "loss": 0.2314,
+      "step": 15372
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.135333738092822e-05,
+      "loss": 0.1923,
+      "step": 15373
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1352349845584023e-05,
+      "loss": 0.2686,
+      "step": 15374
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1351362296805487e-05,
+      "loss": 0.2465,
+      "step": 15375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1350374734602418e-05,
+      "loss": 0.2182,
+      "step": 15376
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1349387158984631e-05,
+      "loss": 0.2521,
+      "step": 15377
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1348399569961935e-05,
+      "loss": 0.2377,
+      "step": 15378
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1347411967544132e-05,
+      "loss": 0.2203,
+      "step": 15379
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.134642435174105e-05,
+      "loss": 0.2457,
+      "step": 15380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1345436722562486e-05,
+      "loss": 0.2293,
+      "step": 15381
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1344449080018259e-05,
+      "loss": 0.2271,
+      "step": 15382
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1343461424118174e-05,
+      "loss": 0.2209,
+      "step": 15383
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1342473754872047e-05,
+      "loss": 0.2616,
+      "step": 15384
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1341486072289686e-05,
+      "loss": 0.3156,
+      "step": 15385
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1340498376380908e-05,
+      "loss": 0.3128,
+      "step": 15386
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1339510667155524e-05,
+      "loss": 0.2822,
+      "step": 15387
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1338522944623338e-05,
+      "loss": 0.2803,
+      "step": 15388
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1337535208794172e-05,
+      "loss": 0.2576,
+      "step": 15389
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1336547459677836e-05,
+      "loss": 0.2204,
+      "step": 15390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1335559697284136e-05,
+      "loss": 0.2634,
+      "step": 15391
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1334571921622893e-05,
+      "loss": 0.2499,
+      "step": 15392
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1333584132703912e-05,
+      "loss": 0.2798,
+      "step": 15393
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1332596330537011e-05,
+      "loss": 0.2942,
+      "step": 15394
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1331608515132001e-05,
+      "loss": 0.2444,
+      "step": 15395
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1330620686498697e-05,
+      "loss": 0.2664,
+      "step": 15396
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1329632844646908e-05,
+      "loss": 0.2347,
+      "step": 15397
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1328644989586453e-05,
+      "loss": 0.236,
+      "step": 15398
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132765712132714e-05,
+      "loss": 0.2363,
+      "step": 15399
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1326669239878785e-05,
+      "loss": 0.2509,
+      "step": 15400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.13256813452512e-05,
+      "loss": 0.2124,
+      "step": 15401
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1324693437454205e-05,
+      "loss": 0.2398,
+      "step": 15402
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1323705516497605e-05,
+      "loss": 0.2277,
+      "step": 15403
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132271758239122e-05,
+      "loss": 0.2352,
+      "step": 15404
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1321729635144862e-05,
+      "loss": 0.2459,
+      "step": 15405
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1320741674768347e-05,
+      "loss": 0.2337,
+      "step": 15406
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1319753701271487e-05,
+      "loss": 0.1837,
+      "step": 15407
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.13187657146641e-05,
+      "loss": 0.2649,
+      "step": 15408
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1317777714955998e-05,
+      "loss": 0.2299,
+      "step": 15409
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1316789702156994e-05,
+      "loss": 0.2315,
+      "step": 15410
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.131580167627691e-05,
+      "loss": 0.2505,
+      "step": 15411
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1314813637325553e-05,
+      "loss": 0.2385,
+      "step": 15412
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1313825585312743e-05,
+      "loss": 0.2355,
+      "step": 15413
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1312837520248296e-05,
+      "loss": 0.2802,
+      "step": 15414
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1311849442142025e-05,
+      "loss": 0.2254,
+      "step": 15415
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1310861351003745e-05,
+      "loss": 0.2104,
+      "step": 15416
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1309873246843274e-05,
+      "loss": 0.2888,
+      "step": 15417
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1308885129670426e-05,
+      "loss": 0.1944,
+      "step": 15418
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.130789699949502e-05,
+      "loss": 0.2626,
+      "step": 15419
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1306908856326871e-05,
+      "loss": 0.2363,
+      "step": 15420
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1305920700175792e-05,
+      "loss": 0.2101,
+      "step": 15421
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1304932531051602e-05,
+      "loss": 0.2653,
+      "step": 15422
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1303944348964122e-05,
+      "loss": 0.2152,
+      "step": 15423
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.130295615392316e-05,
+      "loss": 0.2231,
+      "step": 15424
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1301967945938538e-05,
+      "loss": 0.1888,
+      "step": 15425
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1300979725020072e-05,
+      "loss": 0.2882,
+      "step": 15426
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.129999149117758e-05,
+      "loss": 0.299,
+      "step": 15427
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1299003244420877e-05,
+      "loss": 0.1774,
+      "step": 15428
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1298014984759782e-05,
+      "loss": 0.2677,
+      "step": 15429
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1297026712204111e-05,
+      "loss": 0.3142,
+      "step": 15430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1296038426763682e-05,
+      "loss": 0.1651,
+      "step": 15431
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1295050128448314e-05,
+      "loss": 0.2069,
+      "step": 15432
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1294061817267829e-05,
+      "loss": 0.2331,
+      "step": 15433
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1293073493232032e-05,
+      "loss": 0.2381,
+      "step": 15434
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1292085156350756e-05,
+      "loss": 0.2246,
+      "step": 15435
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1291096806633811e-05,
+      "loss": 0.2656,
+      "step": 15436
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.129010844409101e-05,
+      "loss": 0.2351,
+      "step": 15437
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1289120068732185e-05,
+      "loss": 0.1748,
+      "step": 15438
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1288131680567145e-05,
+      "loss": 0.2786,
+      "step": 15439
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1287143279605713e-05,
+      "loss": 0.2198,
+      "step": 15440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1286154865857704e-05,
+      "loss": 0.3125,
+      "step": 15441
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1285166439332942e-05,
+      "loss": 0.2033,
+      "step": 15442
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1284178000041241e-05,
+      "loss": 0.1839,
+      "step": 15443
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1283189547992423e-05,
+      "loss": 0.2214,
+      "step": 15444
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.128220108319631e-05,
+      "loss": 0.2121,
+      "step": 15445
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1281212605662712e-05,
+      "loss": 0.2554,
+      "step": 15446
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.128022411540146e-05,
+      "loss": 0.1794,
+      "step": 15447
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.127923561242237e-05,
+      "loss": 0.2368,
+      "step": 15448
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1278247096735257e-05,
+      "loss": 0.2779,
+      "step": 15449
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1277258568349947e-05,
+      "loss": 0.2222,
+      "step": 15450
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1276270027276257e-05,
+      "loss": 0.2226,
+      "step": 15451
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1275281473524006e-05,
+      "loss": 0.2686,
+      "step": 15452
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1274292907103017e-05,
+      "loss": 0.2662,
+      "step": 15453
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1273304328023111e-05,
+      "loss": 0.2077,
+      "step": 15454
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1272315736294108e-05,
+      "loss": 0.2427,
+      "step": 15455
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1271327131925826e-05,
+      "loss": 0.2426,
+      "step": 15456
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1270338514928092e-05,
+      "loss": 0.2521,
+      "step": 15457
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1269349885310722e-05,
+      "loss": 0.2581,
+      "step": 15458
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1268361243083532e-05,
+      "loss": 0.221,
+      "step": 15459
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1267372588256355e-05,
+      "loss": 0.1734,
+      "step": 15460
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1266383920839005e-05,
+      "loss": 0.2191,
+      "step": 15461
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1265395240841308e-05,
+      "loss": 0.2544,
+      "step": 15462
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.126440654827308e-05,
+      "loss": 0.2497,
+      "step": 15463
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1263417843144147e-05,
+      "loss": 0.2203,
+      "step": 15464
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1262429125464328e-05,
+      "loss": 0.1989,
+      "step": 15465
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1261440395243449e-05,
+      "loss": 0.2368,
+      "step": 15466
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1260451652491329e-05,
+      "loss": 0.2258,
+      "step": 15467
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1259462897217785e-05,
+      "loss": 0.2358,
+      "step": 15468
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1258474129432653e-05,
+      "loss": 0.2345,
+      "step": 15469
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1257485349145745e-05,
+      "loss": 0.2948,
+      "step": 15470
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1256496556366883e-05,
+      "loss": 0.2233,
+      "step": 15471
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1255507751105899e-05,
+      "loss": 0.2376,
+      "step": 15472
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1254518933372608e-05,
+      "loss": 0.2308,
+      "step": 15473
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1253530103176833e-05,
+      "loss": 0.2075,
+      "step": 15474
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1252541260528398e-05,
+      "loss": 0.2011,
+      "step": 15475
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1251552405437131e-05,
+      "loss": 0.2966,
+      "step": 15476
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1250563537912848e-05,
+      "loss": 0.2331,
+      "step": 15477
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1249574657965378e-05,
+      "loss": 0.2347,
+      "step": 15478
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1248585765604544e-05,
+      "loss": 0.2601,
+      "step": 15479
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1247596860840165e-05,
+      "loss": 0.2138,
+      "step": 15480
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1246607943682072e-05,
+      "loss": 0.2005,
+      "step": 15481
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1245619014140083e-05,
+      "loss": 0.2126,
+      "step": 15482
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1244630072224024e-05,
+      "loss": 0.3256,
+      "step": 15483
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1243641117943723e-05,
+      "loss": 0.2151,
+      "step": 15484
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1242652151308997e-05,
+      "loss": 0.2115,
+      "step": 15485
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1241663172329675e-05,
+      "loss": 0.2046,
+      "step": 15486
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.124067418101558e-05,
+      "loss": 0.2456,
+      "step": 15487
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.123968517737654e-05,
+      "loss": 0.1734,
+      "step": 15488
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1238696161422378e-05,
+      "loss": 0.2773,
+      "step": 15489
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1237707133162917e-05,
+      "loss": 0.2126,
+      "step": 15490
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1236718092607985e-05,
+      "loss": 0.2657,
+      "step": 15491
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1235729039767403e-05,
+      "loss": 0.213,
+      "step": 15492
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1234739974651001e-05,
+      "loss": 0.3205,
+      "step": 15493
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1233750897268603e-05,
+      "loss": 0.2836,
+      "step": 15494
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1232761807630032e-05,
+      "loss": 0.2114,
+      "step": 15495
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1231772705745117e-05,
+      "loss": 0.2463,
+      "step": 15496
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1230783591623682e-05,
+      "loss": 0.2453,
+      "step": 15497
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1229794465275554e-05,
+      "loss": 0.196,
+      "step": 15498
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1228805326710559e-05,
+      "loss": 0.2382,
+      "step": 15499
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1227816175938521e-05,
+      "loss": 0.2656,
+      "step": 15500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.122682701296927e-05,
+      "loss": 0.2449,
+      "step": 15501
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1225837837812629e-05,
+      "loss": 0.1798,
+      "step": 15502
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1224848650478431e-05,
+      "loss": 0.2321,
+      "step": 15503
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1223859450976493e-05,
+      "loss": 0.1705,
+      "step": 15504
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1222870239316645e-05,
+      "loss": 0.2514,
+      "step": 15505
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1221881015508717e-05,
+      "loss": 0.254,
+      "step": 15506
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1220891779562534e-05,
+      "loss": 0.2267,
+      "step": 15507
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1219902531487926e-05,
+      "loss": 0.2042,
+      "step": 15508
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1218913271294713e-05,
+      "loss": 0.2057,
+      "step": 15509
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1217923998992733e-05,
+      "loss": 0.2502,
+      "step": 15510
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1216934714591803e-05,
+      "loss": 0.2483,
+      "step": 15511
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1215945418101756e-05,
+      "loss": 0.3002,
+      "step": 15512
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1214956109532422e-05,
+      "loss": 0.3048,
+      "step": 15513
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1213966788893622e-05,
+      "loss": 0.2742,
+      "step": 15514
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.121297745619519e-05,
+      "loss": 0.2963,
+      "step": 15515
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1211988111446951e-05,
+      "loss": 0.2448,
+      "step": 15516
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1210998754658734e-05,
+      "loss": 0.2207,
+      "step": 15517
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1210009385840371e-05,
+      "loss": 0.2489,
+      "step": 15518
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1209020005001682e-05,
+      "loss": 0.2652,
+      "step": 15519
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1208030612152502e-05,
+      "loss": 0.1894,
+      "step": 15520
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.120704120730266e-05,
+      "loss": 0.2547,
+      "step": 15521
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1206051790461982e-05,
+      "loss": 0.2202,
+      "step": 15522
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1205062361640298e-05,
+      "loss": 0.2295,
+      "step": 15523
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1204072920847435e-05,
+      "loss": 0.2405,
+      "step": 15524
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.120308346809323e-05,
+      "loss": 0.2301,
+      "step": 15525
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1202094003387501e-05,
+      "loss": 0.3334,
+      "step": 15526
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1201104526740086e-05,
+      "loss": 0.1816,
+      "step": 15527
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.120011503816081e-05,
+      "loss": 0.2534,
+      "step": 15528
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1199125537659504e-05,
+      "loss": 0.2408,
+      "step": 15529
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1198136025245999e-05,
+      "loss": 0.1966,
+      "step": 15530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1197146500930122e-05,
+      "loss": 0.212,
+      "step": 15531
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1196156964721707e-05,
+      "loss": 0.1981,
+      "step": 15532
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1195167416630579e-05,
+      "loss": 0.1955,
+      "step": 15533
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1194177856666576e-05,
+      "loss": 0.2416,
+      "step": 15534
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1193188284839518e-05,
+      "loss": 0.2957,
+      "step": 15535
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1192198701159243e-05,
+      "loss": 0.2172,
+      "step": 15536
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1191209105635583e-05,
+      "loss": 0.2275,
+      "step": 15537
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1190219498278363e-05,
+      "loss": 0.3069,
+      "step": 15538
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1189229879097414e-05,
+      "loss": 0.2382,
+      "step": 15539
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.118824024810257e-05,
+      "loss": 0.2662,
+      "step": 15540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1187250605303663e-05,
+      "loss": 0.2248,
+      "step": 15541
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.118626095071052e-05,
+      "loss": 0.2228,
+      "step": 15542
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1185271284332977e-05,
+      "loss": 0.2841,
+      "step": 15543
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1184281606180862e-05,
+      "loss": 0.2148,
+      "step": 15544
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1183291916264007e-05,
+      "loss": 0.2614,
+      "step": 15545
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1182302214592247e-05,
+      "loss": 0.177,
+      "step": 15546
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1181312501175409e-05,
+      "loss": 0.2538,
+      "step": 15547
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1180322776023325e-05,
+      "loss": 0.301,
+      "step": 15548
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1179333039145833e-05,
+      "loss": 0.2582,
+      "step": 15549
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1178343290552757e-05,
+      "loss": 0.1945,
+      "step": 15550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1177353530253937e-05,
+      "loss": 0.2493,
+      "step": 15551
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.11763637582592e-05,
+      "loss": 0.2794,
+      "step": 15552
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1175373974578378e-05,
+      "loss": 0.2153,
+      "step": 15553
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1174384179221306e-05,
+      "loss": 0.2448,
+      "step": 15554
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.117339437219782e-05,
+      "loss": 0.1735,
+      "step": 15555
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1172404553517748e-05,
+      "loss": 0.2036,
+      "step": 15556
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.117141472319092e-05,
+      "loss": 0.2048,
+      "step": 15557
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1170424881227176e-05,
+      "loss": 0.2633,
+      "step": 15558
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1169435027636348e-05,
+      "loss": 0.2489,
+      "step": 15559
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1168445162428266e-05,
+      "loss": 0.2956,
+      "step": 15560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1167455285612765e-05,
+      "loss": 0.2497,
+      "step": 15561
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.116646539719968e-05,
+      "loss": 0.1774,
+      "step": 15562
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1165475497198841e-05,
+      "loss": 0.2452,
+      "step": 15563
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1164485585620084e-05,
+      "loss": 0.2413,
+      "step": 15564
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1163495662473242e-05,
+      "loss": 0.2046,
+      "step": 15565
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1162505727768151e-05,
+      "loss": 0.2371,
+      "step": 15566
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1161515781514645e-05,
+      "loss": 0.2244,
+      "step": 15567
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1160525823722556e-05,
+      "loss": 0.2532,
+      "step": 15568
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1159535854401717e-05,
+      "loss": 0.2119,
+      "step": 15569
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1158545873561968e-05,
+      "loss": 0.2014,
+      "step": 15570
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.115755588121314e-05,
+      "loss": 0.2397,
+      "step": 15571
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1156565877365063e-05,
+      "loss": 0.2183,
+      "step": 15572
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1155575862027584e-05,
+      "loss": 0.2304,
+      "step": 15573
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1154585835210528e-05,
+      "loss": 0.1801,
+      "step": 15574
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.115359579692373e-05,
+      "loss": 0.264,
+      "step": 15575
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.115260574717703e-05,
+      "loss": 0.1987,
+      "step": 15576
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1151615685980259e-05,
+      "loss": 0.2305,
+      "step": 15577
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1150625613343255e-05,
+      "loss": 0.2313,
+      "step": 15578
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1149635529275854e-05,
+      "loss": 0.1978,
+      "step": 15579
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.114864543378789e-05,
+      "loss": 0.1852,
+      "step": 15580
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1147655326889195e-05,
+      "loss": 0.2708,
+      "step": 15581
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1146665208589612e-05,
+      "loss": 0.2182,
+      "step": 15582
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1145675078898977e-05,
+      "loss": 0.273,
+      "step": 15583
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1144684937827117e-05,
+      "loss": 0.1997,
+      "step": 15584
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1143694785383877e-05,
+      "loss": 0.2328,
+      "step": 15585
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1142704621579087e-05,
+      "loss": 0.2925,
+      "step": 15586
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.114171444642259e-05,
+      "loss": 0.2172,
+      "step": 15587
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1140724259924215e-05,
+      "loss": 0.1738,
+      "step": 15588
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1139734062093806e-05,
+      "loss": 0.2075,
+      "step": 15589
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1138743852941197e-05,
+      "loss": 0.2261,
+      "step": 15590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.113775363247622e-05,
+      "loss": 0.2189,
+      "step": 15591
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.113676340070872e-05,
+      "loss": 0.2911,
+      "step": 15592
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1135773157648526e-05,
+      "loss": 0.2103,
+      "step": 15593
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1134782903305479e-05,
+      "loss": 0.2348,
+      "step": 15594
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.113379263768942e-05,
+      "loss": 0.2569,
+      "step": 15595
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1132802360810179e-05,
+      "loss": 0.2298,
+      "step": 15596
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1131812072677599e-05,
+      "loss": 0.2512,
+      "step": 15597
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1130821773301517e-05,
+      "loss": 0.1751,
+      "step": 15598
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1129831462691767e-05,
+      "loss": 0.263,
+      "step": 15599
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1128841140858189e-05,
+      "loss": 0.2313,
+      "step": 15600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1127850807810623e-05,
+      "loss": 0.2143,
+      "step": 15601
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1126860463558907e-05,
+      "loss": 0.2367,
+      "step": 15602
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1125870108112872e-05,
+      "loss": 0.2591,
+      "step": 15603
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1124879741482366e-05,
+      "loss": 0.2663,
+      "step": 15604
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1123889363677223e-05,
+      "loss": 0.2543,
+      "step": 15605
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1122898974707276e-05,
+      "loss": 0.2499,
+      "step": 15606
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1121908574582377e-05,
+      "loss": 0.1873,
+      "step": 15607
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1120918163312353e-05,
+      "loss": 0.2806,
+      "step": 15608
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1119927740907046e-05,
+      "loss": 0.2731,
+      "step": 15609
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1118937307376296e-05,
+      "loss": 0.2583,
+      "step": 15610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1117946862729943e-05,
+      "loss": 0.3016,
+      "step": 15611
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1116956406977823e-05,
+      "loss": 0.1634,
+      "step": 15612
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1115965940129779e-05,
+      "loss": 0.2405,
+      "step": 15613
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1114975462195648e-05,
+      "loss": 0.2359,
+      "step": 15614
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1113984973185268e-05,
+      "loss": 0.2106,
+      "step": 15615
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1112994473108481e-05,
+      "loss": 0.2174,
+      "step": 15616
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1112003961975128e-05,
+      "loss": 0.2821,
+      "step": 15617
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1111013439795046e-05,
+      "loss": 0.2415,
+      "step": 15618
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1110022906578075e-05,
+      "loss": 0.1916,
+      "step": 15619
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1109032362334056e-05,
+      "loss": 0.2516,
+      "step": 15620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.110804180707283e-05,
+      "loss": 0.2935,
+      "step": 15621
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1107051240804237e-05,
+      "loss": 0.273,
+      "step": 15622
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1106060663538114e-05,
+      "loss": 0.2873,
+      "step": 15623
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1105070075284305e-05,
+      "loss": 0.2575,
+      "step": 15624
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1104079476052649e-05,
+      "loss": 0.2562,
+      "step": 15625
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.110308886585299e-05,
+      "loss": 0.2561,
+      "step": 15626
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1102098244695167e-05,
+      "loss": 0.2253,
+      "step": 15627
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1101107612589015e-05,
+      "loss": 0.295,
+      "step": 15628
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1100116969544385e-05,
+      "loss": 0.2802,
+      "step": 15629
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1099126315571111e-05,
+      "loss": 0.2406,
+      "step": 15630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1098135650679037e-05,
+      "loss": 0.224,
+      "step": 15631
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1097144974878003e-05,
+      "loss": 0.2394,
+      "step": 15632
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1096154288177852e-05,
+      "loss": 0.2073,
+      "step": 15633
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1095163590588424e-05,
+      "loss": 0.2638,
+      "step": 15634
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1094172882119562e-05,
+      "loss": 0.1859,
+      "step": 15635
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.109318216278111e-05,
+      "loss": 0.2062,
+      "step": 15636
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1092191432582902e-05,
+      "loss": 0.2202,
+      "step": 15637
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1091200691534788e-05,
+      "loss": 0.197,
+      "step": 15638
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1090209939646605e-05,
+      "loss": 0.2255,
+      "step": 15639
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1089219176928197e-05,
+      "loss": 0.1888,
+      "step": 15640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1088228403389412e-05,
+      "loss": 0.2481,
+      "step": 15641
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1087237619040083e-05,
+      "loss": 0.2449,
+      "step": 15642
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1086246823890057e-05,
+      "loss": 0.1858,
+      "step": 15643
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1085256017949175e-05,
+      "loss": 0.2231,
+      "step": 15644
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1084265201227284e-05,
+      "loss": 0.2811,
+      "step": 15645
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1083274373734221e-05,
+      "loss": 0.2421,
+      "step": 15646
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1082283535479832e-05,
+      "loss": 0.2635,
+      "step": 15647
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1081292686473964e-05,
+      "loss": 0.2191,
+      "step": 15648
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108030182672645e-05,
+      "loss": 0.2569,
+      "step": 15649
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1079310956247144e-05,
+      "loss": 0.2362,
+      "step": 15650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1078320075045882e-05,
+      "loss": 0.2251,
+      "step": 15651
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1077329183132508e-05,
+      "loss": 0.2378,
+      "step": 15652
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1076338280516873e-05,
+      "loss": 0.2234,
+      "step": 15653
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1075347367208812e-05,
+      "loss": 0.2354,
+      "step": 15654
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1074356443218175e-05,
+      "loss": 0.2536,
+      "step": 15655
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1073365508554801e-05,
+      "loss": 0.2825,
+      "step": 15656
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1072374563228536e-05,
+      "loss": 0.2027,
+      "step": 15657
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1071383607249225e-05,
+      "loss": 0.2179,
+      "step": 15658
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1070392640626713e-05,
+      "loss": 0.2517,
+      "step": 15659
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1069401663370843e-05,
+      "loss": 0.2851,
+      "step": 15660
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1068410675491455e-05,
+      "loss": 0.2438,
+      "step": 15661
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.10674196769984e-05,
+      "loss": 0.2507,
+      "step": 15662
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1066428667901523e-05,
+      "loss": 0.2456,
+      "step": 15663
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1065437648210664e-05,
+      "loss": 0.2439,
+      "step": 15664
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.106444661793567e-05,
+      "loss": 0.2489,
+      "step": 15665
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1063455577086386e-05,
+      "loss": 0.2239,
+      "step": 15666
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1062464525672656e-05,
+      "loss": 0.228,
+      "step": 15667
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1061473463704328e-05,
+      "loss": 0.2272,
+      "step": 15668
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1060482391191244e-05,
+      "loss": 0.2092,
+      "step": 15669
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1059491308143256e-05,
+      "loss": 0.3128,
+      "step": 15670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1058500214570197e-05,
+      "loss": 0.2253,
+      "step": 15671
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1057509110481928e-05,
+      "loss": 0.2966,
+      "step": 15672
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1056517995888283e-05,
+      "loss": 0.2455,
+      "step": 15673
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.105552687079911e-05,
+      "loss": 0.2373,
+      "step": 15674
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1054535735224257e-05,
+      "loss": 0.244,
+      "step": 15675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.105354458917357e-05,
+      "loss": 0.2505,
+      "step": 15676
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1052553432656896e-05,
+      "loss": 0.2378,
+      "step": 15677
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1051562265684077e-05,
+      "loss": 0.2538,
+      "step": 15678
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1050571088264968e-05,
+      "loss": 0.1806,
+      "step": 15679
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1049579900409402e-05,
+      "loss": 0.2166,
+      "step": 15680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1048588702127236e-05,
+      "loss": 0.1953,
+      "step": 15681
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1047597493428317e-05,
+      "loss": 0.2595,
+      "step": 15682
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1046606274322482e-05,
+      "loss": 0.2791,
+      "step": 15683
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.104561504481959e-05,
+      "loss": 0.2089,
+      "step": 15684
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1044623804929481e-05,
+      "loss": 0.2468,
+      "step": 15685
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1043632554662e-05,
+      "loss": 0.2645,
+      "step": 15686
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1042641294027003e-05,
+      "loss": 0.24,
+      "step": 15687
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1041650023034329e-05,
+      "loss": 0.2344,
+      "step": 15688
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1040658741693829e-05,
+      "loss": 0.2313,
+      "step": 15689
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1039667450015347e-05,
+      "loss": 0.2493,
+      "step": 15690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1038676148008737e-05,
+      "loss": 0.2272,
+      "step": 15691
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.103768483568384e-05,
+      "loss": 0.2365,
+      "step": 15692
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1036693513050505e-05,
+      "loss": 0.2736,
+      "step": 15693
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1035702180118585e-05,
+      "loss": 0.2097,
+      "step": 15694
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1034710836897922e-05,
+      "loss": 0.2548,
+      "step": 15695
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1033719483398368e-05,
+      "loss": 0.2684,
+      "step": 15696
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.103272811962977e-05,
+      "loss": 0.1997,
+      "step": 15697
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1031736745601975e-05,
+      "loss": 0.2303,
+      "step": 15698
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1030745361324834e-05,
+      "loss": 0.1979,
+      "step": 15699
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.102975396680819e-05,
+      "loss": 0.2824,
+      "step": 15700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1028762562061898e-05,
+      "loss": 0.222,
+      "step": 15701
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1027771147095803e-05,
+      "loss": 0.2242,
+      "step": 15702
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1026779721919759e-05,
+      "loss": 0.2348,
+      "step": 15703
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1025788286543603e-05,
+      "loss": 0.2483,
+      "step": 15704
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1024796840977196e-05,
+      "loss": 0.2735,
+      "step": 15705
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1023805385230386e-05,
+      "loss": 0.2218,
+      "step": 15706
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1022813919313016e-05,
+      "loss": 0.1932,
+      "step": 15707
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1021822443234937e-05,
+      "loss": 0.2028,
+      "step": 15708
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1020830957006001e-05,
+      "loss": 0.2468,
+      "step": 15709
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1019839460636055e-05,
+      "loss": 0.2408,
+      "step": 15710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1018847954134953e-05,
+      "loss": 0.1712,
+      "step": 15711
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1017856437512537e-05,
+      "loss": 0.2726,
+      "step": 15712
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1016864910778664e-05,
+      "loss": 0.1883,
+      "step": 15713
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.101587337394318e-05,
+      "loss": 0.2277,
+      "step": 15714
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1014881827015939e-05,
+      "loss": 0.1727,
+      "step": 15715
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.101389027000679e-05,
+      "loss": 0.2107,
+      "step": 15716
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1012898702925572e-05,
+      "loss": 0.2217,
+      "step": 15717
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1011907125782153e-05,
+      "loss": 0.2511,
+      "step": 15718
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1010915538586373e-05,
+      "loss": 0.1833,
+      "step": 15719
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1009923941348085e-05,
+      "loss": 0.1923,
+      "step": 15720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1008932334077138e-05,
+      "loss": 0.2321,
+      "step": 15721
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1007940716783384e-05,
+      "loss": 0.2252,
+      "step": 15722
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1006949089476675e-05,
+      "loss": 0.2187,
+      "step": 15723
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.100595745216686e-05,
+      "loss": 0.2464,
+      "step": 15724
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1004965804863795e-05,
+      "loss": 0.2704,
+      "step": 15725
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.100397414757732e-05,
+      "loss": 0.2377,
+      "step": 15726
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1002982480317295e-05,
+      "loss": 0.2166,
+      "step": 15727
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1001990803093572e-05,
+      "loss": 0.2262,
+      "step": 15728
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1000999115915995e-05,
+      "loss": 0.2467,
+      "step": 15729
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1000007418794425e-05,
+      "loss": 0.2301,
+      "step": 15730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0999015711738708e-05,
+      "loss": 0.2363,
+      "step": 15731
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0998023994758693e-05,
+      "loss": 0.211,
+      "step": 15732
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0997032267864238e-05,
+      "loss": 0.1663,
+      "step": 15733
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0996040531065191e-05,
+      "loss": 0.2429,
+      "step": 15734
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0995048784371405e-05,
+      "loss": 0.2427,
+      "step": 15735
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0994057027792733e-05,
+      "loss": 0.2648,
+      "step": 15736
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0993065261339028e-05,
+      "loss": 0.2486,
+      "step": 15737
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0992073485020138e-05,
+      "loss": 0.2547,
+      "step": 15738
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0991081698845917e-05,
+      "loss": 0.2182,
+      "step": 15739
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0990089902826221e-05,
+      "loss": 0.2385,
+      "step": 15740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0989098096970899e-05,
+      "loss": 0.1839,
+      "step": 15741
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0988106281289808e-05,
+      "loss": 0.2197,
+      "step": 15742
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0987114455792793e-05,
+      "loss": 0.2127,
+      "step": 15743
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0986122620489715e-05,
+      "loss": 0.3047,
+      "step": 15744
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.098513077539042e-05,
+      "loss": 0.2335,
+      "step": 15745
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0984138920504766e-05,
+      "loss": 0.2062,
+      "step": 15746
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0983147055842602e-05,
+      "loss": 0.2961,
+      "step": 15747
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0982155181413788e-05,
+      "loss": 0.2717,
+      "step": 15748
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0981163297228172e-05,
+      "loss": 0.2368,
+      "step": 15749
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.2042,
+      "step": 15750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0979179499625949e-05,
+      "loss": 0.2027,
+      "step": 15751
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0978187586229052e-05,
+      "loss": 0.223,
+      "step": 15752
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0977195663114768e-05,
+      "loss": 0.2433,
+      "step": 15753
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0976203730292952e-05,
+      "loss": 0.2717,
+      "step": 15754
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0975211787773455e-05,
+      "loss": 0.2171,
+      "step": 15755
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0974219835566133e-05,
+      "loss": 0.2341,
+      "step": 15756
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0973227873680843e-05,
+      "loss": 0.2106,
+      "step": 15757
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0972235902127435e-05,
+      "loss": 0.2447,
+      "step": 15758
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0971243920915769e-05,
+      "loss": 0.2344,
+      "step": 15759
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0970251930055688e-05,
+      "loss": 0.2245,
+      "step": 15760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0969259929557061e-05,
+      "loss": 0.2346,
+      "step": 15761
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0968267919429732e-05,
+      "loss": 0.2164,
+      "step": 15762
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0967275899683559e-05,
+      "loss": 0.2748,
+      "step": 15763
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0966283870328399e-05,
+      "loss": 0.2586,
+      "step": 15764
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0965291831374104e-05,
+      "loss": 0.2317,
+      "step": 15765
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.096429978283053e-05,
+      "loss": 0.2221,
+      "step": 15766
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.096330772470753e-05,
+      "loss": 0.2487,
+      "step": 15767
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0962315657014964e-05,
+      "loss": 0.2476,
+      "step": 15768
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.096132357976268e-05,
+      "loss": 0.215,
+      "step": 15769
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0960331492960543e-05,
+      "loss": 0.2366,
+      "step": 15770
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0959339396618402e-05,
+      "loss": 0.2811,
+      "step": 15771
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0958347290746112e-05,
+      "loss": 0.295,
+      "step": 15772
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0957355175353535e-05,
+      "loss": 0.2568,
+      "step": 15773
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0956363050450518e-05,
+      "loss": 0.2612,
+      "step": 15774
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.095537091604692e-05,
+      "loss": 0.3251,
+      "step": 15775
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.09543787721526e-05,
+      "loss": 0.2339,
+      "step": 15776
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0953386618777413e-05,
+      "loss": 0.195,
+      "step": 15777
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0952394455931214e-05,
+      "loss": 0.1904,
+      "step": 15778
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0951402283623858e-05,
+      "loss": 0.2517,
+      "step": 15779
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0950410101865204e-05,
+      "loss": 0.2079,
+      "step": 15780
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0949417910665105e-05,
+      "loss": 0.2251,
+      "step": 15781
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0948425710033422e-05,
+      "loss": 0.2547,
+      "step": 15782
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.094743349998001e-05,
+      "loss": 0.2595,
+      "step": 15783
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0946441280514721e-05,
+      "loss": 0.2727,
+      "step": 15784
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0945449051647417e-05,
+      "loss": 0.3351,
+      "step": 15785
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0944456813387956e-05,
+      "loss": 0.2324,
+      "step": 15786
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0943464565746192e-05,
+      "loss": 0.2773,
+      "step": 15787
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.094247230873198e-05,
+      "loss": 0.2018,
+      "step": 15788
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0941480042355182e-05,
+      "loss": 0.2379,
+      "step": 15789
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0940487766625652e-05,
+      "loss": 0.1757,
+      "step": 15790
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0939495481553246e-05,
+      "loss": 0.2546,
+      "step": 15791
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0938503187147827e-05,
+      "loss": 0.2091,
+      "step": 15792
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0937510883419247e-05,
+      "loss": 0.1818,
+      "step": 15793
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0936518570377367e-05,
+      "loss": 0.2035,
+      "step": 15794
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0935526248032047e-05,
+      "loss": 0.2641,
+      "step": 15795
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0934533916393137e-05,
+      "loss": 0.2145,
+      "step": 15796
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0933541575470498e-05,
+      "loss": 0.2219,
+      "step": 15797
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0932549225273994e-05,
+      "loss": 0.248,
+      "step": 15798
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0931556865813475e-05,
+      "loss": 0.207,
+      "step": 15799
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0930564497098803e-05,
+      "loss": 0.2606,
+      "step": 15800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0929572119139837e-05,
+      "loss": 0.2113,
+      "step": 15801
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0928579731946433e-05,
+      "loss": 0.2093,
+      "step": 15802
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.092758733552845e-05,
+      "loss": 0.3014,
+      "step": 15803
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0926594929895748e-05,
+      "loss": 0.1991,
+      "step": 15804
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0925602515058187e-05,
+      "loss": 0.2441,
+      "step": 15805
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0924610091025621e-05,
+      "loss": 0.2224,
+      "step": 15806
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0923617657807913e-05,
+      "loss": 0.2562,
+      "step": 15807
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0922625215414916e-05,
+      "loss": 0.2462,
+      "step": 15808
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0921632763856495e-05,
+      "loss": 0.2987,
+      "step": 15809
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0920640303142512e-05,
+      "loss": 0.2459,
+      "step": 15810
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0919647833282816e-05,
+      "loss": 0.2563,
+      "step": 15811
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0918655354287273e-05,
+      "loss": 0.248,
+      "step": 15812
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0917662866165741e-05,
+      "loss": 0.2002,
+      "step": 15813
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0916670368928081e-05,
+      "loss": 0.2561,
+      "step": 15814
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0915677862584149e-05,
+      "loss": 0.207,
+      "step": 15815
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0914685347143807e-05,
+      "loss": 0.248,
+      "step": 15816
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0913692822616918e-05,
+      "loss": 0.2596,
+      "step": 15817
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0912700289013332e-05,
+      "loss": 0.2066,
+      "step": 15818
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.091170774634292e-05,
+      "loss": 0.1909,
+      "step": 15819
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0910715194615535e-05,
+      "loss": 0.2138,
+      "step": 15820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0909722633841038e-05,
+      "loss": 0.2281,
+      "step": 15821
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0908730064029294e-05,
+      "loss": 0.2593,
+      "step": 15822
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0907737485190157e-05,
+      "loss": 0.2556,
+      "step": 15823
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.090674489733349e-05,
+      "loss": 0.2586,
+      "step": 15824
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0905752300469155e-05,
+      "loss": 0.259,
+      "step": 15825
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.090475969460701e-05,
+      "loss": 0.2365,
+      "step": 15826
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0903767079756916e-05,
+      "loss": 0.242,
+      "step": 15827
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0902774455928734e-05,
+      "loss": 0.2138,
+      "step": 15828
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0901781823132326e-05,
+      "loss": 0.2678,
+      "step": 15829
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0900789181377551e-05,
+      "loss": 0.2624,
+      "step": 15830
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0899796530674274e-05,
+      "loss": 0.2659,
+      "step": 15831
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0898803871032351e-05,
+      "loss": 0.2407,
+      "step": 15832
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0897811202461643e-05,
+      "loss": 0.2635,
+      "step": 15833
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0896818524972015e-05,
+      "loss": 0.2902,
+      "step": 15834
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0895825838573326e-05,
+      "loss": 0.2508,
+      "step": 15835
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.089483314327544e-05,
+      "loss": 0.2761,
+      "step": 15836
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0893840439088214e-05,
+      "loss": 0.217,
+      "step": 15837
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0892847726021514e-05,
+      "loss": 0.2084,
+      "step": 15838
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0891855004085203e-05,
+      "loss": 0.2813,
+      "step": 15839
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0890862273289133e-05,
+      "loss": 0.2324,
+      "step": 15840
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0889869533643178e-05,
+      "loss": 0.2299,
+      "step": 15841
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0888876785157192e-05,
+      "loss": 0.2344,
+      "step": 15842
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.088788402784104e-05,
+      "loss": 0.3214,
+      "step": 15843
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0886891261704584e-05,
+      "loss": 0.2327,
+      "step": 15844
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0885898486757686e-05,
+      "loss": 0.2665,
+      "step": 15845
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0884905703010207e-05,
+      "loss": 0.2349,
+      "step": 15846
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0883912910472012e-05,
+      "loss": 0.1934,
+      "step": 15847
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0882920109152962e-05,
+      "loss": 0.2313,
+      "step": 15848
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0881927299062918e-05,
+      "loss": 0.2486,
+      "step": 15849
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0880934480211744e-05,
+      "loss": 0.2045,
+      "step": 15850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0879941652609305e-05,
+      "loss": 0.2186,
+      "step": 15851
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087894881626546e-05,
+      "loss": 0.2433,
+      "step": 15852
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0877955971190075e-05,
+      "loss": 0.188,
+      "step": 15853
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0876963117393012e-05,
+      "loss": 0.241,
+      "step": 15854
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087597025488413e-05,
+      "loss": 0.2097,
+      "step": 15855
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.08749773836733e-05,
+      "loss": 0.2531,
+      "step": 15856
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0873984503770381e-05,
+      "loss": 0.2792,
+      "step": 15857
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0872991615185235e-05,
+      "loss": 0.2799,
+      "step": 15858
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0871998717927728e-05,
+      "loss": 0.2654,
+      "step": 15859
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0871005812007725e-05,
+      "loss": 0.2631,
+      "step": 15860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087001289743508e-05,
+      "loss": 0.2368,
+      "step": 15861
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0869019974219668e-05,
+      "loss": 0.2459,
+      "step": 15862
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.086802704237135e-05,
+      "loss": 0.23,
+      "step": 15863
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0867034101899985e-05,
+      "loss": 0.1749,
+      "step": 15864
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0866041152815443e-05,
+      "loss": 0.2339,
+      "step": 15865
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0865048195127585e-05,
+      "loss": 0.2136,
+      "step": 15866
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0864055228846275e-05,
+      "loss": 0.2058,
+      "step": 15867
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0863062253981378e-05,
+      "loss": 0.2255,
+      "step": 15868
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0862069270542756e-05,
+      "loss": 0.2951,
+      "step": 15869
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0861076278540277e-05,
+      "loss": 0.1813,
+      "step": 15870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0860083277983804e-05,
+      "loss": 0.2326,
+      "step": 15871
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0859090268883203e-05,
+      "loss": 0.1943,
+      "step": 15872
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0858097251248331e-05,
+      "loss": 0.2727,
+      "step": 15873
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0857104225089064e-05,
+      "loss": 0.2908,
+      "step": 15874
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0856111190415262e-05,
+      "loss": 0.2199,
+      "step": 15875
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0855118147236788e-05,
+      "loss": 0.1992,
+      "step": 15876
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0854125095563506e-05,
+      "loss": 0.2551,
+      "step": 15877
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0853132035405285e-05,
+      "loss": 0.3042,
+      "step": 15878
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0852138966771987e-05,
+      "loss": 0.2395,
+      "step": 15879
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.085114588967348e-05,
+      "loss": 0.2361,
+      "step": 15880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.085015280411963e-05,
+      "loss": 0.2661,
+      "step": 15881
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0849159710120296e-05,
+      "loss": 0.2487,
+      "step": 15882
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.084816660768535e-05,
+      "loss": 0.2271,
+      "step": 15883
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0847173496824659e-05,
+      "loss": 0.2302,
+      "step": 15884
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0846180377548082e-05,
+      "loss": 0.1897,
+      "step": 15885
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0845187249865485e-05,
+      "loss": 0.2459,
+      "step": 15886
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.084419411378674e-05,
+      "loss": 0.2369,
+      "step": 15887
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0843200969321708e-05,
+      "loss": 0.2562,
+      "step": 15888
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0842207816480257e-05,
+      "loss": 0.2454,
+      "step": 15889
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0841214655272252e-05,
+      "loss": 0.2307,
+      "step": 15890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0840221485707561e-05,
+      "loss": 0.3004,
+      "step": 15891
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0839228307796046e-05,
+      "loss": 0.1691,
+      "step": 15892
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0838235121547578e-05,
+      "loss": 0.202,
+      "step": 15893
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0837241926972021e-05,
+      "loss": 0.2911,
+      "step": 15894
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0836248724079241e-05,
+      "loss": 0.2337,
+      "step": 15895
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0835255512879108e-05,
+      "loss": 0.2888,
+      "step": 15896
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0834262293381486e-05,
+      "loss": 0.1693,
+      "step": 15897
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.083326906559624e-05,
+      "loss": 0.2088,
+      "step": 15898
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.083227582953324e-05,
+      "loss": 0.2128,
+      "step": 15899
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.083128258520235e-05,
+      "loss": 0.2226,
+      "step": 15900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.083028933261344e-05,
+      "loss": 0.2125,
+      "step": 15901
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0829296071776375e-05,
+      "loss": 0.2537,
+      "step": 15902
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0828302802701021e-05,
+      "loss": 0.2839,
+      "step": 15903
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082730952539725e-05,
+      "loss": 0.2847,
+      "step": 15904
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0826316239874923e-05,
+      "loss": 0.2217,
+      "step": 15905
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0825322946143915e-05,
+      "loss": 0.2445,
+      "step": 15906
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0824329644214083e-05,
+      "loss": 0.1944,
+      "step": 15907
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0823336334095302e-05,
+      "loss": 0.2103,
+      "step": 15908
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0822343015797443e-05,
+      "loss": 0.2393,
+      "step": 15909
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0821349689330363e-05,
+      "loss": 0.2678,
+      "step": 15910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082035635470394e-05,
+      "loss": 0.2351,
+      "step": 15911
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0819363011928035e-05,
+      "loss": 0.2291,
+      "step": 15912
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.081836966101252e-05,
+      "loss": 0.2465,
+      "step": 15913
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.081737630196726e-05,
+      "loss": 0.2206,
+      "step": 15914
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0816382934802123e-05,
+      "loss": 0.2278,
+      "step": 15915
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0815389559526981e-05,
+      "loss": 0.2442,
+      "step": 15916
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0814396176151701e-05,
+      "loss": 0.2344,
+      "step": 15917
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.081340278468615e-05,
+      "loss": 0.2612,
+      "step": 15918
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0812409385140194e-05,
+      "loss": 0.1797,
+      "step": 15919
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0811415977523705e-05,
+      "loss": 0.2381,
+      "step": 15920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0810422561846552e-05,
+      "loss": 0.2172,
+      "step": 15921
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0809429138118603e-05,
+      "loss": 0.1973,
+      "step": 15922
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0808435706349725e-05,
+      "loss": 0.1875,
+      "step": 15923
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0807442266549789e-05,
+      "loss": 0.2424,
+      "step": 15924
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.080644881872866e-05,
+      "loss": 0.2493,
+      "step": 15925
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0805455362896213e-05,
+      "loss": 0.2486,
+      "step": 15926
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0804461899062313e-05,
+      "loss": 0.2108,
+      "step": 15927
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0803468427236832e-05,
+      "loss": 0.2213,
+      "step": 15928
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0802474947429634e-05,
+      "loss": 0.1871,
+      "step": 15929
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0801481459650595e-05,
+      "loss": 0.222,
+      "step": 15930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0800487963909578e-05,
+      "loss": 0.1615,
+      "step": 15931
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0799494460216455e-05,
+      "loss": 0.2697,
+      "step": 15932
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0798500948581102e-05,
+      "loss": 0.2792,
+      "step": 15933
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0797507429013376e-05,
+      "loss": 0.2331,
+      "step": 15934
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0796513901523156e-05,
+      "loss": 0.2066,
+      "step": 15935
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.079552036612031e-05,
+      "loss": 0.3005,
+      "step": 15936
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0794526822814704e-05,
+      "loss": 0.225,
+      "step": 15937
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0793533271616212e-05,
+      "loss": 0.245,
+      "step": 15938
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0792539712534704e-05,
+      "loss": 0.2363,
+      "step": 15939
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0791546145580051e-05,
+      "loss": 0.2085,
+      "step": 15940
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0790552570762116e-05,
+      "loss": 0.2054,
+      "step": 15941
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.078955898809078e-05,
+      "loss": 0.2426,
+      "step": 15942
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0788565397575905e-05,
+      "loss": 0.2606,
+      "step": 15943
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.078757179922736e-05,
+      "loss": 0.1957,
+      "step": 15944
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0786578193055029e-05,
+      "loss": 0.2353,
+      "step": 15945
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0785584579068767e-05,
+      "loss": 0.1817,
+      "step": 15946
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0784590957278452e-05,
+      "loss": 0.2276,
+      "step": 15947
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0783597327693952e-05,
+      "loss": 0.2917,
+      "step": 15948
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0782603690325139e-05,
+      "loss": 0.3276,
+      "step": 15949
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0781610045181887e-05,
+      "loss": 0.2363,
+      "step": 15950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0780616392274062e-05,
+      "loss": 0.2583,
+      "step": 15951
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0779622731611541e-05,
+      "loss": 0.2404,
+      "step": 15952
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0778629063204184e-05,
+      "loss": 0.2276,
+      "step": 15953
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0777635387061875e-05,
+      "loss": 0.2421,
+      "step": 15954
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.077664170319448e-05,
+      "loss": 0.1713,
+      "step": 15955
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0775648011611868e-05,
+      "loss": 0.2274,
+      "step": 15956
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.077465431232391e-05,
+      "loss": 0.2771,
+      "step": 15957
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0773660605340484e-05,
+      "loss": 0.2612,
+      "step": 15958
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0772666890671454e-05,
+      "loss": 0.2981,
+      "step": 15959
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0771673168326699e-05,
+      "loss": 0.2674,
+      "step": 15960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0770679438316085e-05,
+      "loss": 0.2299,
+      "step": 15961
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0769685700649484e-05,
+      "loss": 0.2549,
+      "step": 15962
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076869195533677e-05,
+      "loss": 0.212,
+      "step": 15963
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0767698202387818e-05,
+      "loss": 0.2879,
+      "step": 15964
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0766704441812493e-05,
+      "loss": 0.2063,
+      "step": 15965
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0765710673620668e-05,
+      "loss": 0.2316,
+      "step": 15966
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0764716897822223e-05,
+      "loss": 0.21,
+      "step": 15967
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076372311442702e-05,
+      "loss": 0.2017,
+      "step": 15968
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076272932344494e-05,
+      "loss": 0.1811,
+      "step": 15969
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076173552488585e-05,
+      "loss": 0.2423,
+      "step": 15970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0760741718759623e-05,
+      "loss": 0.2069,
+      "step": 15971
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0759747905076134e-05,
+      "loss": 0.2364,
+      "step": 15972
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0758754083845252e-05,
+      "loss": 0.2229,
+      "step": 15973
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0757760255076856e-05,
+      "loss": 0.224,
+      "step": 15974
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0756766418780811e-05,
+      "loss": 0.2694,
+      "step": 15975
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0755772574966996e-05,
+      "loss": 0.2628,
+      "step": 15976
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.075477872364528e-05,
+      "loss": 0.2303,
+      "step": 15977
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0753784864825537e-05,
+      "loss": 0.2609,
+      "step": 15978
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0752790998517642e-05,
+      "loss": 0.2746,
+      "step": 15979
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0751797124731464e-05,
+      "loss": 0.2377,
+      "step": 15980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0750803243476882e-05,
+      "loss": 0.2402,
+      "step": 15981
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0749809354763764e-05,
+      "loss": 0.2696,
+      "step": 15982
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0748815458601988e-05,
+      "loss": 0.2611,
+      "step": 15983
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0747821555001422e-05,
+      "loss": 0.2336,
+      "step": 15984
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0746827643971942e-05,
+      "loss": 0.2431,
+      "step": 15985
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0745833725523425e-05,
+      "loss": 0.2333,
+      "step": 15986
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0744839799665737e-05,
+      "loss": 0.2001,
+      "step": 15987
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0743845866408762e-05,
+      "loss": 0.2178,
+      "step": 15988
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0742851925762365e-05,
+      "loss": 0.183,
+      "step": 15989
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0741857977736423e-05,
+      "loss": 0.2726,
+      "step": 15990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0740864022340811e-05,
+      "loss": 0.1772,
+      "step": 15991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0739870059585402e-05,
+      "loss": 0.2049,
+      "step": 15992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073887608948007e-05,
+      "loss": 0.2171,
+      "step": 15993
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073788211203469e-05,
+      "loss": 0.2364,
+      "step": 15994
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0736888127259132e-05,
+      "loss": 0.2306,
+      "step": 15995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0735894135163279e-05,
+      "loss": 0.2571,
+      "step": 15996
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0734900135756999e-05,
+      "loss": 0.2634,
+      "step": 15997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0733906129050166e-05,
+      "loss": 0.199,
+      "step": 15998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0732912115052658e-05,
+      "loss": 0.2236,
+      "step": 15999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0731918093774347e-05,
+      "loss": 0.2304,
+      "step": 16000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0730924065225108e-05,
+      "loss": 0.3063,
+      "step": 16001
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0729930029414817e-05,
+      "loss": 0.2283,
+      "step": 16002
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0728935986353345e-05,
+      "loss": 0.2098,
+      "step": 16003
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0727941936050574e-05,
+      "loss": 0.2077,
+      "step": 16004
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0726947878516373e-05,
+      "loss": 0.2606,
+      "step": 16005
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.072595381376062e-05,
+      "loss": 0.2183,
+      "step": 16006
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.072495974179319e-05,
+      "loss": 0.2271,
+      "step": 16007
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0723965662623958e-05,
+      "loss": 0.3225,
+      "step": 16008
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0722971576262795e-05,
+      "loss": 0.185,
+      "step": 16009
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0721977482719583e-05,
+      "loss": 0.2559,
+      "step": 16010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0720983382004193e-05,
+      "loss": 0.2166,
+      "step": 16011
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0719989274126503e-05,
+      "loss": 0.2384,
+      "step": 16012
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0718995159096386e-05,
+      "loss": 0.2733,
+      "step": 16013
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.071800103692372e-05,
+      "loss": 0.1629,
+      "step": 16014
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0717006907618377e-05,
+      "loss": 0.2213,
+      "step": 16015
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0716012771190239e-05,
+      "loss": 0.1881,
+      "step": 16016
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0715018627649177e-05,
+      "loss": 0.221,
+      "step": 16017
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0714024477005064e-05,
+      "loss": 0.2441,
+      "step": 16018
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0713030319267783e-05,
+      "loss": 0.2645,
+      "step": 16019
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0712036154447212e-05,
+      "loss": 0.2096,
+      "step": 16020
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0711041982553215e-05,
+      "loss": 0.2562,
+      "step": 16021
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.071004780359568e-05,
+      "loss": 0.2451,
+      "step": 16022
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0709053617584475e-05,
+      "loss": 0.2697,
+      "step": 16023
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070805942452948e-05,
+      "loss": 0.3126,
+      "step": 16024
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0707065224440575e-05,
+      "loss": 0.2497,
+      "step": 16025
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070607101732763e-05,
+      "loss": 0.2512,
+      "step": 16026
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0705076803200525e-05,
+      "loss": 0.2214,
+      "step": 16027
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0704082582069136e-05,
+      "loss": 0.2304,
+      "step": 16028
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070308835394334e-05,
+      "loss": 0.2675,
+      "step": 16029
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070209411883301e-05,
+      "loss": 0.1923,
+      "step": 16030
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070109987674803e-05,
+      "loss": 0.2162,
+      "step": 16031
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0700105627698273e-05,
+      "loss": 0.1881,
+      "step": 16032
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0699111371693613e-05,
+      "loss": 0.2522,
+      "step": 16033
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0698117108743933e-05,
+      "loss": 0.2304,
+      "step": 16034
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0697122838859106e-05,
+      "loss": 0.2045,
+      "step": 16035
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0696128562049011e-05,
+      "loss": 0.2805,
+      "step": 16036
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0695134278323522e-05,
+      "loss": 0.238,
+      "step": 16037
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069413998769252e-05,
+      "loss": 0.1729,
+      "step": 16038
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069314569016588e-05,
+      "loss": 0.2463,
+      "step": 16039
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069215138575348e-05,
+      "loss": 0.2099,
+      "step": 16040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0691157074465202e-05,
+      "loss": 0.2077,
+      "step": 16041
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0690162756310913e-05,
+      "loss": 0.2205,
+      "step": 16042
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0689168431300501e-05,
+      "loss": 0.2812,
+      "step": 16043
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0688174099443842e-05,
+      "loss": 0.2466,
+      "step": 16044
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0687179760750808e-05,
+      "loss": 0.2719,
+      "step": 16045
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0686185415231282e-05,
+      "loss": 0.2223,
+      "step": 16046
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068519106289514e-05,
+      "loss": 0.2357,
+      "step": 16047
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0684196703752258e-05,
+      "loss": 0.2876,
+      "step": 16048
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068320233781252e-05,
+      "loss": 0.2343,
+      "step": 16049
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0682207965085798e-05,
+      "loss": 0.2247,
+      "step": 16050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0681213585581973e-05,
+      "loss": 0.248,
+      "step": 16051
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0680219199310923e-05,
+      "loss": 0.2501,
+      "step": 16052
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.067922480628253e-05,
+      "loss": 0.231,
+      "step": 16053
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0678230406506662e-05,
+      "loss": 0.2195,
+      "step": 16054
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0677235999993205e-05,
+      "loss": 0.2657,
+      "step": 16055
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.067624158675204e-05,
+      "loss": 0.2437,
+      "step": 16056
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.067524716679304e-05,
+      "loss": 0.2443,
+      "step": 16057
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0674252740126086e-05,
+      "loss": 0.2595,
+      "step": 16058
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0673258306761056e-05,
+      "loss": 0.2396,
+      "step": 16059
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0672263866707828e-05,
+      "loss": 0.2534,
+      "step": 16060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0671269419976285e-05,
+      "loss": 0.1631,
+      "step": 16061
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.06702749665763e-05,
+      "loss": 0.2332,
+      "step": 16062
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0669280506517759e-05,
+      "loss": 0.2329,
+      "step": 16063
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0668286039810529e-05,
+      "loss": 0.1948,
+      "step": 16064
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0667291566464504e-05,
+      "loss": 0.2536,
+      "step": 16065
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0666297086489555e-05,
+      "loss": 0.2296,
+      "step": 16066
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.066530259989556e-05,
+      "loss": 0.1503,
+      "step": 16067
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0664308106692403e-05,
+      "loss": 0.2819,
+      "step": 16068
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.066331360688996e-05,
+      "loss": 0.3004,
+      "step": 16069
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0662319100498111e-05,
+      "loss": 0.2138,
+      "step": 16070
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0661324587526738e-05,
+      "loss": 0.2889,
+      "step": 16071
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0660330067985716e-05,
+      "loss": 0.195,
+      "step": 16072
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0659335541884927e-05,
+      "loss": 0.1915,
+      "step": 16073
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0658341009234255e-05,
+      "loss": 0.2829,
+      "step": 16074
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0657346470043575e-05,
+      "loss": 0.317,
+      "step": 16075
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0656351924322763e-05,
+      "loss": 0.2568,
+      "step": 16076
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0655357372081704e-05,
+      "loss": 0.2827,
+      "step": 16077
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.065436281333028e-05,
+      "loss": 0.2387,
+      "step": 16078
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0653368248078369e-05,
+      "loss": 0.2554,
+      "step": 16079
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.065237367633585e-05,
+      "loss": 0.2156,
+      "step": 16080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0651379098112602e-05,
+      "loss": 0.2154,
+      "step": 16081
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0650384513418506e-05,
+      "loss": 0.2453,
+      "step": 16082
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0649389922263444e-05,
+      "loss": 0.1992,
+      "step": 16083
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0648395324657298e-05,
+      "loss": 0.1866,
+      "step": 16084
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0647400720609942e-05,
+      "loss": 0.2211,
+      "step": 16085
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0646406110131263e-05,
+      "loss": 0.2404,
+      "step": 16086
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0645411493231137e-05,
+      "loss": 0.2192,
+      "step": 16087
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0644416869919447e-05,
+      "loss": 0.2711,
+      "step": 16088
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0643422240206071e-05,
+      "loss": 0.1991,
+      "step": 16089
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0642427604100895e-05,
+      "loss": 0.2223,
+      "step": 16090
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0641432961613796e-05,
+      "loss": 0.2909,
+      "step": 16091
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0640438312754655e-05,
+      "loss": 0.2463,
+      "step": 16092
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0639443657533353e-05,
+      "loss": 0.2772,
+      "step": 16093
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0638448995959772e-05,
+      "loss": 0.2061,
+      "step": 16094
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0637454328043792e-05,
+      "loss": 0.2188,
+      "step": 16095
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0636459653795294e-05,
+      "loss": 0.2034,
+      "step": 16096
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0635464973224161e-05,
+      "loss": 0.2657,
+      "step": 16097
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.063447028634027e-05,
+      "loss": 0.2382,
+      "step": 16098
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0633475593153508e-05,
+      "loss": 0.2651,
+      "step": 16099
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0632480893673753e-05,
+      "loss": 0.2413,
+      "step": 16100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0631486187910885e-05,
+      "loss": 0.1988,
+      "step": 16101
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0630491475874792e-05,
+      "loss": 0.2133,
+      "step": 16102
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0629496757575346e-05,
+      "loss": 0.2738,
+      "step": 16103
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0628502033022434e-05,
+      "loss": 0.2042,
+      "step": 16104
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0627507302225938e-05,
+      "loss": 0.2765,
+      "step": 16105
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0626512565195741e-05,
+      "loss": 0.2148,
+      "step": 16106
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.062551782194172e-05,
+      "loss": 0.2693,
+      "step": 16107
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.062452307247376e-05,
+      "loss": 0.1811,
+      "step": 16108
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0623528316801746e-05,
+      "loss": 0.2222,
+      "step": 16109
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0622533554935552e-05,
+      "loss": 0.2382,
+      "step": 16110
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0621538786885069e-05,
+      "loss": 0.2708,
+      "step": 16111
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.062054401266017e-05,
+      "loss": 0.1949,
+      "step": 16112
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061954923227074e-05,
+      "loss": 0.3003,
+      "step": 16113
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061855444572667e-05,
+      "loss": 0.2025,
+      "step": 16114
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0617559653037833e-05,
+      "loss": 0.2179,
+      "step": 16115
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061656485421411e-05,
+      "loss": 0.2442,
+      "step": 16116
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061557004926539e-05,
+      "loss": 0.2384,
+      "step": 16117
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0614575238201552e-05,
+      "loss": 0.2424,
+      "step": 16118
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0613580421032478e-05,
+      "loss": 0.2434,
+      "step": 16119
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0612585597768052e-05,
+      "loss": 0.2137,
+      "step": 16120
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0611590768418159e-05,
+      "loss": 0.2109,
+      "step": 16121
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0610595932992674e-05,
+      "loss": 0.2453,
+      "step": 16122
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0609601091501488e-05,
+      "loss": 0.2808,
+      "step": 16123
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.060860624395448e-05,
+      "loss": 0.2747,
+      "step": 16124
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0607611390361534e-05,
+      "loss": 0.2536,
+      "step": 16125
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0606616530732532e-05,
+      "loss": 0.1889,
+      "step": 16126
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0605621665077358e-05,
+      "loss": 0.1985,
+      "step": 16127
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0604626793405892e-05,
+      "loss": 0.1919,
+      "step": 16128
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0603631915728022e-05,
+      "loss": 0.2434,
+      "step": 16129
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0602637032053626e-05,
+      "loss": 0.2062,
+      "step": 16130
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0601642142392591e-05,
+      "loss": 0.3844,
+      "step": 16131
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.06006472467548e-05,
+      "loss": 0.2159,
+      "step": 16132
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0599652345150137e-05,
+      "loss": 0.2842,
+      "step": 16133
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0598657437588484e-05,
+      "loss": 0.2643,
+      "step": 16134
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0597662524079721e-05,
+      "loss": 0.1981,
+      "step": 16135
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0596667604633739e-05,
+      "loss": 0.2707,
+      "step": 16136
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0595672679260416e-05,
+      "loss": 0.1843,
+      "step": 16137
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0594677747969638e-05,
+      "loss": 0.2133,
+      "step": 16138
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0593682810771286e-05,
+      "loss": 0.2734,
+      "step": 16139
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.059268786767525e-05,
+      "loss": 0.2609,
+      "step": 16140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0591692918691405e-05,
+      "loss": 0.2344,
+      "step": 16141
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0590697963829641e-05,
+      "loss": 0.2808,
+      "step": 16142
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0589703003099845e-05,
+      "loss": 0.2612,
+      "step": 16143
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.058870803651189e-05,
+      "loss": 0.2352,
+      "step": 16144
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0587713064075673e-05,
+      "loss": 0.2668,
+      "step": 16145
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0586718085801069e-05,
+      "loss": 0.2515,
+      "step": 16146
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.058572310169796e-05,
+      "loss": 0.3072,
+      "step": 16147
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0584728111776244e-05,
+      "loss": 0.2343,
+      "step": 16148
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0583733116045791e-05,
+      "loss": 0.2604,
+      "step": 16149
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0582738114516491e-05,
+      "loss": 0.2842,
+      "step": 16150
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.058174310719823e-05,
+      "loss": 0.2801,
+      "step": 16151
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0580748094100893e-05,
+      "loss": 0.2062,
+      "step": 16152
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0579753075234355e-05,
+      "loss": 0.2472,
+      "step": 16153
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0578758050608512e-05,
+      "loss": 0.2019,
+      "step": 16154
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057776302023325e-05,
+      "loss": 0.2499,
+      "step": 16155
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057676798411844e-05,
+      "loss": 0.2624,
+      "step": 16156
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0575772942273979e-05,
+      "loss": 0.2047,
+      "step": 16157
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0574777894709746e-05,
+      "loss": 0.2658,
+      "step": 16158
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057378284143563e-05,
+      "loss": 0.2266,
+      "step": 16159
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057278778246151e-05,
+      "loss": 0.2477,
+      "step": 16160
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0571792717797277e-05,
+      "loss": 0.2739,
+      "step": 16161
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0570797647452815e-05,
+      "loss": 0.2221,
+      "step": 16162
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0569802571438005e-05,
+      "loss": 0.3196,
+      "step": 16163
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0568807489762738e-05,
+      "loss": 0.2391,
+      "step": 16164
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0567812402436895e-05,
+      "loss": 0.2553,
+      "step": 16165
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056681730947036e-05,
+      "loss": 0.2644,
+      "step": 16166
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0565822210873025e-05,
+      "loss": 0.2429,
+      "step": 16167
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0564827106654767e-05,
+      "loss": 0.2645,
+      "step": 16168
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0563831996825478e-05,
+      "loss": 0.2308,
+      "step": 16169
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056283688139504e-05,
+      "loss": 0.2005,
+      "step": 16170
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0561841760373338e-05,
+      "loss": 0.2426,
+      "step": 16171
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0560846633770263e-05,
+      "loss": 0.2038,
+      "step": 16172
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0559851501595694e-05,
+      "loss": 0.2099,
+      "step": 16173
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0558856363859519e-05,
+      "loss": 0.2321,
+      "step": 16174
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0557861220571626e-05,
+      "loss": 0.1853,
+      "step": 16175
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0556866071741901e-05,
+      "loss": 0.2372,
+      "step": 16176
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0555870917380226e-05,
+      "loss": 0.2888,
+      "step": 16177
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0554875757496487e-05,
+      "loss": 0.2196,
+      "step": 16178
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0553880592100576e-05,
+      "loss": 0.2939,
+      "step": 16179
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0552885421202372e-05,
+      "loss": 0.2484,
+      "step": 16180
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0551890244811764e-05,
+      "loss": 0.235,
+      "step": 16181
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.055089506293864e-05,
+      "loss": 0.1995,
+      "step": 16182
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0549899875592883e-05,
+      "loss": 0.2465,
+      "step": 16183
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0548904682784383e-05,
+      "loss": 0.2168,
+      "step": 16184
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0547909484523023e-05,
+      "loss": 0.2211,
+      "step": 16185
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054691428081869e-05,
+      "loss": 0.2502,
+      "step": 16186
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054591907168127e-05,
+      "loss": 0.257,
+      "step": 16187
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054492385712065e-05,
+      "loss": 0.254,
+      "step": 16188
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054392863714672e-05,
+      "loss": 0.2664,
+      "step": 16189
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0542933411769359e-05,
+      "loss": 0.2355,
+      "step": 16190
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0541938180998463e-05,
+      "loss": 0.2499,
+      "step": 16191
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054094294484391e-05,
+      "loss": 0.2819,
+      "step": 16192
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0539947703315589e-05,
+      "loss": 0.2992,
+      "step": 16193
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0538952456423394e-05,
+      "loss": 0.2566,
+      "step": 16194
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0537957204177204e-05,
+      "loss": 0.1761,
+      "step": 16195
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0536961946586907e-05,
+      "loss": 0.236,
+      "step": 16196
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0535966683662392e-05,
+      "loss": 0.216,
+      "step": 16197
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0534971415413547e-05,
+      "loss": 0.2201,
+      "step": 16198
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0533976141850253e-05,
+      "loss": 0.2827,
+      "step": 16199
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0532980862982402e-05,
+      "loss": 0.2574,
+      "step": 16200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0531985578819884e-05,
+      "loss": 0.2006,
+      "step": 16201
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0530990289372578e-05,
+      "loss": 0.2279,
+      "step": 16202
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.052999499465038e-05,
+      "loss": 0.2285,
+      "step": 16203
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0528999694663171e-05,
+      "loss": 0.2242,
+      "step": 16204
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.052800438942084e-05,
+      "loss": 0.2146,
+      "step": 16205
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0527009078933276e-05,
+      "loss": 0.2165,
+      "step": 16206
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0526013763210365e-05,
+      "loss": 0.2212,
+      "step": 16207
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0525018442261996e-05,
+      "loss": 0.2141,
+      "step": 16208
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0524023116098057e-05,
+      "loss": 0.2176,
+      "step": 16209
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0523027784728434e-05,
+      "loss": 0.2694,
+      "step": 16210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.052203244816301e-05,
+      "loss": 0.2498,
+      "step": 16211
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0521037106411683e-05,
+      "loss": 0.2259,
+      "step": 16212
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0520041759484336e-05,
+      "loss": 0.1985,
+      "step": 16213
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0519046407390856e-05,
+      "loss": 0.2616,
+      "step": 16214
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.051805105014113e-05,
+      "loss": 0.2446,
+      "step": 16215
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0517055687745045e-05,
+      "loss": 0.1993,
+      "step": 16216
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0516060320212496e-05,
+      "loss": 0.234,
+      "step": 16217
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0515064947553361e-05,
+      "loss": 0.2919,
+      "step": 16218
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0514069569777539e-05,
+      "loss": 0.2388,
+      "step": 16219
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0513074186894911e-05,
+      "loss": 0.2962,
+      "step": 16220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0512078798915364e-05,
+      "loss": 0.242,
+      "step": 16221
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0511083405848791e-05,
+      "loss": 0.1843,
+      "step": 16222
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.051008800770508e-05,
+      "loss": 0.1999,
+      "step": 16223
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0509092604494113e-05,
+      "loss": 0.2412,
+      "step": 16224
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0508097196225787e-05,
+      "loss": 0.2117,
+      "step": 16225
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0507101782909987e-05,
+      "loss": 0.219,
+      "step": 16226
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0506106364556598e-05,
+      "loss": 0.1898,
+      "step": 16227
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0505110941175514e-05,
+      "loss": 0.24,
+      "step": 16228
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.050411551277662e-05,
+      "loss": 0.2515,
+      "step": 16229
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0503120079369809e-05,
+      "loss": 0.2736,
+      "step": 16230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0502124640964964e-05,
+      "loss": 0.2988,
+      "step": 16231
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0501129197571978e-05,
+      "loss": 0.1953,
+      "step": 16232
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0500133749200733e-05,
+      "loss": 0.2502,
+      "step": 16233
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0499138295861129e-05,
+      "loss": 0.221,
+      "step": 16234
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0498142837563048e-05,
+      "loss": 0.1999,
+      "step": 16235
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0497147374316377e-05,
+      "loss": 0.1923,
+      "step": 16236
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049615190613101e-05,
+      "loss": 0.2524,
+      "step": 16237
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0495156433016836e-05,
+      "loss": 0.176,
+      "step": 16238
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0494160954983738e-05,
+      "loss": 0.2198,
+      "step": 16239
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0493165472041611e-05,
+      "loss": 0.2443,
+      "step": 16240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0492169984200342e-05,
+      "loss": 0.2988,
+      "step": 16241
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049117449146982e-05,
+      "loss": 0.2918,
+      "step": 16242
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0490178993859937e-05,
+      "loss": 0.2766,
+      "step": 16243
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0489183491380582e-05,
+      "loss": 0.2724,
+      "step": 16244
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0488187984041636e-05,
+      "loss": 0.2309,
+      "step": 16245
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0487192471853e-05,
+      "loss": 0.2785,
+      "step": 16246
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048619695482456e-05,
+      "loss": 0.2019,
+      "step": 16247
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.04852014329662e-05,
+      "loss": 0.2557,
+      "step": 16248
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0484205906287817e-05,
+      "loss": 0.2889,
+      "step": 16249
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0483210374799295e-05,
+      "loss": 0.2219,
+      "step": 16250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0482214838510528e-05,
+      "loss": 0.2751,
+      "step": 16251
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0481219297431402e-05,
+      "loss": 0.2174,
+      "step": 16252
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048022375157181e-05,
+      "loss": 0.3077,
+      "step": 16253
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.047922820094164e-05,
+      "loss": 0.2406,
+      "step": 16254
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0478232645550784e-05,
+      "loss": 0.2262,
+      "step": 16255
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0477237085409129e-05,
+      "loss": 0.217,
+      "step": 16256
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0476241520526566e-05,
+      "loss": 0.2219,
+      "step": 16257
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0475245950912982e-05,
+      "loss": 0.2409,
+      "step": 16258
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0474250376578274e-05,
+      "loss": 0.1854,
+      "step": 16259
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.047325479753233e-05,
+      "loss": 0.2393,
+      "step": 16260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0472259213785035e-05,
+      "loss": 0.233,
+      "step": 16261
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0471263625346281e-05,
+      "loss": 0.2431,
+      "step": 16262
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0470268032225966e-05,
+      "loss": 0.263,
+      "step": 16263
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0469272434433968e-05,
+      "loss": 0.191,
+      "step": 16264
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0468276831980186e-05,
+      "loss": 0.2705,
+      "step": 16265
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0467281224874512e-05,
+      "loss": 0.2285,
+      "step": 16266
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0466285613126827e-05,
+      "loss": 0.2497,
+      "step": 16267
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.046528999674703e-05,
+      "loss": 0.2452,
+      "step": 16268
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0464294375745007e-05,
+      "loss": 0.2569,
+      "step": 16269
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0463298750130648e-05,
+      "loss": 0.1849,
+      "step": 16270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0462303119913851e-05,
+      "loss": 0.2469,
+      "step": 16271
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0461307485104497e-05,
+      "loss": 0.2153,
+      "step": 16272
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0460311845712481e-05,
+      "loss": 0.2628,
+      "step": 16273
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0459316201747695e-05,
+      "loss": 0.273,
+      "step": 16274
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.045832055322003e-05,
+      "loss": 0.2896,
+      "step": 16275
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0457324900139372e-05,
+      "loss": 0.2969,
+      "step": 16276
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0456329242515617e-05,
+      "loss": 0.2357,
+      "step": 16277
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0455333580358656e-05,
+      "loss": 0.2165,
+      "step": 16278
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0454337913678374e-05,
+      "loss": 0.2283,
+      "step": 16279
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.045334224248467e-05,
+      "loss": 0.2729,
+      "step": 16280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.045234656678743e-05,
+      "loss": 0.2267,
+      "step": 16281
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0451350886596543e-05,
+      "loss": 0.2399,
+      "step": 16282
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0450355201921908e-05,
+      "loss": 0.2121,
+      "step": 16283
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0449359512773412e-05,
+      "loss": 0.2031,
+      "step": 16284
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0448363819160946e-05,
+      "loss": 0.2881,
+      "step": 16285
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0447368121094397e-05,
+      "loss": 0.2628,
+      "step": 16286
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0446372418583664e-05,
+      "loss": 0.2923,
+      "step": 16287
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0445376711638633e-05,
+      "loss": 0.1792,
+      "step": 16288
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.04443810002692e-05,
+      "loss": 0.2649,
+      "step": 16289
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0443385284485254e-05,
+      "loss": 0.2202,
+      "step": 16290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0442389564296682e-05,
+      "loss": 0.1687,
+      "step": 16291
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0441393839713385e-05,
+      "loss": 0.2549,
+      "step": 16292
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0440398110745249e-05,
+      "loss": 0.307,
+      "step": 16293
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0439402377402163e-05,
+      "loss": 0.2389,
+      "step": 16294
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0438406639694025e-05,
+      "loss": 0.2567,
+      "step": 16295
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0437410897630721e-05,
+      "loss": 0.2201,
+      "step": 16296
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0436415151222147e-05,
+      "loss": 0.2703,
+      "step": 16297
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0435419400478193e-05,
+      "loss": 0.2225,
+      "step": 16298
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043442364540875e-05,
+      "loss": 0.219,
+      "step": 16299
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0433427886023717e-05,
+      "loss": 0.2741,
+      "step": 16300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0432432122332972e-05,
+      "loss": 0.2403,
+      "step": 16301
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043143635434642e-05,
+      "loss": 0.2173,
+      "step": 16302
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0430440582073947e-05,
+      "loss": 0.2498,
+      "step": 16303
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0429444805525442e-05,
+      "loss": 0.2577,
+      "step": 16304
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0428449024710805e-05,
+      "loss": 0.2244,
+      "step": 16305
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0427453239639925e-05,
+      "loss": 0.2502,
+      "step": 16306
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0426457450322691e-05,
+      "loss": 0.215,
+      "step": 16307
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0425461656768999e-05,
+      "loss": 0.2028,
+      "step": 16308
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0424465858988742e-05,
+      "loss": 0.2465,
+      "step": 16309
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0423470056991804e-05,
+      "loss": 0.1887,
+      "step": 16310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0422474250788088e-05,
+      "loss": 0.2275,
+      "step": 16311
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0421478440387484e-05,
+      "loss": 0.2631,
+      "step": 16312
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.042048262579988e-05,
+      "loss": 0.2363,
+      "step": 16313
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0419486807035173e-05,
+      "loss": 0.268,
+      "step": 16314
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041849098410325e-05,
+      "loss": 0.2415,
+      "step": 16315
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0417495157014009e-05,
+      "loss": 0.1798,
+      "step": 16316
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0416499325777343e-05,
+      "loss": 0.2798,
+      "step": 16317
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041550349040314e-05,
+      "loss": 0.1651,
+      "step": 16318
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0414507650901295e-05,
+      "loss": 0.2514,
+      "step": 16319
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0413511807281702e-05,
+      "loss": 0.2243,
+      "step": 16320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0412515959554254e-05,
+      "loss": 0.2551,
+      "step": 16321
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041152010772884e-05,
+      "loss": 0.2697,
+      "step": 16322
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0410524251815357e-05,
+      "loss": 0.2182,
+      "step": 16323
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0409528391823698e-05,
+      "loss": 0.2139,
+      "step": 16324
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0408532527763749e-05,
+      "loss": 0.3145,
+      "step": 16325
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0407536659645414e-05,
+      "loss": 0.2355,
+      "step": 16326
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0406540787478576e-05,
+      "loss": 0.3247,
+      "step": 16327
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0405544911273133e-05,
+      "loss": 0.2292,
+      "step": 16328
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.040454903103898e-05,
+      "loss": 0.2775,
+      "step": 16329
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0403553146786005e-05,
+      "loss": 0.2019,
+      "step": 16330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0402557258524105e-05,
+      "loss": 0.2344,
+      "step": 16331
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0401561366263171e-05,
+      "loss": 0.2136,
+      "step": 16332
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0400565470013098e-05,
+      "loss": 0.2502,
+      "step": 16333
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0399569569783778e-05,
+      "loss": 0.2059,
+      "step": 16334
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0398573665585105e-05,
+      "loss": 0.2186,
+      "step": 16335
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0397577757426976e-05,
+      "loss": 0.2361,
+      "step": 16336
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0396581845319275e-05,
+      "loss": 0.2559,
+      "step": 16337
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0395585929271904e-05,
+      "loss": 0.3321,
+      "step": 16338
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0394590009294753e-05,
+      "loss": 0.231,
+      "step": 16339
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0393594085397716e-05,
+      "loss": 0.2339,
+      "step": 16340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0392598157590687e-05,
+      "loss": 0.2317,
+      "step": 16341
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.039160222588356e-05,
+      "loss": 0.2384,
+      "step": 16342
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0390606290286228e-05,
+      "loss": 0.2178,
+      "step": 16343
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0389610350808584e-05,
+      "loss": 0.2281,
+      "step": 16344
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0388614407460523e-05,
+      "loss": 0.2064,
+      "step": 16345
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.038761846025194e-05,
+      "loss": 0.2505,
+      "step": 16346
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0386622509192723e-05,
+      "loss": 0.1823,
+      "step": 16347
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0385626554292774e-05,
+      "loss": 0.2264,
+      "step": 16348
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0384630595561982e-05,
+      "loss": 0.2393,
+      "step": 16349
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.038363463301024e-05,
+      "loss": 0.2725,
+      "step": 16350
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0382638666647444e-05,
+      "loss": 0.2367,
+      "step": 16351
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0381642696483489e-05,
+      "loss": 0.249,
+      "step": 16352
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0380646722528267e-05,
+      "loss": 0.209,
+      "step": 16353
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.037965074479167e-05,
+      "loss": 0.184,
+      "step": 16354
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0378654763283602e-05,
+      "loss": 0.2318,
+      "step": 16355
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0377658778013944e-05,
+      "loss": 0.278,
+      "step": 16356
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0376662788992597e-05,
+      "loss": 0.1754,
+      "step": 16357
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.037566679622946e-05,
+      "loss": 0.2286,
+      "step": 16358
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0374670799734414e-05,
+      "loss": 0.1966,
+      "step": 16359
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0373674799517368e-05,
+      "loss": 0.2119,
+      "step": 16360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0372678795588204e-05,
+      "loss": 0.2154,
+      "step": 16361
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0371682787956822e-05,
+      "loss": 0.2557,
+      "step": 16362
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.037068677663312e-05,
+      "loss": 0.2798,
+      "step": 16363
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0369690761626987e-05,
+      "loss": 0.1749,
+      "step": 16364
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.036869474294832e-05,
+      "loss": 0.2629,
+      "step": 16365
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0367698720607012e-05,
+      "loss": 0.2597,
+      "step": 16366
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.036670269461296e-05,
+      "loss": 0.2579,
+      "step": 16367
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0365706664976052e-05,
+      "loss": 0.2063,
+      "step": 16368
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.036471063170619e-05,
+      "loss": 0.302,
+      "step": 16369
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0363714594813268e-05,
+      "loss": 0.2428,
+      "step": 16370
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0362718554307174e-05,
+      "loss": 0.2292,
+      "step": 16371
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0361722510197813e-05,
+      "loss": 0.2456,
+      "step": 16372
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0360726462495072e-05,
+      "loss": 0.2558,
+      "step": 16373
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0359730411208848e-05,
+      "loss": 0.245,
+      "step": 16374
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0358734356349035e-05,
+      "loss": 0.2368,
+      "step": 16375
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0357738297925531e-05,
+      "loss": 0.2345,
+      "step": 16376
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0356742235948227e-05,
+      "loss": 0.253,
+      "step": 16377
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0355746170427021e-05,
+      "loss": 0.2613,
+      "step": 16378
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0354750101371808e-05,
+      "loss": 0.3161,
+      "step": 16379
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.035375402879248e-05,
+      "loss": 0.2407,
+      "step": 16380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0352757952698933e-05,
+      "loss": 0.193,
+      "step": 16381
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0351761873101065e-05,
+      "loss": 0.2488,
+      "step": 16382
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0350765790008768e-05,
+      "loss": 0.2529,
+      "step": 16383
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0349769703431937e-05,
+      "loss": 0.2596,
+      "step": 16384
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.034877361338047e-05,
+      "loss": 0.3084,
+      "step": 16385
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0347777519864259e-05,
+      "loss": 0.2272,
+      "step": 16386
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0346781422893203e-05,
+      "loss": 0.2697,
+      "step": 16387
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0345785322477194e-05,
+      "loss": 0.2665,
+      "step": 16388
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.034478921862613e-05,
+      "loss": 0.2361,
+      "step": 16389
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0343793111349903e-05,
+      "loss": 0.2199,
+      "step": 16390
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0342797000658411e-05,
+      "loss": 0.23,
+      "step": 16391
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0341800886561549e-05,
+      "loss": 0.2438,
+      "step": 16392
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0340804769069211e-05,
+      "loss": 0.2991,
+      "step": 16393
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0339808648191294e-05,
+      "loss": 0.2352,
+      "step": 16394
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0338812523937694e-05,
+      "loss": 0.2217,
+      "step": 16395
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0337816396318305e-05,
+      "loss": 0.192,
+      "step": 16396
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0336820265343024e-05,
+      "loss": 0.2216,
+      "step": 16397
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0335824131021746e-05,
+      "loss": 0.2565,
+      "step": 16398
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0334827993364367e-05,
+      "loss": 0.2722,
+      "step": 16399
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0333831852380782e-05,
+      "loss": 0.2943,
+      "step": 16400
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0332835708080888e-05,
+      "loss": 0.2516,
+      "step": 16401
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.033183956047458e-05,
+      "loss": 0.2361,
+      "step": 16402
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0330843409571752e-05,
+      "loss": 0.2494,
+      "step": 16403
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0329847255382303e-05,
+      "loss": 0.1882,
+      "step": 16404
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0328851097916124e-05,
+      "loss": 0.1738,
+      "step": 16405
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.032785493718312e-05,
+      "loss": 0.2744,
+      "step": 16406
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0326858773193178e-05,
+      "loss": 0.2259,
+      "step": 16407
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0325862605956198e-05,
+      "loss": 0.2415,
+      "step": 16408
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0324866435482075e-05,
+      "loss": 0.3108,
+      "step": 16409
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0323870261780703e-05,
+      "loss": 0.2184,
+      "step": 16410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0322874084861984e-05,
+      "loss": 0.2571,
+      "step": 16411
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0321877904735809e-05,
+      "loss": 0.2489,
+      "step": 16412
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0320881721412077e-05,
+      "loss": 0.2112,
+      "step": 16413
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.031988553490068e-05,
+      "loss": 0.2293,
+      "step": 16414
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0318889345211519e-05,
+      "loss": 0.2338,
+      "step": 16415
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0317893152354487e-05,
+      "loss": 0.2413,
+      "step": 16416
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0316896956339481e-05,
+      "loss": 0.2054,
+      "step": 16417
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.03159007571764e-05,
+      "loss": 0.2405,
+      "step": 16418
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0314904554875135e-05,
+      "loss": 0.2555,
+      "step": 16419
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0313908349445587e-05,
+      "loss": 0.2661,
+      "step": 16420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0312912140897651e-05,
+      "loss": 0.2447,
+      "step": 16421
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0311915929241221e-05,
+      "loss": 0.246,
+      "step": 16422
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0310919714486196e-05,
+      "loss": 0.2269,
+      "step": 16423
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0309923496642473e-05,
+      "loss": 0.2752,
+      "step": 16424
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0308927275719949e-05,
+      "loss": 0.2803,
+      "step": 16425
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0307931051728517e-05,
+      "loss": 0.2593,
+      "step": 16426
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0306934824678075e-05,
+      "loss": 0.2052,
+      "step": 16427
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0305938594578523e-05,
+      "loss": 0.2346,
+      "step": 16428
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0304942361439752e-05,
+      "loss": 0.1718,
+      "step": 16429
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0303946125271662e-05,
+      "loss": 0.2501,
+      "step": 16430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0302949886084151e-05,
+      "loss": 0.3037,
+      "step": 16431
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030195364388711e-05,
+      "loss": 0.2171,
+      "step": 16432
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0300957398690444e-05,
+      "loss": 0.1835,
+      "step": 16433
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0299961150504043e-05,
+      "loss": 0.2623,
+      "step": 16434
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0298964899337807e-05,
+      "loss": 0.28,
+      "step": 16435
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0297968645201631e-05,
+      "loss": 0.223,
+      "step": 16436
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0296972388105416e-05,
+      "loss": 0.2057,
+      "step": 16437
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0295976128059051e-05,
+      "loss": 0.2016,
+      "step": 16438
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0294979865072436e-05,
+      "loss": 0.2215,
+      "step": 16439
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0293983599155477e-05,
+      "loss": 0.2296,
+      "step": 16440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.029298733031806e-05,
+      "loss": 0.2401,
+      "step": 16441
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0291991058570085e-05,
+      "loss": 0.229,
+      "step": 16442
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0290994783921451e-05,
+      "loss": 0.2555,
+      "step": 16443
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0289998506382053e-05,
+      "loss": 0.2692,
+      "step": 16444
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0289002225961788e-05,
+      "loss": 0.2121,
+      "step": 16445
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0288005942670554e-05,
+      "loss": 0.2328,
+      "step": 16446
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028700965651825e-05,
+      "loss": 0.1902,
+      "step": 16447
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0286013367514768e-05,
+      "loss": 0.2342,
+      "step": 16448
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0285017075670014e-05,
+      "loss": 0.1828,
+      "step": 16449
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0284020780993876e-05,
+      "loss": 0.24,
+      "step": 16450
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028302448349625e-05,
+      "loss": 0.2195,
+      "step": 16451
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0282028183187047e-05,
+      "loss": 0.2215,
+      "step": 16452
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0281031880076152e-05,
+      "loss": 0.3037,
+      "step": 16453
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0280035574173466e-05,
+      "loss": 0.2419,
+      "step": 16454
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0279039265488885e-05,
+      "loss": 0.2798,
+      "step": 16455
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.027804295403231e-05,
+      "loss": 0.2456,
+      "step": 16456
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0277046639813636e-05,
+      "loss": 0.1845,
+      "step": 16457
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0276050322842757e-05,
+      "loss": 0.2569,
+      "step": 16458
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0275054003129578e-05,
+      "loss": 0.2747,
+      "step": 16459
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0274057680683989e-05,
+      "loss": 0.2972,
+      "step": 16460
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0273061355515896e-05,
+      "loss": 0.2324,
+      "step": 16461
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0272065027635189e-05,
+      "loss": 0.1671,
+      "step": 16462
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0271068697051767e-05,
+      "loss": 0.2251,
+      "step": 16463
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.027007236377553e-05,
+      "loss": 0.2459,
+      "step": 16464
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0269076027816376e-05,
+      "loss": 0.213,
+      "step": 16465
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0268079689184199e-05,
+      "loss": 0.211,
+      "step": 16466
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.02670833478889e-05,
+      "loss": 0.2242,
+      "step": 16467
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0266087003940375e-05,
+      "loss": 0.2138,
+      "step": 16468
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0265090657348525e-05,
+      "loss": 0.2216,
+      "step": 16469
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.026409430812324e-05,
+      "loss": 0.2415,
+      "step": 16470
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0263097956274427e-05,
+      "loss": 0.204,
+      "step": 16471
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.026210160181198e-05,
+      "loss": 0.2403,
+      "step": 16472
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0261105244745792e-05,
+      "loss": 0.2363,
+      "step": 16473
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0260108885085774e-05,
+      "loss": 0.2431,
+      "step": 16474
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0259112522841808e-05,
+      "loss": 0.2283,
+      "step": 16475
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0258116158023803e-05,
+      "loss": 0.197,
+      "step": 16476
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0257119790641652e-05,
+      "loss": 0.2301,
+      "step": 16477
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0256123420705255e-05,
+      "loss": 0.1919,
+      "step": 16478
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0255127048224506e-05,
+      "loss": 0.2633,
+      "step": 16479
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.025413067320931e-05,
+      "loss": 0.2203,
+      "step": 16480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0253134295669564e-05,
+      "loss": 0.2794,
+      "step": 16481
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0252137915615156e-05,
+      "loss": 0.2505,
+      "step": 16482
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0251141533055999e-05,
+      "loss": 0.2355,
+      "step": 16483
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0250145148001982e-05,
+      "loss": 0.2523,
+      "step": 16484
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0249148760463001e-05,
+      "loss": 0.2697,
+      "step": 16485
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0248152370448963e-05,
+      "loss": 0.2385,
+      "step": 16486
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0247155977969758e-05,
+      "loss": 0.1821,
+      "step": 16487
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0246159583035288e-05,
+      "loss": 0.2344,
+      "step": 16488
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0245163185655453e-05,
+      "loss": 0.1963,
+      "step": 16489
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0244166785840151e-05,
+      "loss": 0.2101,
+      "step": 16490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0243170383599272e-05,
+      "loss": 0.1742,
+      "step": 16491
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0242173978942724e-05,
+      "loss": 0.2275,
+      "step": 16492
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0241177571880403e-05,
+      "loss": 0.207,
+      "step": 16493
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0240181162422205e-05,
+      "loss": 0.2277,
+      "step": 16494
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.023918475057803e-05,
+      "loss": 0.208,
+      "step": 16495
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.023818833635778e-05,
+      "loss": 0.2157,
+      "step": 16496
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0237191919771345e-05,
+      "loss": 0.2149,
+      "step": 16497
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.023619550082863e-05,
+      "loss": 0.2293,
+      "step": 16498
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.023519907953953e-05,
+      "loss": 0.1838,
+      "step": 16499
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0234202655913947e-05,
+      "loss": 0.2226,
+      "step": 16500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0233206229961778e-05,
+      "loss": 0.2528,
+      "step": 16501
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.023220980169292e-05,
+      "loss": 0.2084,
+      "step": 16502
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0231213371117274e-05,
+      "loss": 0.2369,
+      "step": 16503
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.023021693824474e-05,
+      "loss": 0.1952,
+      "step": 16504
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0229220503085214e-05,
+      "loss": 0.2671,
+      "step": 16505
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0228224065648591e-05,
+      "loss": 0.2149,
+      "step": 16506
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0227227625944776e-05,
+      "loss": 0.2618,
+      "step": 16507
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0226231183983664e-05,
+      "loss": 0.3128,
+      "step": 16508
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0225234739775155e-05,
+      "loss": 0.2966,
+      "step": 16509
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022423829332915e-05,
+      "loss": 0.3213,
+      "step": 16510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0223241844655542e-05,
+      "loss": 0.3113,
+      "step": 16511
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0222245393764236e-05,
+      "loss": 0.1869,
+      "step": 16512
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0221248940665127e-05,
+      "loss": 0.2589,
+      "step": 16513
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0220252485368117e-05,
+      "loss": 0.2201,
+      "step": 16514
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0219256027883102e-05,
+      "loss": 0.2144,
+      "step": 16515
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0218259568219979e-05,
+      "loss": 0.2692,
+      "step": 16516
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0217263106388653e-05,
+      "loss": 0.2149,
+      "step": 16517
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.021626664239902e-05,
+      "loss": 0.2559,
+      "step": 16518
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0215270176260976e-05,
+      "loss": 0.2543,
+      "step": 16519
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0214273707984421e-05,
+      "loss": 0.2355,
+      "step": 16520
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0213277237579258e-05,
+      "loss": 0.2786,
+      "step": 16521
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0212280765055382e-05,
+      "loss": 0.2581,
+      "step": 16522
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0211284290422696e-05,
+      "loss": 0.2656,
+      "step": 16523
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0210287813691098e-05,
+      "loss": 0.2191,
+      "step": 16524
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0209291334870479e-05,
+      "loss": 0.2106,
+      "step": 16525
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0208294853970749e-05,
+      "loss": 0.2387,
+      "step": 16526
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0207298371001804e-05,
+      "loss": 0.23,
+      "step": 16527
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0206301885973537e-05,
+      "loss": 0.3,
+      "step": 16528
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0205305398895857e-05,
+      "loss": 0.2758,
+      "step": 16529
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0204308909778654e-05,
+      "loss": 0.2968,
+      "step": 16530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0203312418631837e-05,
+      "loss": 0.2105,
+      "step": 16531
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0202315925465296e-05,
+      "loss": 0.3438,
+      "step": 16532
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0201319430288934e-05,
+      "loss": 0.2446,
+      "step": 16533
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.020032293311265e-05,
+      "loss": 0.2251,
+      "step": 16534
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0199326433946345e-05,
+      "loss": 0.2259,
+      "step": 16535
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0198329932799917e-05,
+      "loss": 0.1804,
+      "step": 16536
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.019733342968326e-05,
+      "loss": 0.2678,
+      "step": 16537
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0196336924606282e-05,
+      "loss": 0.2482,
+      "step": 16538
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0195340417578883e-05,
+      "loss": 0.2215,
+      "step": 16539
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0194343908610951e-05,
+      "loss": 0.26,
+      "step": 16540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0193347397712399e-05,
+      "loss": 0.2636,
+      "step": 16541
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0192350884893117e-05,
+      "loss": 0.2362,
+      "step": 16542
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0191354370163005e-05,
+      "loss": 0.2086,
+      "step": 16543
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0190357853531967e-05,
+      "loss": 0.2431,
+      "step": 16544
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.01893613350099e-05,
+      "loss": 0.2433,
+      "step": 16545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0188364814606704e-05,
+      "loss": 0.2266,
+      "step": 16546
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0187368292332277e-05,
+      "loss": 0.2488,
+      "step": 16547
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0186371768196523e-05,
+      "loss": 0.266,
+      "step": 16548
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0185375242209336e-05,
+      "loss": 0.2422,
+      "step": 16549
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0184378714380617e-05,
+      "loss": 0.2295,
+      "step": 16550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0183382184720268e-05,
+      "loss": 0.2505,
+      "step": 16551
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0182385653238187e-05,
+      "loss": 0.2989,
+      "step": 16552
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0181389119944272e-05,
+      "loss": 0.2797,
+      "step": 16553
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0180392584848426e-05,
+      "loss": 0.1989,
+      "step": 16554
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0179396047960546e-05,
+      "loss": 0.2247,
+      "step": 16555
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0178399509290533e-05,
+      "loss": 0.2251,
+      "step": 16556
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0177402968848286e-05,
+      "loss": 0.205,
+      "step": 16557
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0176406426643709e-05,
+      "loss": 0.2894,
+      "step": 16558
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.017540988268669e-05,
+      "loss": 0.2386,
+      "step": 16559
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0174413336987146e-05,
+      "loss": 0.3303,
+      "step": 16560
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.017341678955496e-05,
+      "loss": 0.2252,
+      "step": 16561
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0172420240400038e-05,
+      "loss": 0.202,
+      "step": 16562
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0171423689532287e-05,
+      "loss": 0.185,
+      "step": 16563
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0170427136961598e-05,
+      "loss": 0.2386,
+      "step": 16564
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0169430582697873e-05,
+      "loss": 0.203,
+      "step": 16565
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0168434026751014e-05,
+      "loss": 0.1971,
+      "step": 16566
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0167437469130914e-05,
+      "loss": 0.2582,
+      "step": 16567
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0166440909847483e-05,
+      "loss": 0.2421,
+      "step": 16568
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0165444348910614e-05,
+      "loss": 0.2195,
+      "step": 16569
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0164447786330212e-05,
+      "loss": 0.2264,
+      "step": 16570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0163451222116167e-05,
+      "loss": 0.2381,
+      "step": 16571
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.016245465627839e-05,
+      "loss": 0.1825,
+      "step": 16572
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.016145808882678e-05,
+      "loss": 0.2608,
+      "step": 16573
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0160461519771228e-05,
+      "loss": 0.1675,
+      "step": 16574
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0159464949121642e-05,
+      "loss": 0.2228,
+      "step": 16575
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.015846837688792e-05,
+      "loss": 0.1917,
+      "step": 16576
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0157471803079962e-05,
+      "loss": 0.1781,
+      "step": 16577
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0156475227707666e-05,
+      "loss": 0.256,
+      "step": 16578
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0155478650780932e-05,
+      "loss": 0.2526,
+      "step": 16579
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0154482072309666e-05,
+      "loss": 0.2428,
+      "step": 16580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0153485492303761e-05,
+      "loss": 0.2529,
+      "step": 16581
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0152488910773124e-05,
+      "loss": 0.2066,
+      "step": 16582
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0151492327727646e-05,
+      "loss": 0.2438,
+      "step": 16583
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0150495743177233e-05,
+      "loss": 0.2408,
+      "step": 16584
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0149499157131788e-05,
+      "loss": 0.2593,
+      "step": 16585
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0148502569601203e-05,
+      "loss": 0.18,
+      "step": 16586
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0147505980595387e-05,
+      "loss": 0.2187,
+      "step": 16587
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0146509390124232e-05,
+      "loss": 0.2587,
+      "step": 16588
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0145512798197645e-05,
+      "loss": 0.2321,
+      "step": 16589
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0144516204825522e-05,
+      "loss": 0.243,
+      "step": 16590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0143519610017764e-05,
+      "loss": 0.2457,
+      "step": 16591
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0142523013784273e-05,
+      "loss": 0.2298,
+      "step": 16592
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0141526416134948e-05,
+      "loss": 0.1922,
+      "step": 16593
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0140529817079692e-05,
+      "loss": 0.2855,
+      "step": 16594
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0139533216628399e-05,
+      "loss": 0.224,
+      "step": 16595
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013853661479097e-05,
+      "loss": 0.2479,
+      "step": 16596
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0137540011577314e-05,
+      "loss": 0.2222,
+      "step": 16597
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0136543406997323e-05,
+      "loss": 0.1983,
+      "step": 16598
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.01355468010609e-05,
+      "loss": 0.2187,
+      "step": 16599
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0134550193777946e-05,
+      "loss": 0.2405,
+      "step": 16600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013355358515836e-05,
+      "loss": 0.3064,
+      "step": 16601
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0132556975212042e-05,
+      "loss": 0.2428,
+      "step": 16602
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0131560363948894e-05,
+      "loss": 0.2056,
+      "step": 16603
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0130563751378819e-05,
+      "loss": 0.1927,
+      "step": 16604
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.012956713751171e-05,
+      "loss": 0.28,
+      "step": 16605
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0128570522357473e-05,
+      "loss": 0.2609,
+      "step": 16606
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0127573905926007e-05,
+      "loss": 0.162,
+      "step": 16607
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.012657728822721e-05,
+      "loss": 0.2072,
+      "step": 16608
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.012558066927099e-05,
+      "loss": 0.2602,
+      "step": 16609
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.012458404906724e-05,
+      "loss": 0.2633,
+      "step": 16610
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0123587427625863e-05,
+      "loss": 0.2341,
+      "step": 16611
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.012259080495676e-05,
+      "loss": 0.2561,
+      "step": 16612
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0121594181069832e-05,
+      "loss": 0.2146,
+      "step": 16613
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0120597555974975e-05,
+      "loss": 0.2586,
+      "step": 16614
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0119600929682096e-05,
+      "loss": 0.2358,
+      "step": 16615
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0118604302201094e-05,
+      "loss": 0.2481,
+      "step": 16616
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0117607673541865e-05,
+      "loss": 0.2748,
+      "step": 16617
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0116611043714316e-05,
+      "loss": 0.2346,
+      "step": 16618
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0115614412728344e-05,
+      "loss": 0.2152,
+      "step": 16619
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0114617780593846e-05,
+      "loss": 0.2422,
+      "step": 16620
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.011362114732073e-05,
+      "loss": 0.2193,
+      "step": 16621
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0112624512918893e-05,
+      "loss": 0.2582,
+      "step": 16622
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0111627877398235e-05,
+      "loss": 0.2514,
+      "step": 16623
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0110631240768659e-05,
+      "loss": 0.2537,
+      "step": 16624
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0109634603040062e-05,
+      "loss": 0.2283,
+      "step": 16625
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.010863796422235e-05,
+      "loss": 0.2464,
+      "step": 16626
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0107641324325417e-05,
+      "loss": 0.1914,
+      "step": 16627
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0106644683359173e-05,
+      "loss": 0.2215,
+      "step": 16628
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0105648041333507e-05,
+      "loss": 0.235,
+      "step": 16629
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.010465139825833e-05,
+      "loss": 0.2312,
+      "step": 16630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0103654754143535e-05,
+      "loss": 0.1877,
+      "step": 16631
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0102658108999026e-05,
+      "loss": 0.2804,
+      "step": 16632
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0101661462834706e-05,
+      "loss": 0.2087,
+      "step": 16633
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0100664815660473e-05,
+      "loss": 0.2504,
+      "step": 16634
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0099668167486227e-05,
+      "loss": 0.209,
+      "step": 16635
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0098671518321872e-05,
+      "loss": 0.2204,
+      "step": 16636
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0097674868177308e-05,
+      "loss": 0.3561,
+      "step": 16637
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0096678217062434e-05,
+      "loss": 0.1865,
+      "step": 16638
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.009568156498715e-05,
+      "loss": 0.2326,
+      "step": 16639
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0094684911961361e-05,
+      "loss": 0.2121,
+      "step": 16640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0093688257994962e-05,
+      "loss": 0.2609,
+      "step": 16641
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0092691603097858e-05,
+      "loss": 0.2671,
+      "step": 16642
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0091694947279951e-05,
+      "loss": 0.2261,
+      "step": 16643
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0090698290551138e-05,
+      "loss": 0.2427,
+      "step": 16644
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0089701632921321e-05,
+      "loss": 0.1945,
+      "step": 16645
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0088704974400404e-05,
+      "loss": 0.2196,
+      "step": 16646
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0087708314998284e-05,
+      "loss": 0.2489,
+      "step": 16647
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0086711654724862e-05,
+      "loss": 0.3147,
+      "step": 16648
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008571499359004e-05,
+      "loss": 0.2482,
+      "step": 16649
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008471833160372e-05,
+      "loss": 0.2114,
+      "step": 16650
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.00837216687758e-05,
+      "loss": 0.2413,
+      "step": 16651
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0082725005116187e-05,
+      "loss": 0.2549,
+      "step": 16652
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0081728340634776e-05,
+      "loss": 0.2648,
+      "step": 16653
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0080731675341466e-05,
+      "loss": 0.1855,
+      "step": 16654
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0079735009246168e-05,
+      "loss": 0.2323,
+      "step": 16655
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0078738342358774e-05,
+      "loss": 0.2795,
+      "step": 16656
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0077741674689187e-05,
+      "loss": 0.2988,
+      "step": 16657
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0076745006247308e-05,
+      "loss": 0.3087,
+      "step": 16658
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0075748337043041e-05,
+      "loss": 0.2282,
+      "step": 16659
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0074751667086283e-05,
+      "loss": 0.2613,
+      "step": 16660
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0073754996386935e-05,
+      "loss": 0.2483,
+      "step": 16661
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0072758324954903e-05,
+      "loss": 0.2727,
+      "step": 16662
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.007176165280008e-05,
+      "loss": 0.2637,
+      "step": 16663
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0070764979932374e-05,
+      "loss": 0.2292,
+      "step": 16664
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0069768306361685e-05,
+      "loss": 0.2478,
+      "step": 16665
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0068771632097912e-05,
+      "loss": 0.232,
+      "step": 16666
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0067774957150955e-05,
+      "loss": 0.2572,
+      "step": 16667
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0066778281530717e-05,
+      "loss": 0.2401,
+      "step": 16668
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0065781605247098e-05,
+      "loss": 0.2221,
+      "step": 16669
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.006478492831e-05,
+      "loss": 0.2033,
+      "step": 16670
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0063788250729327e-05,
+      "loss": 0.2787,
+      "step": 16671
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0062791572514973e-05,
+      "loss": 0.1916,
+      "step": 16672
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0061794893676844e-05,
+      "loss": 0.2159,
+      "step": 16673
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0060798214224842e-05,
+      "loss": 0.2439,
+      "step": 16674
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0059801534168868e-05,
+      "loss": 0.2926,
+      "step": 16675
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0058804853518816e-05,
+      "loss": 0.2218,
+      "step": 16676
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0057808172284593e-05,
+      "loss": 0.2808,
+      "step": 16677
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0056811490476102e-05,
+      "loss": 0.2646,
+      "step": 16678
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.005581480810324e-05,
+      "loss": 0.1819,
+      "step": 16679
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.005481812517591e-05,
+      "loss": 0.2653,
+      "step": 16680
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0053821441704012e-05,
+      "loss": 0.2754,
+      "step": 16681
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0052824757697448e-05,
+      "loss": 0.1807,
+      "step": 16682
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0051828073166119e-05,
+      "loss": 0.26,
+      "step": 16683
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0050831388119927e-05,
+      "loss": 0.2627,
+      "step": 16684
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0049834702568772e-05,
+      "loss": 0.2334,
+      "step": 16685
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0048838016522557e-05,
+      "loss": 0.2555,
+      "step": 16686
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0047841329991181e-05,
+      "loss": 0.2167,
+      "step": 16687
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0046844642984544e-05,
+      "loss": 0.2152,
+      "step": 16688
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0045847955512548e-05,
+      "loss": 0.1964,
+      "step": 16689
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0044851267585098e-05,
+      "loss": 0.1847,
+      "step": 16690
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0043854579212091e-05,
+      "loss": 0.24,
+      "step": 16691
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.004285789040343e-05,
+      "loss": 0.2017,
+      "step": 16692
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0041861201169017e-05,
+      "loss": 0.2374,
+      "step": 16693
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0040864511518747e-05,
+      "loss": 0.2687,
+      "step": 16694
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0039867821462529e-05,
+      "loss": 0.2512,
+      "step": 16695
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0038871131010264e-05,
+      "loss": 0.2438,
+      "step": 16696
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0037874440171845e-05,
+      "loss": 0.2348,
+      "step": 16697
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0036877748957182e-05,
+      "loss": 0.2257,
+      "step": 16698
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0035881057376172e-05,
+      "loss": 0.2402,
+      "step": 16699
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0034884365438715e-05,
+      "loss": 0.2507,
+      "step": 16700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0033887673154717e-05,
+      "loss": 0.2431,
+      "step": 16701
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0032890980534075e-05,
+      "loss": 0.2111,
+      "step": 16702
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0031894287586691e-05,
+      "loss": 0.1801,
+      "step": 16703
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.003089759432247e-05,
+      "loss": 0.2646,
+      "step": 16704
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0029900900751308e-05,
+      "loss": 0.2039,
+      "step": 16705
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0028904206883106e-05,
+      "loss": 0.2222,
+      "step": 16706
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0027907512727771e-05,
+      "loss": 0.2309,
+      "step": 16707
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.00269108182952e-05,
+      "loss": 0.2182,
+      "step": 16708
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0025914123595294e-05,
+      "loss": 0.1892,
+      "step": 16709
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0024917428637954e-05,
+      "loss": 0.2487,
+      "step": 16710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0023920733433085e-05,
+      "loss": 0.2836,
+      "step": 16711
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0022924037990584e-05,
+      "loss": 0.2925,
+      "step": 16712
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0021927342320355e-05,
+      "loss": 0.255,
+      "step": 16713
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0020930646432299e-05,
+      "loss": 0.2107,
+      "step": 16714
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0019933950336314e-05,
+      "loss": 0.2457,
+      "step": 16715
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0018937254042304e-05,
+      "loss": 0.2776,
+      "step": 16716
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0017940557560173e-05,
+      "loss": 0.2386,
+      "step": 16717
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0016943860899817e-05,
+      "loss": 0.1993,
+      "step": 16718
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0015947164071138e-05,
+      "loss": 0.2526,
+      "step": 16719
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0014950467084042e-05,
+      "loss": 0.1913,
+      "step": 16720
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0013953769948425e-05,
+      "loss": 0.2323,
+      "step": 16721
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.001295707267419e-05,
+      "loss": 0.2528,
+      "step": 16722
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.001196037527124e-05,
+      "loss": 0.242,
+      "step": 16723
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0010963677749472e-05,
+      "loss": 0.203,
+      "step": 16724
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0009966980118792e-05,
+      "loss": 0.2211,
+      "step": 16725
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0008970282389099e-05,
+      "loss": 0.2202,
+      "step": 16726
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0007973584570296e-05,
+      "loss": 0.2539,
+      "step": 16727
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.000697688667228e-05,
+      "loss": 0.1948,
+      "step": 16728
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0005980188704959e-05,
+      "loss": 0.2928,
+      "step": 16729
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0004983490678228e-05,
+      "loss": 0.1897,
+      "step": 16730
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0003986792601988e-05,
+      "loss": 0.2795,
+      "step": 16731
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0002990094486147e-05,
+      "loss": 0.3065,
+      "step": 16732
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0001993396340602e-05,
+      "loss": 0.2887,
+      "step": 16733
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0000996698175252e-05,
+      "loss": 0.2223,
+      "step": 16734
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1e-05,
+      "loss": 0.2142,
+      "step": 16735
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.999003301824751e-06,
+      "loss": 0.272,
+      "step": 16736
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.998006603659402e-06,
+      "loss": 0.2528,
+      "step": 16737
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.997009905513858e-06,
+      "loss": 0.2909,
+      "step": 16738
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.996013207398016e-06,
+      "loss": 0.2564,
+      "step": 16739
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.995016509321776e-06,
+      "loss": 0.2153,
+      "step": 16740
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.994019811295043e-06,
+      "loss": 0.2208,
+      "step": 16741
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99302311332772e-06,
+      "loss": 0.2573,
+      "step": 16742
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.992026415429707e-06,
+      "loss": 0.2601,
+      "step": 16743
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.991029717610905e-06,
+      "loss": 0.276,
+      "step": 16744
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99003301988121e-06,
+      "loss": 0.2177,
+      "step": 16745
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.989036322250531e-06,
+      "loss": 0.1932,
+      "step": 16746
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.988039624728762e-06,
+      "loss": 0.1895,
+      "step": 16747
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.987042927325815e-06,
+      "loss": 0.2394,
+      "step": 16748
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.986046230051578e-06,
+      "loss": 0.2325,
+      "step": 16749
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.985049532915962e-06,
+      "loss": 0.2676,
+      "step": 16750
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.984052835928862e-06,
+      "loss": 0.2186,
+      "step": 16751
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983056139100188e-06,
+      "loss": 0.2455,
+      "step": 16752
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.982059442439832e-06,
+      "loss": 0.2029,
+      "step": 16753
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.981062745957698e-06,
+      "loss": 0.2237,
+      "step": 16754
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98006604966369e-06,
+      "loss": 0.2136,
+      "step": 16755
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.979069353567703e-06,
+      "loss": 0.2135,
+      "step": 16756
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97807265767965e-06,
+      "loss": 0.1521,
+      "step": 16757
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.977075962009418e-06,
+      "loss": 0.2228,
+      "step": 16758
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.976079266566918e-06,
+      "loss": 0.2571,
+      "step": 16759
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.975082571362047e-06,
+      "loss": 0.2346,
+      "step": 16760
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.974085876404708e-06,
+      "loss": 0.2008,
+      "step": 16761
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.973089181704804e-06,
+      "loss": 0.2638,
+      "step": 16762
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.972092487272234e-06,
+      "loss": 0.217,
+      "step": 16763
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.971095793116898e-06,
+      "loss": 0.2641,
+      "step": 16764
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.970099099248694e-06,
+      "loss": 0.184,
+      "step": 16765
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.969102405677532e-06,
+      "loss": 0.1887,
+      "step": 16766
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96810571241331e-06,
+      "loss": 0.3042,
+      "step": 16767
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.967109019465928e-06,
+      "loss": 0.2126,
+      "step": 16768
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.966112326845287e-06,
+      "loss": 0.169,
+      "step": 16769
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.965115634561287e-06,
+      "loss": 0.2822,
+      "step": 16770
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96411894262383e-06,
+      "loss": 0.2426,
+      "step": 16771
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.963122251042823e-06,
+      "loss": 0.2385,
+      "step": 16772
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.962125559828158e-06,
+      "loss": 0.2776,
+      "step": 16773
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96112886898974e-06,
+      "loss": 0.2455,
+      "step": 16774
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960132178537471e-06,
+      "loss": 0.2846,
+      "step": 16775
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.959135488481253e-06,
+      "loss": 0.2718,
+      "step": 16776
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.958138798830988e-06,
+      "loss": 0.2454,
+      "step": 16777
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.957142109596573e-06,
+      "loss": 0.2442,
+      "step": 16778
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.956145420787912e-06,
+      "loss": 0.2591,
+      "step": 16779
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.955148732414902e-06,
+      "loss": 0.2151,
+      "step": 16780
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.954152044487455e-06,
+      "loss": 0.263,
+      "step": 16781
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.953155357015461e-06,
+      "loss": 0.2592,
+      "step": 16782
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.952158670008822e-06,
+      "loss": 0.2473,
+      "step": 16783
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.951161983477445e-06,
+      "loss": 0.2238,
+      "step": 16784
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95016529743123e-06,
+      "loss": 0.1597,
+      "step": 16785
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.949168611880076e-06,
+      "loss": 0.234,
+      "step": 16786
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.948171926833883e-06,
+      "loss": 0.2719,
+      "step": 16787
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.947175242302555e-06,
+      "loss": 0.2636,
+      "step": 16788
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.946178558295992e-06,
+      "loss": 0.2573,
+      "step": 16789
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.945181874824092e-06,
+      "loss": 0.2702,
+      "step": 16790
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.944185191896765e-06,
+      "loss": 0.3294,
+      "step": 16791
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943188509523901e-06,
+      "loss": 0.2809,
+      "step": 16792
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.942191827715409e-06,
+      "loss": 0.2012,
+      "step": 16793
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.941195146481187e-06,
+      "loss": 0.235,
+      "step": 16794
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.940198465831137e-06,
+      "loss": 0.2461,
+      "step": 16795
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.939201785775161e-06,
+      "loss": 0.2536,
+      "step": 16796
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.938205106323158e-06,
+      "loss": 0.2066,
+      "step": 16797
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.937208427485029e-06,
+      "loss": 0.2036,
+      "step": 16798
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.936211749270674e-06,
+      "loss": 0.2259,
+      "step": 16799
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.93521507169e-06,
+      "loss": 0.25,
+      "step": 16800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.934218394752903e-06,
+      "loss": 0.2627,
+      "step": 16801
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.933221718469286e-06,
+      "loss": 0.26,
+      "step": 16802
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.932225042849048e-06,
+      "loss": 0.2276,
+      "step": 16803
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.931228367902092e-06,
+      "loss": 0.2376,
+      "step": 16804
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.930231693638316e-06,
+      "loss": 0.219,
+      "step": 16805
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.929235020067629e-06,
+      "loss": 0.3055,
+      "step": 16806
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.928238347199923e-06,
+      "loss": 0.2881,
+      "step": 16807
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.9272416750451e-06,
+      "loss": 0.2416,
+      "step": 16808
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.926245003613066e-06,
+      "loss": 0.181,
+      "step": 16809
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.925248332913722e-06,
+      "loss": 0.2645,
+      "step": 16810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.924251662956964e-06,
+      "loss": 0.2254,
+      "step": 16811
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.923254993752694e-06,
+      "loss": 0.1914,
+      "step": 16812
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.922258325310816e-06,
+      "loss": 0.1982,
+      "step": 16813
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.921261657641226e-06,
+      "loss": 0.2046,
+      "step": 16814
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.920264990753837e-06,
+      "loss": 0.3033,
+      "step": 16815
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.919268324658536e-06,
+      "loss": 0.274,
+      "step": 16816
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.918271659365227e-06,
+      "loss": 0.2647,
+      "step": 16817
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.917274994883815e-06,
+      "loss": 0.2582,
+      "step": 16818
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.9162783312242e-06,
+      "loss": 0.1782,
+      "step": 16819
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.915281668396283e-06,
+      "loss": 0.2234,
+      "step": 16820
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.914285006409964e-06,
+      "loss": 0.2617,
+      "step": 16821
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.913288345275143e-06,
+      "loss": 0.2176,
+      "step": 16822
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.91229168500172e-06,
+      "loss": 0.2479,
+      "step": 16823
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.911295025599598e-06,
+      "loss": 0.1668,
+      "step": 16824
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.910298367078684e-06,
+      "loss": 0.2404,
+      "step": 16825
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.909301709448865e-06,
+      "loss": 0.1983,
+      "step": 16826
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.908305052720052e-06,
+      "loss": 0.2231,
+      "step": 16827
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.907308396902144e-06,
+      "loss": 0.2666,
+      "step": 16828
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.90631174200504e-06,
+      "loss": 0.2469,
+      "step": 16829
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.905315088038645e-06,
+      "loss": 0.246,
+      "step": 16830
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.904318435012854e-06,
+      "loss": 0.2728,
+      "step": 16831
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.903321782937569e-06,
+      "loss": 0.2159,
+      "step": 16832
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.902325131822694e-06,
+      "loss": 0.2765,
+      "step": 16833
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.901328481678133e-06,
+      "loss": 0.2722,
+      "step": 16834
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.900331832513775e-06,
+      "loss": 0.2598,
+      "step": 16835
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89933518433953e-06,
+      "loss": 0.2264,
+      "step": 16836
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.898338537165297e-06,
+      "loss": 0.2618,
+      "step": 16837
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.897341891000975e-06,
+      "loss": 0.228,
+      "step": 16838
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89634524585647e-06,
+      "loss": 0.2416,
+      "step": 16839
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.895348601741677e-06,
+      "loss": 0.2416,
+      "step": 16840
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.894351958666495e-06,
+      "loss": 0.2323,
+      "step": 16841
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.893355316640832e-06,
+      "loss": 0.2752,
+      "step": 16842
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.892358675674583e-06,
+      "loss": 0.2168,
+      "step": 16843
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.891362035777654e-06,
+      "loss": 0.2171,
+      "step": 16844
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89036539695994e-06,
+      "loss": 0.3015,
+      "step": 16845
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.889368759231345e-06,
+      "loss": 0.2209,
+      "step": 16846
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.888372122601768e-06,
+      "loss": 0.2477,
+      "step": 16847
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.887375487081109e-06,
+      "loss": 0.246,
+      "step": 16848
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.886378852679275e-06,
+      "loss": 0.2134,
+      "step": 16849
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.885382219406158e-06,
+      "loss": 0.2731,
+      "step": 16850
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.884385587271661e-06,
+      "loss": 0.2464,
+      "step": 16851
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.883388956285686e-06,
+      "loss": 0.246,
+      "step": 16852
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.882392326458136e-06,
+      "loss": 0.2353,
+      "step": 16853
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.881395697798911e-06,
+      "loss": 0.2127,
+      "step": 16854
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.880399070317907e-06,
+      "loss": 0.2461,
+      "step": 16855
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.879402444025026e-06,
+      "loss": 0.2229,
+      "step": 16856
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.87840581893017e-06,
+      "loss": 0.2183,
+      "step": 16857
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.87740919504324e-06,
+      "loss": 0.1665,
+      "step": 16858
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.876412572374142e-06,
+      "loss": 0.2791,
+      "step": 16859
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.875415950932763e-06,
+      "loss": 0.215,
+      "step": 16860
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.874419330729012e-06,
+      "loss": 0.2387,
+      "step": 16861
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.87342271177279e-06,
+      "loss": 0.2118,
+      "step": 16862
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.872426094073998e-06,
+      "loss": 0.2379,
+      "step": 16863
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.871429477642532e-06,
+      "loss": 0.2976,
+      "step": 16864
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.870432862488295e-06,
+      "loss": 0.3153,
+      "step": 16865
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.869436248621184e-06,
+      "loss": 0.2177,
+      "step": 16866
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.868439636051106e-06,
+      "loss": 0.2145,
+      "step": 16867
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.867443024787964e-06,
+      "loss": 0.1944,
+      "step": 16868
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.866446414841643e-06,
+      "loss": 0.2266,
+      "step": 16869
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.865449806222059e-06,
+      "loss": 0.2231,
+      "step": 16870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.864453198939103e-06,
+      "loss": 0.2325,
+      "step": 16871
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.86345659300268e-06,
+      "loss": 0.2405,
+      "step": 16872
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.86245998842269e-06,
+      "loss": 0.1987,
+      "step": 16873
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.861463385209032e-06,
+      "loss": 0.2374,
+      "step": 16874
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.860466783371605e-06,
+      "loss": 0.2293,
+      "step": 16875
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.859470182920311e-06,
+      "loss": 0.2595,
+      "step": 16876
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.858473583865052e-06,
+      "loss": 0.1893,
+      "step": 16877
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.857476986215729e-06,
+      "loss": 0.179,
+      "step": 16878
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.856480389982238e-06,
+      "loss": 0.2381,
+      "step": 16879
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.85548379517448e-06,
+      "loss": 0.2242,
+      "step": 16880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.854487201802357e-06,
+      "loss": 0.1972,
+      "step": 16881
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.853490609875766e-06,
+      "loss": 0.2221,
+      "step": 16882
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.852494019404618e-06,
+      "loss": 0.2332,
+      "step": 16883
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.851497430398798e-06,
+      "loss": 0.2153,
+      "step": 16884
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.850500842868215e-06,
+      "loss": 0.2121,
+      "step": 16885
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.849504256822767e-06,
+      "loss": 0.2456,
+      "step": 16886
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.848507672272356e-06,
+      "loss": 0.2364,
+      "step": 16887
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.847511089226883e-06,
+      "loss": 0.2177,
+      "step": 16888
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.846514507696242e-06,
+      "loss": 0.1672,
+      "step": 16889
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.845517927690337e-06,
+      "loss": 0.2637,
+      "step": 16890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.844521349219068e-06,
+      "loss": 0.2499,
+      "step": 16891
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84352477229234e-06,
+      "loss": 0.211,
+      "step": 16892
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.842528196920043e-06,
+      "loss": 0.2668,
+      "step": 16893
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.841531623112083e-06,
+      "loss": 0.2724,
+      "step": 16894
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84053505087836e-06,
+      "loss": 0.1842,
+      "step": 16895
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.839538480228774e-06,
+      "loss": 0.2055,
+      "step": 16896
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.838541911173226e-06,
+      "loss": 0.2077,
+      "step": 16897
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.837545343721611e-06,
+      "loss": 0.234,
+      "step": 16898
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.836548777883835e-06,
+      "loss": 0.2903,
+      "step": 16899
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.835552213669792e-06,
+      "loss": 0.1934,
+      "step": 16900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.834555651089387e-06,
+      "loss": 0.3159,
+      "step": 16901
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.83355909015252e-06,
+      "loss": 0.2224,
+      "step": 16902
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.832562530869088e-06,
+      "loss": 0.2149,
+      "step": 16903
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.831565973248991e-06,
+      "loss": 0.2501,
+      "step": 16904
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.83056941730213e-06,
+      "loss": 0.2118,
+      "step": 16905
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.829572863038404e-06,
+      "loss": 0.2461,
+      "step": 16906
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.828576310467718e-06,
+      "loss": 0.2342,
+      "step": 16907
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.827579759599963e-06,
+      "loss": 0.2515,
+      "step": 16908
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.826583210445041e-06,
+      "loss": 0.158,
+      "step": 16909
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.825586663012859e-06,
+      "loss": 0.2295,
+      "step": 16910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.824590117313308e-06,
+      "loss": 0.2144,
+      "step": 16911
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.823593573356297e-06,
+      "loss": 0.2542,
+      "step": 16912
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.822597031151717e-06,
+      "loss": 0.2496,
+      "step": 16913
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.82160049070947e-06,
+      "loss": 0.2738,
+      "step": 16914
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.820603952039456e-06,
+      "loss": 0.2021,
+      "step": 16915
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.819607415151573e-06,
+      "loss": 0.325,
+      "step": 16916
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.818610880055733e-06,
+      "loss": 0.2218,
+      "step": 16917
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.817614346761816e-06,
+      "loss": 0.1889,
+      "step": 16918
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.816617815279734e-06,
+      "loss": 0.2513,
+      "step": 16919
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.815621285619385e-06,
+      "loss": 0.2381,
+      "step": 16920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.814624757790669e-06,
+      "loss": 0.2279,
+      "step": 16921
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.813628231803482e-06,
+      "loss": 0.2098,
+      "step": 16922
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.812631707667726e-06,
+      "loss": 0.2049,
+      "step": 16923
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.8116351853933e-06,
+      "loss": 0.2099,
+      "step": 16924
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.810638664990101e-06,
+      "loss": 0.2221,
+      "step": 16925
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.809642146468038e-06,
+      "loss": 0.2505,
+      "step": 16926
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.808645629836999e-06,
+      "loss": 0.1896,
+      "step": 16927
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.807649115106888e-06,
+      "loss": 0.2302,
+      "step": 16928
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.806652602287604e-06,
+      "loss": 0.2348,
+      "step": 16929
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.805656091389049e-06,
+      "loss": 0.2498,
+      "step": 16930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.804659582421122e-06,
+      "loss": 0.2529,
+      "step": 16931
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.80366307539372e-06,
+      "loss": 0.1997,
+      "step": 16932
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.802666570316742e-06,
+      "loss": 0.2819,
+      "step": 16933
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.801670067200087e-06,
+      "loss": 0.282,
+      "step": 16934
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.800673566053657e-06,
+      "loss": 0.2223,
+      "step": 16935
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.799677066887353e-06,
+      "loss": 0.2177,
+      "step": 16936
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.79868056971107e-06,
+      "loss": 0.1966,
+      "step": 16937
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.797684074534707e-06,
+      "loss": 0.2376,
+      "step": 16938
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.796687581368167e-06,
+      "loss": 0.2596,
+      "step": 16939
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.795691090221344e-06,
+      "loss": 0.212,
+      "step": 16940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.794694601104148e-06,
+      "loss": 0.1979,
+      "step": 16941
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.793698114026466e-06,
+      "loss": 0.2371,
+      "step": 16942
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.7927016289982e-06,
+      "loss": 0.2439,
+      "step": 16943
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.791705146029253e-06,
+      "loss": 0.2544,
+      "step": 16944
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.790708665129521e-06,
+      "loss": 0.2399,
+      "step": 16945
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.789712186308908e-06,
+      "loss": 0.278,
+      "step": 16946
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.788715709577307e-06,
+      "loss": 0.1952,
+      "step": 16947
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.78771923494462e-06,
+      "loss": 0.2359,
+      "step": 16948
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.786722762420743e-06,
+      "loss": 0.2969,
+      "step": 16949
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.785726292015582e-06,
+      "loss": 0.1885,
+      "step": 16950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.78472982373903e-06,
+      "loss": 0.2108,
+      "step": 16951
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.783733357600984e-06,
+      "loss": 0.22,
+      "step": 16952
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.782736893611349e-06,
+      "loss": 0.2921,
+      "step": 16953
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.781740431780021e-06,
+      "loss": 0.2352,
+      "step": 16954
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.780743972116903e-06,
+      "loss": 0.2503,
+      "step": 16955
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.779747514631887e-06,
+      "loss": 0.2146,
+      "step": 16956
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.778751059334874e-06,
+      "loss": 0.1538,
+      "step": 16957
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.777754606235767e-06,
+      "loss": 0.1945,
+      "step": 16958
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.776758155344456e-06,
+      "loss": 0.2291,
+      "step": 16959
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.775761706670855e-06,
+      "loss": 0.1993,
+      "step": 16960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.774765260224846e-06,
+      "loss": 0.2264,
+      "step": 16961
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.773768816016337e-06,
+      "loss": 0.1787,
+      "step": 16962
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.772772374055227e-06,
+      "loss": 0.194,
+      "step": 16963
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.77177593435141e-06,
+      "loss": 0.2365,
+      "step": 16964
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.770779496914791e-06,
+      "loss": 0.2337,
+      "step": 16965
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.769783061755264e-06,
+      "loss": 0.2309,
+      "step": 16966
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.768786628882728e-06,
+      "loss": 0.2169,
+      "step": 16967
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.767790198307078e-06,
+      "loss": 0.2053,
+      "step": 16968
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.766793770038222e-06,
+      "loss": 0.1981,
+      "step": 16969
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.765797344086055e-06,
+      "loss": 0.2419,
+      "step": 16970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.764800920460472e-06,
+      "loss": 0.2728,
+      "step": 16971
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.763804499171373e-06,
+      "loss": 0.21,
+      "step": 16972
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.762808080228658e-06,
+      "loss": 0.2233,
+      "step": 16973
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.761811663642222e-06,
+      "loss": 0.2378,
+      "step": 16974
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.760815249421973e-06,
+      "loss": 0.2292,
+      "step": 16975
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.7598188375778e-06,
+      "loss": 0.2838,
+      "step": 16976
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.758822428119599e-06,
+      "loss": 0.2179,
+      "step": 16977
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.757826021057277e-06,
+      "loss": 0.2053,
+      "step": 16978
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.756829616400731e-06,
+      "loss": 0.2862,
+      "step": 16979
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.755833214159856e-06,
+      "loss": 0.2642,
+      "step": 16980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.754836814344549e-06,
+      "loss": 0.2442,
+      "step": 16981
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.753840416964714e-06,
+      "loss": 0.1735,
+      "step": 16982
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.752844022030242e-06,
+      "loss": 0.1728,
+      "step": 16983
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.751847629551042e-06,
+      "loss": 0.2367,
+      "step": 16984
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.750851239537002e-06,
+      "loss": 0.1882,
+      "step": 16985
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.749854851998021e-06,
+      "loss": 0.1979,
+      "step": 16986
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.748858466944003e-06,
+      "loss": 0.1932,
+      "step": 16987
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.747862084384842e-06,
+      "loss": 0.2366,
+      "step": 16988
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.746865704330441e-06,
+      "loss": 0.2196,
+      "step": 16989
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.745869326790693e-06,
+      "loss": 0.2839,
+      "step": 16990
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.744872951775495e-06,
+      "loss": 0.1736,
+      "step": 16991
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.743876579294746e-06,
+      "loss": 0.1973,
+      "step": 16992
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.74288020935835e-06,
+      "loss": 0.2444,
+      "step": 16993
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.741883841976204e-06,
+      "loss": 0.3268,
+      "step": 16994
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.740887477158194e-06,
+      "loss": 0.2453,
+      "step": 16995
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.739891114914231e-06,
+      "loss": 0.2833,
+      "step": 16996
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.738894755254208e-06,
+      "loss": 0.2717,
+      "step": 16997
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.737898398188023e-06,
+      "loss": 0.2143,
+      "step": 16998
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.736902043725578e-06,
+      "loss": 0.1888,
+      "step": 16999
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.735905691876762e-06,
+      "loss": 0.254,
+      "step": 17000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.734909342651478e-06,
+      "loss": 0.2563,
+      "step": 17001
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.733912996059625e-06,
+      "loss": 0.2421,
+      "step": 17002
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.732916652111105e-06,
+      "loss": 0.2068,
+      "step": 17003
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.731920310815805e-06,
+      "loss": 0.309,
+      "step": 17004
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.730923972183629e-06,
+      "loss": 0.1988,
+      "step": 17005
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.729927636224472e-06,
+      "loss": 0.2454,
+      "step": 17006
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.728931302948236e-06,
+      "loss": 0.3233,
+      "step": 17007
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.727934972364816e-06,
+      "loss": 0.2302,
+      "step": 17008
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.726938644484109e-06,
+      "loss": 0.2298,
+      "step": 17009
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.725942319316013e-06,
+      "loss": 0.2291,
+      "step": 17010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.724945996870425e-06,
+      "loss": 0.3021,
+      "step": 17011
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.723949677157245e-06,
+      "loss": 0.223,
+      "step": 17012
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.72295336018637e-06,
+      "loss": 0.24,
+      "step": 17013
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.721957045967694e-06,
+      "loss": 0.217,
+      "step": 17014
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.720960734511118e-06,
+      "loss": 0.2217,
+      "step": 17015
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.719964425826537e-06,
+      "loss": 0.2251,
+      "step": 17016
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.718968119923848e-06,
+      "loss": 0.2424,
+      "step": 17017
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.717971816812958e-06,
+      "loss": 0.2711,
+      "step": 17018
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.71697551650375e-06,
+      "loss": 0.2645,
+      "step": 17019
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.715979219006127e-06,
+      "loss": 0.2032,
+      "step": 17020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.71498292432999e-06,
+      "loss": 0.236,
+      "step": 17021
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.713986632485232e-06,
+      "loss": 0.2295,
+      "step": 17022
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.712990343481754e-06,
+      "loss": 0.3054,
+      "step": 17023
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.71199405732945e-06,
+      "loss": 0.2097,
+      "step": 17024
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.710997774038215e-06,
+      "loss": 0.2025,
+      "step": 17025
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.710001493617948e-06,
+      "loss": 0.1909,
+      "step": 17026
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.70900521607855e-06,
+      "loss": 0.2583,
+      "step": 17027
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.70800894142992e-06,
+      "loss": 0.2118,
+      "step": 17028
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.707012669681943e-06,
+      "loss": 0.27,
+      "step": 17029
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.706016400844526e-06,
+      "loss": 0.1544,
+      "step": 17030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.705020134927564e-06,
+      "loss": 0.2543,
+      "step": 17031
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.704023871940954e-06,
+      "loss": 0.2125,
+      "step": 17032
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.703027611894591e-06,
+      "loss": 0.2152,
+      "step": 17033
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.702031354798374e-06,
+      "loss": 0.2015,
+      "step": 17034
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.701035100662195e-06,
+      "loss": 0.1876,
+      "step": 17035
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.700038849495959e-06,
+      "loss": 0.2495,
+      "step": 17036
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.699042601309561e-06,
+      "loss": 0.2808,
+      "step": 17037
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.698046356112892e-06,
+      "loss": 0.2107,
+      "step": 17038
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.697050113915854e-06,
+      "loss": 0.2342,
+      "step": 17039
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.696053874728341e-06,
+      "loss": 0.2274,
+      "step": 17040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.69505763856025e-06,
+      "loss": 0.1919,
+      "step": 17041
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.694061405421482e-06,
+      "loss": 0.2633,
+      "step": 17042
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.693065175321928e-06,
+      "loss": 0.2636,
+      "step": 17043
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.692068948271485e-06,
+      "loss": 0.2365,
+      "step": 17044
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.691072724280053e-06,
+      "loss": 0.205,
+      "step": 17045
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.690076503357526e-06,
+      "loss": 0.1938,
+      "step": 17046
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.689080285513807e-06,
+      "loss": 0.2069,
+      "step": 17047
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.688084070758782e-06,
+      "loss": 0.1455,
+      "step": 17048
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.687087859102354e-06,
+      "loss": 0.2097,
+      "step": 17049
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.686091650554416e-06,
+      "loss": 0.2869,
+      "step": 17050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.685095445124865e-06,
+      "loss": 0.2078,
+      "step": 17051
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.684099242823605e-06,
+      "loss": 0.2463,
+      "step": 17052
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68310304366052e-06,
+      "loss": 0.1844,
+      "step": 17053
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.682106847645514e-06,
+      "loss": 0.24,
+      "step": 17054
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.681110654788483e-06,
+      "loss": 0.2409,
+      "step": 17055
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68011446509932e-06,
+      "loss": 0.2716,
+      "step": 17056
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.679118278587926e-06,
+      "loss": 0.2305,
+      "step": 17057
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.678122095264193e-06,
+      "loss": 0.2247,
+      "step": 17058
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.677125915138017e-06,
+      "loss": 0.2398,
+      "step": 17059
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.676129738219295e-06,
+      "loss": 0.292,
+      "step": 17060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.67513356451793e-06,
+      "loss": 0.24,
+      "step": 17061
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.674137394043804e-06,
+      "loss": 0.2055,
+      "step": 17062
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.673141226806825e-06,
+      "loss": 0.2126,
+      "step": 17063
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.672145062816881e-06,
+      "loss": 0.185,
+      "step": 17064
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.671148902083876e-06,
+      "loss": 0.2605,
+      "step": 17065
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.670152744617702e-06,
+      "loss": 0.2529,
+      "step": 17066
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.669156590428253e-06,
+      "loss": 0.2326,
+      "step": 17067
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.668160439525424e-06,
+      "loss": 0.2216,
+      "step": 17068
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.667164291919113e-06,
+      "loss": 0.2746,
+      "step": 17069
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.66616814761922e-06,
+      "loss": 0.2479,
+      "step": 17070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.665172006635636e-06,
+      "loss": 0.1838,
+      "step": 17071
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.664175868978256e-06,
+      "loss": 0.2199,
+      "step": 17072
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.663179734656978e-06,
+      "loss": 0.1915,
+      "step": 17073
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.662183603681697e-06,
+      "loss": 0.2218,
+      "step": 17074
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.661187476062308e-06,
+      "loss": 0.2253,
+      "step": 17075
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.66019135180871e-06,
+      "loss": 0.2516,
+      "step": 17076
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.659195230930794e-06,
+      "loss": 0.2625,
+      "step": 17077
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.658199113438453e-06,
+      "loss": 0.2143,
+      "step": 17078
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.65720299934159e-06,
+      "loss": 0.1932,
+      "step": 17079
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.656206888650099e-06,
+      "loss": 0.2189,
+      "step": 17080
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.655210781373875e-06,
+      "loss": 0.2455,
+      "step": 17081
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.654214677522808e-06,
+      "loss": 0.2344,
+      "step": 17082
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.6532185771068e-06,
+      "loss": 0.2263,
+      "step": 17083
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.652222480135743e-06,
+      "loss": 0.2534,
+      "step": 17084
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.65122638661953e-06,
+      "loss": 0.2403,
+      "step": 17085
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.650230296568068e-06,
+      "loss": 0.2817,
+      "step": 17086
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.649234209991235e-06,
+      "loss": 0.2704,
+      "step": 17087
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.648238126898938e-06,
+      "loss": 0.2204,
+      "step": 17088
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.647242047301069e-06,
+      "loss": 0.2653,
+      "step": 17089
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.646245971207524e-06,
+      "loss": 0.2767,
+      "step": 17090
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.645249898628197e-06,
+      "loss": 0.2222,
+      "step": 17091
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.64425382957298e-06,
+      "loss": 0.1522,
+      "step": 17092
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.643257764051774e-06,
+      "loss": 0.2016,
+      "step": 17093
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.642261702074469e-06,
+      "loss": 0.2027,
+      "step": 17094
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.641265643650968e-06,
+      "loss": 0.2645,
+      "step": 17095
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.640269588791155e-06,
+      "loss": 0.2122,
+      "step": 17096
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63927353750493e-06,
+      "loss": 0.2211,
+      "step": 17097
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63827748980219e-06,
+      "loss": 0.2065,
+      "step": 17098
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.637281445692826e-06,
+      "loss": 0.2254,
+      "step": 17099
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.636285405186737e-06,
+      "loss": 0.2329,
+      "step": 17100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.635289368293812e-06,
+      "loss": 0.2719,
+      "step": 17101
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.634293335023951e-06,
+      "loss": 0.222,
+      "step": 17102
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.633297305387045e-06,
+      "loss": 0.2383,
+      "step": 17103
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63230127939299e-06,
+      "loss": 0.2535,
+      "step": 17104
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.631305257051683e-06,
+      "loss": 0.2459,
+      "step": 17105
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.630309238373014e-06,
+      "loss": 0.2026,
+      "step": 17106
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.629313223366881e-06,
+      "loss": 0.248,
+      "step": 17107
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.628317212043177e-06,
+      "loss": 0.2955,
+      "step": 17108
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.627321204411797e-06,
+      "loss": 0.2531,
+      "step": 17109
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.626325200482637e-06,
+      "loss": 0.2695,
+      "step": 17110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.625329200265589e-06,
+      "loss": 0.2608,
+      "step": 17111
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.624333203770544e-06,
+      "loss": 0.2799,
+      "step": 17112
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.623337211007402e-06,
+      "loss": 0.2561,
+      "step": 17113
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.622341221986056e-06,
+      "loss": 0.1785,
+      "step": 17114
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.621345236716403e-06,
+      "loss": 0.2532,
+      "step": 17115
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62034925520833e-06,
+      "loss": 0.2805,
+      "step": 17116
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.619353277471735e-06,
+      "loss": 0.1902,
+      "step": 17117
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.618357303516515e-06,
+      "loss": 0.2,
+      "step": 17118
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.617361333352561e-06,
+      "loss": 0.2648,
+      "step": 17119
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.616365366989765e-06,
+      "loss": 0.2443,
+      "step": 17120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.615369404438022e-06,
+      "loss": 0.1933,
+      "step": 17121
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.614373445707228e-06,
+      "loss": 0.2204,
+      "step": 17122
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.613377490807277e-06,
+      "loss": 0.2233,
+      "step": 17123
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.612381539748065e-06,
+      "loss": 0.267,
+      "step": 17124
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.61138559253948e-06,
+      "loss": 0.2196,
+      "step": 17125
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.61038964919142e-06,
+      "loss": 0.2608,
+      "step": 17126
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.609393709713776e-06,
+      "loss": 0.2423,
+      "step": 17127
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60839777411644e-06,
+      "loss": 0.2666,
+      "step": 17128
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.607401842409318e-06,
+      "loss": 0.1956,
+      "step": 17129
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.606405914602285e-06,
+      "loss": 0.1829,
+      "step": 17130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60540999070525e-06,
+      "loss": 0.1673,
+      "step": 17131
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.604414070728099e-06,
+      "loss": 0.224,
+      "step": 17132
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.603418154680726e-06,
+      "loss": 0.2343,
+      "step": 17133
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.602422242573031e-06,
+      "loss": 0.2346,
+      "step": 17134
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.601426334414898e-06,
+      "loss": 0.2408,
+      "step": 17135
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.600430430216226e-06,
+      "loss": 0.1952,
+      "step": 17136
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.599434529986904e-06,
+      "loss": 0.2796,
+      "step": 17137
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.598438633736829e-06,
+      "loss": 0.2502,
+      "step": 17138
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.597442741475898e-06,
+      "loss": 0.3215,
+      "step": 17139
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.596446853213998e-06,
+      "loss": 0.2231,
+      "step": 17140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.595450968961022e-06,
+      "loss": 0.2602,
+      "step": 17141
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.594455088726869e-06,
+      "loss": 0.2253,
+      "step": 17142
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.593459212521424e-06,
+      "loss": 0.2356,
+      "step": 17143
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.592463340354591e-06,
+      "loss": 0.2688,
+      "step": 17144
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.591467472236255e-06,
+      "loss": 0.2475,
+      "step": 17145
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.590471608176307e-06,
+      "loss": 0.2301,
+      "step": 17146
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.589475748184645e-06,
+      "loss": 0.2536,
+      "step": 17147
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.588479892271163e-06,
+      "loss": 0.2056,
+      "step": 17148
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.587484040445751e-06,
+      "loss": 0.2548,
+      "step": 17149
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.586488192718301e-06,
+      "loss": 0.1839,
+      "step": 17150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.585492349098706e-06,
+      "loss": 0.2588,
+      "step": 17151
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.584496509596861e-06,
+      "loss": 0.2153,
+      "step": 17152
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.583500674222662e-06,
+      "loss": 0.2555,
+      "step": 17153
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.582504842985994e-06,
+      "loss": 0.1739,
+      "step": 17154
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.581509015896751e-06,
+      "loss": 0.2773,
+      "step": 17155
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.58051319296483e-06,
+      "loss": 0.2565,
+      "step": 17156
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.579517374200122e-06,
+      "loss": 0.2277,
+      "step": 17157
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57852155961252e-06,
+      "loss": 0.2611,
+      "step": 17158
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.577525749211915e-06,
+      "loss": 0.2628,
+      "step": 17159
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.576529943008199e-06,
+      "loss": 0.2619,
+      "step": 17160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.575534141011263e-06,
+      "loss": 0.2392,
+      "step": 17161
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.574538343231003e-06,
+      "loss": 0.2294,
+      "step": 17162
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573542549677315e-06,
+      "loss": 0.2499,
+      "step": 17163
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57254676036008e-06,
+      "loss": 0.2474,
+      "step": 17164
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.571550975289197e-06,
+      "loss": 0.2251,
+      "step": 17165
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57055519447456e-06,
+      "loss": 0.2585,
+      "step": 17166
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.569559417926058e-06,
+      "loss": 0.3263,
+      "step": 17167
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.568563645653585e-06,
+      "loss": 0.2359,
+      "step": 17168
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.567567877667032e-06,
+      "loss": 0.2829,
+      "step": 17169
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.566572113976287e-06,
+      "loss": 0.2766,
+      "step": 17170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.565576354591248e-06,
+      "loss": 0.2524,
+      "step": 17171
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.56458059952181e-06,
+      "loss": 0.2264,
+      "step": 17172
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.563584848777854e-06,
+      "loss": 0.2186,
+      "step": 17173
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.56258910236928e-06,
+      "loss": 0.3084,
+      "step": 17174
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.561593360305977e-06,
+      "loss": 0.2339,
+      "step": 17175
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.560597622597839e-06,
+      "loss": 0.1984,
+      "step": 17176
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.559601889254756e-06,
+      "loss": 0.2458,
+      "step": 17177
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.558606160286618e-06,
+      "loss": 0.2153,
+      "step": 17178
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55761043570332e-06,
+      "loss": 0.2412,
+      "step": 17179
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.556614715514749e-06,
+      "loss": 0.2814,
+      "step": 17180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.555618999730802e-06,
+      "loss": 0.2308,
+      "step": 17181
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.554623288361369e-06,
+      "loss": 0.2381,
+      "step": 17182
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.553627581416339e-06,
+      "loss": 0.1946,
+      "step": 17183
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.552631878905605e-06,
+      "loss": 0.2678,
+      "step": 17184
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55163618083906e-06,
+      "loss": 0.2867,
+      "step": 17185
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55064048722659e-06,
+      "loss": 0.2601,
+      "step": 17186
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.549644798078097e-06,
+      "loss": 0.2225,
+      "step": 17187
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.54864911340346e-06,
+      "loss": 0.2713,
+      "step": 17188
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.547653433212573e-06,
+      "loss": 0.1981,
+      "step": 17189
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.546657757515333e-06,
+      "loss": 0.2391,
+      "step": 17190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.545662086321628e-06,
+      "loss": 0.231,
+      "step": 17191
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.54466641964135e-06,
+      "loss": 0.2351,
+      "step": 17192
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.543670757484386e-06,
+      "loss": 0.1983,
+      "step": 17193
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.542675099860632e-06,
+      "loss": 0.2673,
+      "step": 17194
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.541679446779972e-06,
+      "loss": 0.1999,
+      "step": 17195
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.540683798252306e-06,
+      "loss": 0.2071,
+      "step": 17196
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.539688154287522e-06,
+      "loss": 0.2142,
+      "step": 17197
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.538692514895505e-06,
+      "loss": 0.2513,
+      "step": 17198
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.537696880086152e-06,
+      "loss": 0.2491,
+      "step": 17199
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.536701249869352e-06,
+      "loss": 0.2174,
+      "step": 17200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.535705624254998e-06,
+      "loss": 0.2276,
+      "step": 17201
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.534710003252975e-06,
+      "loss": 0.2231,
+      "step": 17202
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.533714386873176e-06,
+      "loss": 0.2819,
+      "step": 17203
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.532718775125492e-06,
+      "loss": 0.2346,
+      "step": 17204
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.531723168019814e-06,
+      "loss": 0.2289,
+      "step": 17205
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.530727565566036e-06,
+      "loss": 0.3571,
+      "step": 17206
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.529731967774039e-06,
+      "loss": 0.1984,
+      "step": 17207
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.52873637465372e-06,
+      "loss": 0.2671,
+      "step": 17208
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.527740786214968e-06,
+      "loss": 0.2258,
+      "step": 17209
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.526745202467674e-06,
+      "loss": 0.22,
+      "step": 17210
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.52574962342173e-06,
+      "loss": 0.207,
+      "step": 17211
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.524754049087021e-06,
+      "loss": 0.2201,
+      "step": 17212
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.523758479473437e-06,
+      "loss": 0.2139,
+      "step": 17213
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.522762914590875e-06,
+      "loss": 0.1893,
+      "step": 17214
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.521767354449218e-06,
+      "loss": 0.2221,
+      "step": 17215
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.520771799058363e-06,
+      "loss": 0.2547,
+      "step": 17216
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.519776248428191e-06,
+      "loss": 0.2491,
+      "step": 17217
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.5187807025686e-06,
+      "loss": 0.203,
+      "step": 17218
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.517785161489474e-06,
+      "loss": 0.2054,
+      "step": 17219
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.516789625200704e-06,
+      "loss": 0.2336,
+      "step": 17220
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.515794093712188e-06,
+      "loss": 0.188,
+      "step": 17221
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.514798567033801e-06,
+      "loss": 0.2875,
+      "step": 17222
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.513803045175444e-06,
+      "loss": 0.2431,
+      "step": 17223
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.512807528147002e-06,
+      "loss": 0.2467,
+      "step": 17224
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.511812015958364e-06,
+      "loss": 0.2925,
+      "step": 17225
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.510816508619423e-06,
+      "loss": 0.264,
+      "step": 17226
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.509821006140066e-06,
+      "loss": 0.2087,
+      "step": 17227
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.508825508530182e-06,
+      "loss": 0.2688,
+      "step": 17228
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.507830015799658e-06,
+      "loss": 0.2089,
+      "step": 17229
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.506834527958394e-06,
+      "loss": 0.247,
+      "step": 17230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.505839045016265e-06,
+      "loss": 0.262,
+      "step": 17231
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.504843566983169e-06,
+      "loss": 0.1679,
+      "step": 17232
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.503848093868993e-06,
+      "loss": 0.2249,
+      "step": 17233
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.502852625683625e-06,
+      "loss": 0.229,
+      "step": 17234
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.501857162436957e-06,
+      "loss": 0.2544,
+      "step": 17235
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.500861704138874e-06,
+      "loss": 0.2777,
+      "step": 17236
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.499866250799268e-06,
+      "loss": 0.288,
+      "step": 17237
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.498870802428026e-06,
+      "loss": 0.1876,
+      "step": 17238
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.497875359035038e-06,
+      "loss": 0.2379,
+      "step": 17239
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.496879920630198e-06,
+      "loss": 0.2501,
+      "step": 17240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.495884487223381e-06,
+      "loss": 0.2494,
+      "step": 17241
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.494889058824488e-06,
+      "loss": 0.216,
+      "step": 17242
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.493893635443404e-06,
+      "loss": 0.2021,
+      "step": 17243
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.492898217090016e-06,
+      "loss": 0.2821,
+      "step": 17244
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.491902803774216e-06,
+      "loss": 0.2422,
+      "step": 17245
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.49090739550589e-06,
+      "loss": 0.2595,
+      "step": 17246
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.489911992294924e-06,
+      "loss": 0.2509,
+      "step": 17247
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.48891659415121e-06,
+      "loss": 0.2791,
+      "step": 17248
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.487921201084636e-06,
+      "loss": 0.2511,
+      "step": 17249
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.486925813105094e-06,
+      "loss": 0.2273,
+      "step": 17250
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.485930430222464e-06,
+      "loss": 0.264,
+      "step": 17251
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.48493505244664e-06,
+      "loss": 0.2405,
+      "step": 17252
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.483939679787509e-06,
+      "loss": 0.2076,
+      "step": 17253
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.482944312254954e-06,
+      "loss": 0.2373,
+      "step": 17254
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.481948949858876e-06,
+      "loss": 0.185,
+      "step": 17255
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.480953592609149e-06,
+      "loss": 0.1833,
+      "step": 17256
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.479958240515667e-06,
+      "loss": 0.2779,
+      "step": 17257
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.478962893588317e-06,
+      "loss": 0.2961,
+      "step": 17258
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.47796755183699e-06,
+      "loss": 0.2291,
+      "step": 17259
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.476972215271571e-06,
+      "loss": 0.243,
+      "step": 17260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.475976883901947e-06,
+      "loss": 0.2411,
+      "step": 17261
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.474981557738005e-06,
+      "loss": 0.1811,
+      "step": 17262
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.473986236789634e-06,
+      "loss": 0.2467,
+      "step": 17263
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.472990921066729e-06,
+      "loss": 0.3265,
+      "step": 17264
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.471995610579161e-06,
+      "loss": 0.2338,
+      "step": 17265
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.471000305336832e-06,
+      "loss": 0.2656,
+      "step": 17266
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.470005005349622e-06,
+      "loss": 0.2535,
+      "step": 17267
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.469009710627424e-06,
+      "loss": 0.2182,
+      "step": 17268
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468014421180121e-06,
+      "loss": 0.2338,
+      "step": 17269
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467019137017601e-06,
+      "loss": 0.2684,
+      "step": 17270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.466023858149752e-06,
+      "loss": 0.255,
+      "step": 17271
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.465028584586456e-06,
+      "loss": 0.2589,
+      "step": 17272
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.464033316337608e-06,
+      "loss": 0.2501,
+      "step": 17273
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.463038053413096e-06,
+      "loss": 0.2001,
+      "step": 17274
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.462042795822799e-06,
+      "loss": 0.2087,
+      "step": 17275
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.461047543576607e-06,
+      "loss": 0.2606,
+      "step": 17276
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.46005229668441e-06,
+      "loss": 0.2417,
+      "step": 17277
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.459057055156092e-06,
+      "loss": 0.322,
+      "step": 17278
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.458061819001542e-06,
+      "loss": 0.2756,
+      "step": 17279
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.457066588230643e-06,
+      "loss": 0.2298,
+      "step": 17280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.456071362853283e-06,
+      "loss": 0.1986,
+      "step": 17281
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45507614287935e-06,
+      "loss": 0.2055,
+      "step": 17282
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.454080928318732e-06,
+      "loss": 0.2861,
+      "step": 17283
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.453085719181315e-06,
+      "loss": 0.1819,
+      "step": 17284
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.452090515476982e-06,
+      "loss": 0.2877,
+      "step": 17285
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45109531721562e-06,
+      "loss": 0.2488,
+      "step": 17286
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.450100124407118e-06,
+      "loss": 0.2182,
+      "step": 17287
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.449104937061365e-06,
+      "loss": 0.2691,
+      "step": 17288
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44810975518824e-06,
+      "loss": 0.1969,
+      "step": 17289
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44711457879763e-06,
+      "loss": 0.2738,
+      "step": 17290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.446119407899427e-06,
+      "loss": 0.233,
+      "step": 17291
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.445124242503515e-06,
+      "loss": 0.2922,
+      "step": 17292
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44412908261978e-06,
+      "loss": 0.261,
+      "step": 17293
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.443133928258104e-06,
+      "loss": 0.1984,
+      "step": 17294
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.442138779428376e-06,
+      "loss": 0.2047,
+      "step": 17295
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.441143636140483e-06,
+      "loss": 0.2018,
+      "step": 17296
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.440148498404307e-06,
+      "loss": 0.2578,
+      "step": 17297
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.439153366229742e-06,
+      "loss": 0.1925,
+      "step": 17298
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.438158239626663e-06,
+      "loss": 0.2591,
+      "step": 17299
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.437163118604962e-06,
+      "loss": 0.2702,
+      "step": 17300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.436168003174526e-06,
+      "loss": 0.2324,
+      "step": 17301
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.435172893345235e-06,
+      "loss": 0.2733,
+      "step": 17302
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43417778912698e-06,
+      "loss": 0.2435,
+      "step": 17303
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.433182690529643e-06,
+      "loss": 0.2178,
+      "step": 17304
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43218759756311e-06,
+      "loss": 0.1698,
+      "step": 17305
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.431192510237264e-06,
+      "loss": 0.2421,
+      "step": 17306
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.430197428561995e-06,
+      "loss": 0.3004,
+      "step": 17307
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.429202352547188e-06,
+      "loss": 0.2287,
+      "step": 17308
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.428207282202724e-06,
+      "loss": 0.2388,
+      "step": 17309
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.427212217538491e-06,
+      "loss": 0.2868,
+      "step": 17310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.426217158564374e-06,
+      "loss": 0.2315,
+      "step": 17311
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.425222105290259e-06,
+      "loss": 0.2278,
+      "step": 17312
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.424227057726025e-06,
+      "loss": 0.2264,
+      "step": 17313
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.423232015881564e-06,
+      "loss": 0.2694,
+      "step": 17314
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.422236979766755e-06,
+      "loss": 0.1684,
+      "step": 17315
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.421241949391488e-06,
+      "loss": 0.2429,
+      "step": 17316
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.420246924765646e-06,
+      "loss": 0.233,
+      "step": 17317
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.419251905899112e-06,
+      "loss": 0.201,
+      "step": 17318
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.418256892801772e-06,
+      "loss": 0.2915,
+      "step": 17319
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.41726188548351e-06,
+      "loss": 0.2272,
+      "step": 17320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.416266883954209e-06,
+      "loss": 0.2703,
+      "step": 17321
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.415271888223763e-06,
+      "loss": 0.2264,
+      "step": 17322
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.414276898302042e-06,
+      "loss": 0.1727,
+      "step": 17323
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.413281914198935e-06,
+      "loss": 0.2301,
+      "step": 17324
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.41228693592433e-06,
+      "loss": 0.2393,
+      "step": 17325
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.41129196348811e-06,
+      "loss": 0.3095,
+      "step": 17326
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.41029699690016e-06,
+      "loss": 0.206,
+      "step": 17327
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.40930203617036e-06,
+      "loss": 0.2575,
+      "step": 17328
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.408307081308597e-06,
+      "loss": 0.25,
+      "step": 17329
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.407312132324753e-06,
+      "loss": 0.2137,
+      "step": 17330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.406317189228714e-06,
+      "loss": 0.2005,
+      "step": 17331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.405322252030367e-06,
+      "loss": 0.3275,
+      "step": 17332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.404327320739588e-06,
+      "loss": 0.2254,
+      "step": 17333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.403332395366263e-06,
+      "loss": 0.2354,
+      "step": 17334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.40233747592028e-06,
+      "loss": 0.2016,
+      "step": 17335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.40134256241152e-06,
+      "loss": 0.2441,
+      "step": 17336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.400347654849868e-06,
+      "loss": 0.2503,
+      "step": 17337
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.399352753245203e-06,
+      "loss": 0.259,
+      "step": 17338
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.398357857607412e-06,
+      "loss": 0.2166,
+      "step": 17339
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.397362967946376e-06,
+      "loss": 0.2237,
+      "step": 17340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.396368084271985e-06,
+      "loss": 0.2933,
+      "step": 17341
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.395373206594112e-06,
+      "loss": 0.2805,
+      "step": 17342
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.394378334922647e-06,
+      "loss": 0.2298,
+      "step": 17343
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.393383469267472e-06,
+      "loss": 0.2453,
+      "step": 17344
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39238860963847e-06,
+      "loss": 0.2188,
+      "step": 17345
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.391393756045524e-06,
+      "loss": 0.2405,
+      "step": 17346
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.390398908498515e-06,
+      "loss": 0.2681,
+      "step": 17347
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.389404067007328e-06,
+      "loss": 0.2625,
+      "step": 17348
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.388409231581845e-06,
+      "loss": 0.2271,
+      "step": 17349
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.38741440223195e-06,
+      "loss": 0.2346,
+      "step": 17350
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.386419578967525e-06,
+      "loss": 0.3187,
+      "step": 17351
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.385424761798452e-06,
+      "loss": 0.2184,
+      "step": 17352
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.384429950734614e-06,
+      "loss": 0.2309,
+      "step": 17353
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.383435145785893e-06,
+      "loss": 0.2353,
+      "step": 17354
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.382440346962169e-06,
+      "loss": 0.1931,
+      "step": 17355
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.381445554273335e-06,
+      "loss": 0.2659,
+      "step": 17356
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.380450767729262e-06,
+      "loss": 0.2802,
+      "step": 17357
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.379455987339832e-06,
+      "loss": 0.2105,
+      "step": 17358
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.378461213114936e-06,
+      "loss": 0.2313,
+      "step": 17359
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37746644506445e-06,
+      "loss": 0.2045,
+      "step": 17360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.376471683198259e-06,
+      "loss": 0.1975,
+      "step": 17361
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.375476927526242e-06,
+      "loss": 0.1832,
+      "step": 17362
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.374482178058283e-06,
+      "loss": 0.2255,
+      "step": 17363
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37348743480426e-06,
+      "loss": 0.2475,
+      "step": 17364
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.372492697774062e-06,
+      "loss": 0.2417,
+      "step": 17365
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37149796697757e-06,
+      "loss": 0.2472,
+      "step": 17366
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.370503242424657e-06,
+      "loss": 0.2889,
+      "step": 17367
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.369508524125213e-06,
+      "loss": 0.1975,
+      "step": 17368
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.368513812089115e-06,
+      "loss": 0.2329,
+      "step": 17369
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.367519106326252e-06,
+      "loss": 0.2894,
+      "step": 17370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.366524406846496e-06,
+      "loss": 0.2227,
+      "step": 17371
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.365529713659732e-06,
+      "loss": 0.2131,
+      "step": 17372
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.36453502677584e-06,
+      "loss": 0.2359,
+      "step": 17373
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.363540346204707e-06,
+      "loss": 0.2265,
+      "step": 17374
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.362545671956213e-06,
+      "loss": 0.1623,
+      "step": 17375
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.361551004040232e-06,
+      "loss": 0.2041,
+      "step": 17376
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.360556342466648e-06,
+      "loss": 0.2687,
+      "step": 17377
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.359561687245348e-06,
+      "loss": 0.2026,
+      "step": 17378
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.358567038386205e-06,
+      "loss": 0.2379,
+      "step": 17379
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.357572395899108e-06,
+      "loss": 0.2828,
+      "step": 17380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.35657775979393e-06,
+      "loss": 0.2781,
+      "step": 17381
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.355583130080555e-06,
+      "loss": 0.2149,
+      "step": 17382
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.354588506768864e-06,
+      "loss": 0.2618,
+      "step": 17383
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.353593889868739e-06,
+      "loss": 0.2901,
+      "step": 17384
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.352599279390062e-06,
+      "loss": 0.267,
+      "step": 17385
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.351604675342707e-06,
+      "loss": 0.2027,
+      "step": 17386
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.350610077736558e-06,
+      "loss": 0.2397,
+      "step": 17387
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.349615486581496e-06,
+      "loss": 0.239,
+      "step": 17388
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3486209018874e-06,
+      "loss": 0.2721,
+      "step": 17389
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.347626323664157e-06,
+      "loss": 0.208,
+      "step": 17390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.346631751921633e-06,
+      "loss": 0.195,
+      "step": 17391
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.345637186669721e-06,
+      "loss": 0.1992,
+      "step": 17392
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.344642627918296e-06,
+      "loss": 0.1939,
+      "step": 17393
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.343648075677238e-06,
+      "loss": 0.1692,
+      "step": 17394
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.342653529956432e-06,
+      "loss": 0.1803,
+      "step": 17395
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.34165899076575e-06,
+      "loss": 0.243,
+      "step": 17396
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.340664458115074e-06,
+      "loss": 0.2307,
+      "step": 17397
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.339669932014286e-06,
+      "loss": 0.2394,
+      "step": 17398
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.338675412473269e-06,
+      "loss": 0.2308,
+      "step": 17399
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.337680899501892e-06,
+      "loss": 0.2003,
+      "step": 17400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.336686393110042e-06,
+      "loss": 0.2683,
+      "step": 17401
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3356918933076e-06,
+      "loss": 0.2325,
+      "step": 17402
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.334697400104441e-06,
+      "loss": 0.1905,
+      "step": 17403
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.33370291351045e-06,
+      "loss": 0.2397,
+      "step": 17404
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3327084335355e-06,
+      "loss": 0.2146,
+      "step": 17405
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.331713960189473e-06,
+      "loss": 0.2142,
+      "step": 17406
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.330719493482246e-06,
+      "loss": 0.2153,
+      "step": 17407
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.329725033423701e-06,
+      "loss": 0.2161,
+      "step": 17408
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.328730580023722e-06,
+      "loss": 0.2275,
+      "step": 17409
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.327736133292175e-06,
+      "loss": 0.1853,
+      "step": 17410
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.326741693238948e-06,
+      "loss": 0.2268,
+      "step": 17411
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.325747259873918e-06,
+      "loss": 0.2183,
+      "step": 17412
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.324752833206962e-06,
+      "loss": 0.2135,
+      "step": 17413
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.323758413247965e-06,
+      "loss": 0.1954,
+      "step": 17414
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.322764000006798e-06,
+      "loss": 0.178,
+      "step": 17415
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.321769593493341e-06,
+      "loss": 0.1994,
+      "step": 17416
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.320775193717476e-06,
+      "loss": 0.1927,
+      "step": 17417
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.319780800689078e-06,
+      "loss": 0.2828,
+      "step": 17418
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.31878641441803e-06,
+      "loss": 0.1944,
+      "step": 17419
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.317792034914204e-06,
+      "loss": 0.2699,
+      "step": 17420
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.316797662187484e-06,
+      "loss": 0.1973,
+      "step": 17421
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.315803296247743e-06,
+      "loss": 0.2416,
+      "step": 17422
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.314808937104862e-06,
+      "loss": 0.239,
+      "step": 17423
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.313814584768723e-06,
+      "loss": 0.2156,
+      "step": 17424
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.312820239249194e-06,
+      "loss": 0.2339,
+      "step": 17425
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.311825900556161e-06,
+      "loss": 0.2092,
+      "step": 17426
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.310831568699498e-06,
+      "loss": 0.2635,
+      "step": 17427
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.30983724368909e-06,
+      "loss": 0.3151,
+      "step": 17428
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.308842925534803e-06,
+      "loss": 0.2534,
+      "step": 17429
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.307848614246522e-06,
+      "loss": 0.2168,
+      "step": 17430
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.306854309834123e-06,
+      "loss": 0.2536,
+      "step": 17431
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.305860012307481e-06,
+      "loss": 0.2274,
+      "step": 17432
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.304865721676483e-06,
+      "loss": 0.2118,
+      "step": 17433
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.303871437950992e-06,
+      "loss": 0.2211,
+      "step": 17434
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.302877161140897e-06,
+      "loss": 0.1757,
+      "step": 17435
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.301882891256069e-06,
+      "loss": 0.2393,
+      "step": 17436
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.300888628306387e-06,
+      "loss": 0.2378,
+      "step": 17437
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.299894372301732e-06,
+      "loss": 0.2591,
+      "step": 17438
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.298900123251973e-06,
+      "loss": 0.2145,
+      "step": 17439
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.297905881166991e-06,
+      "loss": 0.2334,
+      "step": 17440
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.296911646056661e-06,
+      "loss": 0.2278,
+      "step": 17441
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.295917417930864e-06,
+      "loss": 0.2589,
+      "step": 17442
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.294923196799478e-06,
+      "loss": 0.2362,
+      "step": 17443
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.293928982672373e-06,
+      "loss": 0.2394,
+      "step": 17444
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.292934775559428e-06,
+      "loss": 0.2385,
+      "step": 17445
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.291940575470521e-06,
+      "loss": 0.2931,
+      "step": 17446
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.290946382415527e-06,
+      "loss": 0.2522,
+      "step": 17447
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.289952196404326e-06,
+      "loss": 0.2274,
+      "step": 17448
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.288958017446789e-06,
+      "loss": 0.2447,
+      "step": 17449
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.287963845552791e-06,
+      "loss": 0.2978,
+      "step": 17450
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.286969680732217e-06,
+      "loss": 0.2222,
+      "step": 17451
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.285975522994936e-06,
+      "loss": 0.2337,
+      "step": 17452
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.284981372350828e-06,
+      "loss": 0.2296,
+      "step": 17453
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.283987228809766e-06,
+      "loss": 0.2046,
+      "step": 17454
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.282993092381626e-06,
+      "loss": 0.3025,
+      "step": 17455
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.281998963076285e-06,
+      "loss": 0.2427,
+      "step": 17456
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.28100484090362e-06,
+      "loss": 0.2276,
+      "step": 17457
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.280010725873503e-06,
+      "loss": 0.2804,
+      "step": 17458
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.279016617995808e-06,
+      "loss": 0.2653,
+      "step": 17459
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.278022517280419e-06,
+      "loss": 0.2553,
+      "step": 17460
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.277028423737205e-06,
+      "loss": 0.2762,
+      "step": 17461
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.276034337376047e-06,
+      "loss": 0.2638,
+      "step": 17462
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.275040258206813e-06,
+      "loss": 0.2577,
+      "step": 17463
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.274046186239381e-06,
+      "loss": 0.2527,
+      "step": 17464
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.273052121483628e-06,
+      "loss": 0.2585,
+      "step": 17465
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.272058063949426e-06,
+      "loss": 0.2387,
+      "step": 17466
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.271064013646658e-06,
+      "loss": 0.2248,
+      "step": 17467
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.270069970585188e-06,
+      "loss": 0.2816,
+      "step": 17468
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.269075934774895e-06,
+      "loss": 0.2333,
+      "step": 17469
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.268081906225656e-06,
+      "loss": 0.2012,
+      "step": 17470
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.267087884947346e-06,
+      "loss": 0.2832,
+      "step": 17471
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.266093870949837e-06,
+      "loss": 0.2178,
+      "step": 17472
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.265099864243006e-06,
+      "loss": 0.2975,
+      "step": 17473
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.264105864836725e-06,
+      "loss": 0.2431,
+      "step": 17474
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.263111872740866e-06,
+      "loss": 0.2594,
+      "step": 17475
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.262117887965312e-06,
+      "loss": 0.2099,
+      "step": 17476
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.261123910519933e-06,
+      "loss": 0.1638,
+      "step": 17477
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2601299404146e-06,
+      "loss": 0.2209,
+      "step": 17478
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.25913597765919e-06,
+      "loss": 0.2617,
+      "step": 17479
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.258142022263579e-06,
+      "loss": 0.2247,
+      "step": 17480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.25714807423764e-06,
+      "loss": 0.2534,
+      "step": 17481
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.256154133591243e-06,
+      "loss": 0.264,
+      "step": 17482
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255160200334266e-06,
+      "loss": 0.2251,
+      "step": 17483
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.254166274476579e-06,
+      "loss": 0.2193,
+      "step": 17484
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.253172356028058e-06,
+      "loss": 0.2486,
+      "step": 17485
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.252178444998582e-06,
+      "loss": 0.2162,
+      "step": 17486
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.251184541398017e-06,
+      "loss": 0.2256,
+      "step": 17487
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.250190645236239e-06,
+      "loss": 0.1706,
+      "step": 17488
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.249196756523122e-06,
+      "loss": 0.239,
+      "step": 17489
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248202875268534e-06,
+      "loss": 0.1771,
+      "step": 17490
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.247209001482363e-06,
+      "loss": 0.2555,
+      "step": 17491
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.246215135174468e-06,
+      "loss": 0.2337,
+      "step": 17492
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.245221276354723e-06,
+      "loss": 0.2479,
+      "step": 17493
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244227425033007e-06,
+      "loss": 0.1964,
+      "step": 17494
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.24323358121919e-06,
+      "loss": 0.1796,
+      "step": 17495
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.242239744923149e-06,
+      "loss": 0.2687,
+      "step": 17496
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.24124591615475e-06,
+      "loss": 0.3704,
+      "step": 17497
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.24025209492387e-06,
+      "loss": 0.2624,
+      "step": 17498
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.23925828124038e-06,
+      "loss": 0.2728,
+      "step": 17499
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.238264475114152e-06,
+      "loss": 0.2283,
+      "step": 17500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.237270676555066e-06,
+      "loss": 0.2376,
+      "step": 17501
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.236276885572981e-06,
+      "loss": 0.1999,
+      "step": 17502
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.23528310217778e-06,
+      "loss": 0.2646,
+      "step": 17503
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234289326379334e-06,
+      "loss": 0.2225,
+      "step": 17504
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.23329555818751e-06,
+      "loss": 0.2424,
+      "step": 17505
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.232301797612187e-06,
+      "loss": 0.2194,
+      "step": 17506
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231308044663232e-06,
+      "loss": 0.2253,
+      "step": 17507
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.230314299350519e-06,
+      "loss": 0.2553,
+      "step": 17508
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.229320561683917e-06,
+      "loss": 0.2355,
+      "step": 17509
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.228326831673306e-06,
+      "loss": 0.2375,
+      "step": 17510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.227333109328547e-06,
+      "loss": 0.2453,
+      "step": 17511
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.226339394659518e-06,
+      "loss": 0.1986,
+      "step": 17512
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.225345687676091e-06,
+      "loss": 0.2557,
+      "step": 17513
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.224351988388135e-06,
+      "loss": 0.2723,
+      "step": 17514
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.223358296805526e-06,
+      "loss": 0.2456,
+      "step": 17515
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.222364612938128e-06,
+      "loss": 0.246,
+      "step": 17516
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.221370936795818e-06,
+      "loss": 0.206,
+      "step": 17517
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.220377268388464e-06,
+      "loss": 0.2299,
+      "step": 17518
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.219383607725938e-06,
+      "loss": 0.2466,
+      "step": 17519
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.218389954818116e-06,
+      "loss": 0.2112,
+      "step": 17520
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.217396309674863e-06,
+      "loss": 0.2654,
+      "step": 17521
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.21640267230605e-06,
+      "loss": 0.2345,
+      "step": 17522
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.215409042721553e-06,
+      "loss": 0.2323,
+      "step": 17523
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.214415420931235e-06,
+      "loss": 0.2753,
+      "step": 17524
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.213421806944978e-06,
+      "loss": 0.2024,
+      "step": 17525
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.212428200772641e-06,
+      "loss": 0.1894,
+      "step": 17526
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.211434602424097e-06,
+      "loss": 0.2829,
+      "step": 17527
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.210441011909222e-06,
+      "loss": 0.3183,
+      "step": 17528
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.209447429237884e-06,
+      "loss": 0.2119,
+      "step": 17529
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.208453854419954e-06,
+      "loss": 0.1952,
+      "step": 17530
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.207460287465298e-06,
+      "loss": 0.245,
+      "step": 17531
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.20646672838379e-06,
+      "loss": 0.2415,
+      "step": 17532
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.205473177185296e-06,
+      "loss": 0.2327,
+      "step": 17533
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.204479633879692e-06,
+      "loss": 0.2664,
+      "step": 17534
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.20348609847685e-06,
+      "loss": 0.1962,
+      "step": 17535
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.202492570986626e-06,
+      "loss": 0.2236,
+      "step": 17536
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.201499051418903e-06,
+      "loss": 0.243,
+      "step": 17537
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.200505539783545e-06,
+      "loss": 0.2757,
+      "step": 17538
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.199512036090427e-06,
+      "loss": 0.1941,
+      "step": 17539
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19851854034941e-06,
+      "loss": 0.295,
+      "step": 17540
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19752505257037e-06,
+      "loss": 0.2531,
+      "step": 17541
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.196531572763171e-06,
+      "loss": 0.2326,
+      "step": 17542
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.195538100937688e-06,
+      "loss": 0.2516,
+      "step": 17543
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.194544637103794e-06,
+      "loss": 0.2464,
+      "step": 17544
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.193551181271341e-06,
+      "loss": 0.2329,
+      "step": 17545
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.192557733450216e-06,
+      "loss": 0.2041,
+      "step": 17546
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.191564293650278e-06,
+      "loss": 0.173,
+      "step": 17547
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.1905708618814e-06,
+      "loss": 0.2695,
+      "step": 17548
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.189577438153452e-06,
+      "loss": 0.2682,
+      "step": 17549
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.188584022476299e-06,
+      "loss": 0.2116,
+      "step": 17550
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.187590614859808e-06,
+      "loss": 0.2224,
+      "step": 17551
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.186597215313854e-06,
+      "loss": 0.2468,
+      "step": 17552
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.1856038238483e-06,
+      "loss": 0.2052,
+      "step": 17553
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.184610440473022e-06,
+      "loss": 0.2154,
+      "step": 17554
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.183617065197879e-06,
+      "loss": 0.2738,
+      "step": 17555
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.182623698032744e-06,
+      "loss": 0.1629,
+      "step": 17556
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.181630338987484e-06,
+      "loss": 0.1958,
+      "step": 17557
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.180636988071966e-06,
+      "loss": 0.2363,
+      "step": 17558
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.179643645296066e-06,
+      "loss": 0.272,
+      "step": 17559
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.17865031066964e-06,
+      "loss": 0.249,
+      "step": 17560
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.17765698420256e-06,
+      "loss": 0.1793,
+      "step": 17561
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.176663665904698e-06,
+      "loss": 0.242,
+      "step": 17562
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175670355785919e-06,
+      "loss": 0.2038,
+      "step": 17563
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.174677053856092e-06,
+      "loss": 0.1963,
+      "step": 17564
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.17368376012508e-06,
+      "loss": 0.1818,
+      "step": 17565
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.172690474602754e-06,
+      "loss": 0.2733,
+      "step": 17566
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.171697197298979e-06,
+      "loss": 0.2426,
+      "step": 17567
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.17070392822363e-06,
+      "loss": 0.2062,
+      "step": 17568
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.169710667386565e-06,
+      "loss": 0.2782,
+      "step": 17569
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.168717414797652e-06,
+      "loss": 0.1718,
+      "step": 17570
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.167724170466762e-06,
+      "loss": 0.2374,
+      "step": 17571
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.166730934403762e-06,
+      "loss": 0.2472,
+      "step": 17572
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.165737706618519e-06,
+      "loss": 0.2403,
+      "step": 17573
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.164744487120895e-06,
+      "loss": 0.2208,
+      "step": 17574
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.163751275920762e-06,
+      "loss": 0.2291,
+      "step": 17575
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.16275807302798e-06,
+      "loss": 0.2559,
+      "step": 17576
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.161764878452422e-06,
+      "loss": 0.2548,
+      "step": 17577
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.160771692203959e-06,
+      "loss": 0.2656,
+      "step": 17578
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.159778514292444e-06,
+      "loss": 0.2574,
+      "step": 17579
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.158785344727751e-06,
+      "loss": 0.2561,
+      "step": 17580
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.157792183519746e-06,
+      "loss": 0.2692,
+      "step": 17581
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.156799030678294e-06,
+      "loss": 0.1967,
+      "step": 17582
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.155805886213264e-06,
+      "loss": 0.2346,
+      "step": 17583
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.154812750134518e-06,
+      "loss": 0.2642,
+      "step": 17584
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.153819622451921e-06,
+      "loss": 0.2565,
+      "step": 17585
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.152826503175344e-06,
+      "loss": 0.2313,
+      "step": 17586
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.15183339231465e-06,
+      "loss": 0.2221,
+      "step": 17587
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.150840289879706e-06,
+      "loss": 0.2553,
+      "step": 17588
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.149847195880374e-06,
+      "loss": 0.2415,
+      "step": 17589
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.148854110326521e-06,
+      "loss": 0.2738,
+      "step": 17590
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.147861033228014e-06,
+      "loss": 0.2066,
+      "step": 17591
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.146867964594715e-06,
+      "loss": 0.2255,
+      "step": 17592
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.1458749044365e-06,
+      "loss": 0.2044,
+      "step": 17593
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.144881852763216e-06,
+      "loss": 0.2009,
+      "step": 17594
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.143888809584741e-06,
+      "loss": 0.2426,
+      "step": 17595
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.142895774910938e-06,
+      "loss": 0.2435,
+      "step": 17596
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14190274875167e-06,
+      "loss": 0.2384,
+      "step": 17597
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.140909731116802e-06,
+      "loss": 0.2338,
+      "step": 17598
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.1399167220162e-06,
+      "loss": 0.2214,
+      "step": 17599
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.138923721459726e-06,
+      "loss": 0.3088,
+      "step": 17600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.137930729457244e-06,
+      "loss": 0.2654,
+      "step": 17601
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.136937746018629e-06,
+      "loss": 0.2015,
+      "step": 17602
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.135944771153729e-06,
+      "loss": 0.24,
+      "step": 17603
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.134951804872418e-06,
+      "loss": 0.1842,
+      "step": 17604
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.133958847184559e-06,
+      "loss": 0.202,
+      "step": 17605
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.132965898100016e-06,
+      "loss": 0.2524,
+      "step": 17606
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.131972957628653e-06,
+      "loss": 0.2363,
+      "step": 17607
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.130980025780335e-06,
+      "loss": 0.2281,
+      "step": 17608
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.129987102564923e-06,
+      "loss": 0.1983,
+      "step": 17609
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.12899418799228e-06,
+      "loss": 0.2463,
+      "step": 17610
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.128001282072274e-06,
+      "loss": 0.2623,
+      "step": 17611
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.127008384814767e-06,
+      "loss": 0.2536,
+      "step": 17612
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.126015496229622e-06,
+      "loss": 0.2415,
+      "step": 17613
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.125022616326702e-06,
+      "loss": 0.2897,
+      "step": 17614
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.124029745115872e-06,
+      "loss": 0.2368,
+      "step": 17615
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.123036882606992e-06,
+      "loss": 0.2267,
+      "step": 17616
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.12204402880993e-06,
+      "loss": 0.2785,
+      "step": 17617
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.121051183734544e-06,
+      "loss": 0.2519,
+      "step": 17618
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.120058347390696e-06,
+      "loss": 0.2009,
+      "step": 17619
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.119065519788258e-06,
+      "loss": 0.1608,
+      "step": 17620
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.118072700937084e-06,
+      "loss": 0.2693,
+      "step": 17621
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.117079890847043e-06,
+      "loss": 0.2526,
+      "step": 17622
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.116087089527991e-06,
+      "loss": 0.2215,
+      "step": 17623
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.115094296989796e-06,
+      "loss": 0.2252,
+      "step": 17624
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.114101513242317e-06,
+      "loss": 0.1712,
+      "step": 17625
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.11310873829542e-06,
+      "loss": 0.1855,
+      "step": 17626
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.112115972158965e-06,
+      "loss": 0.2474,
+      "step": 17627
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.11112321484281e-06,
+      "loss": 0.2693,
+      "step": 17628
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.110130466356825e-06,
+      "loss": 0.3123,
+      "step": 17629
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.109137726710867e-06,
+      "loss": 0.2152,
+      "step": 17630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.108144995914802e-06,
+      "loss": 0.2394,
+      "step": 17631
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.107152273978487e-06,
+      "loss": 0.2249,
+      "step": 17632
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.106159560911788e-06,
+      "loss": 0.258,
+      "step": 17633
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.105166856724564e-06,
+      "loss": 0.2189,
+      "step": 17634
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.104174161426674e-06,
+      "loss": 0.2318,
+      "step": 17635
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.10318147502799e-06,
+      "loss": 0.2621,
+      "step": 17636
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.102188797538359e-06,
+      "loss": 0.2215,
+      "step": 17637
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.101196128967654e-06,
+      "loss": 0.2221,
+      "step": 17638
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.10020346932573e-06,
+      "loss": 0.2293,
+      "step": 17639
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.09921081862245e-06,
+      "loss": 0.2362,
+      "step": 17640
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.098218176867677e-06,
+      "loss": 0.2084,
+      "step": 17641
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.097225544071268e-06,
+      "loss": 0.2288,
+      "step": 17642
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.096232920243088e-06,
+      "loss": 0.2352,
+      "step": 17643
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.09524030539299e-06,
+      "loss": 0.2483,
+      "step": 17644
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.094247699530845e-06,
+      "loss": 0.2004,
+      "step": 17645
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.093255102666512e-06,
+      "loss": 0.2467,
+      "step": 17646
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.092262514809846e-06,
+      "loss": 0.2363,
+      "step": 17647
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.09126993597071e-06,
+      "loss": 0.2746,
+      "step": 17648
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.090277366158962e-06,
+      "loss": 0.199,
+      "step": 17649
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08928480538447e-06,
+      "loss": 0.2934,
+      "step": 17650
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.088292253657085e-06,
+      "loss": 0.268,
+      "step": 17651
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.087299710986672e-06,
+      "loss": 0.2815,
+      "step": 17652
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.086307177383086e-06,
+      "loss": 0.2237,
+      "step": 17653
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.085314652856193e-06,
+      "loss": 0.214,
+      "step": 17654
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.084322137415855e-06,
+      "loss": 0.2466,
+      "step": 17655
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.083329631071924e-06,
+      "loss": 0.2673,
+      "step": 17656
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08233713383426e-06,
+      "loss": 0.239,
+      "step": 17657
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.081344645712729e-06,
+      "loss": 0.2593,
+      "step": 17658
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.080352166717186e-06,
+      "loss": 0.2128,
+      "step": 17659
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.079359696857495e-06,
+      "loss": 0.2754,
+      "step": 17660
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.078367236143506e-06,
+      "loss": 0.2344,
+      "step": 17661
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.077374784585085e-06,
+      "loss": 0.2318,
+      "step": 17662
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07638234219209e-06,
+      "loss": 0.2566,
+      "step": 17663
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07538990897438e-06,
+      "loss": 0.2207,
+      "step": 17664
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.074397484941816e-06,
+      "loss": 0.2775,
+      "step": 17665
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.073405070104254e-06,
+      "loss": 0.3116,
+      "step": 17666
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07241266447155e-06,
+      "loss": 0.2049,
+      "step": 17667
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.071420268053568e-06,
+      "loss": 0.198,
+      "step": 17668
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.070427880860163e-06,
+      "loss": 0.2823,
+      "step": 17669
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0694355029012e-06,
+      "loss": 0.2655,
+      "step": 17670
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.068443134186526e-06,
+      "loss": 0.2436,
+      "step": 17671
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.06745077472601e-06,
+      "loss": 0.2008,
+      "step": 17672
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.066458424529502e-06,
+      "loss": 0.2447,
+      "step": 17673
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.065466083606865e-06,
+      "loss": 0.2323,
+      "step": 17674
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.064473751967958e-06,
+      "loss": 0.2145,
+      "step": 17675
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.063481429622634e-06,
+      "loss": 0.2403,
+      "step": 17676
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.062489116580755e-06,
+      "loss": 0.2072,
+      "step": 17677
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.061496812852175e-06,
+      "loss": 0.2499,
+      "step": 17678
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.060504518446758e-06,
+      "loss": 0.2331,
+      "step": 17679
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.059512233374353e-06,
+      "loss": 0.2216,
+      "step": 17680
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.058519957644822e-06,
+      "loss": 0.2309,
+      "step": 17681
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.057527691268021e-06,
+      "loss": 0.2705,
+      "step": 17682
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.056535434253812e-06,
+      "loss": 0.2284,
+      "step": 17683
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.055543186612049e-06,
+      "loss": 0.199,
+      "step": 17684
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.054550948352587e-06,
+      "loss": 0.2477,
+      "step": 17685
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.053558719485282e-06,
+      "loss": 0.2366,
+      "step": 17686
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.052566500019992e-06,
+      "loss": 0.2211,
+      "step": 17687
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.05157428996658e-06,
+      "loss": 0.2537,
+      "step": 17688
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.050582089334896e-06,
+      "loss": 0.2015,
+      "step": 17689
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0495898981348e-06,
+      "loss": 0.2651,
+      "step": 17690
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.048597716376143e-06,
+      "loss": 0.2497,
+      "step": 17691
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.04760554406879e-06,
+      "loss": 0.2537,
+      "step": 17692
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.046613381222587e-06,
+      "loss": 0.2328,
+      "step": 17693
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.045621227847403e-06,
+      "loss": 0.2164,
+      "step": 17694
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.044629083953082e-06,
+      "loss": 0.2789,
+      "step": 17695
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.043636949549485e-06,
+      "loss": 0.1997,
+      "step": 17696
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.042644824646469e-06,
+      "loss": 0.1597,
+      "step": 17697
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.04165270925389e-06,
+      "loss": 0.2513,
+      "step": 17698
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.040660603381601e-06,
+      "loss": 0.2289,
+      "step": 17699
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03966850703946e-06,
+      "loss": 0.2008,
+      "step": 17700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03867642023732e-06,
+      "loss": 0.2953,
+      "step": 17701
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.037684342985038e-06,
+      "loss": 0.211,
+      "step": 17702
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03669227529247e-06,
+      "loss": 0.2261,
+      "step": 17703
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.035700217169477e-06,
+      "loss": 0.2292,
+      "step": 17704
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0347081686259e-06,
+      "loss": 0.1983,
+      "step": 17705
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.033716129671605e-06,
+      "loss": 0.1836,
+      "step": 17706
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.032724100316443e-06,
+      "loss": 0.229,
+      "step": 17707
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.031732080570273e-06,
+      "loss": 0.2413,
+      "step": 17708
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.030740070442945e-06,
+      "loss": 0.2502,
+      "step": 17709
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.029748069944314e-06,
+      "loss": 0.1838,
+      "step": 17710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.028756079084236e-06,
+      "loss": 0.2614,
+      "step": 17711
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.027764097872565e-06,
+      "loss": 0.2546,
+      "step": 17712
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.026772126319163e-06,
+      "loss": 0.2619,
+      "step": 17713
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.025780164433869e-06,
+      "loss": 0.2156,
+      "step": 17714
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.024788212226549e-06,
+      "loss": 0.2251,
+      "step": 17715
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.023796269707053e-06,
+      "loss": 0.2321,
+      "step": 17716
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.022804336885236e-06,
+      "loss": 0.2234,
+      "step": 17717
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.021812413770953e-06,
+      "loss": 0.1938,
+      "step": 17718
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.020820500374054e-06,
+      "loss": 0.2425,
+      "step": 17719
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.019828596704394e-06,
+      "loss": 0.2206,
+      "step": 17720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.018836702771831e-06,
+      "loss": 0.2227,
+      "step": 17721
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.017844818586214e-06,
+      "loss": 0.2587,
+      "step": 17722
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.0168529441574e-06,
+      "loss": 0.1934,
+      "step": 17723
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.015861079495238e-06,
+      "loss": 0.2997,
+      "step": 17724
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.014869224609584e-06,
+      "loss": 0.2638,
+      "step": 17725
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.01387737951029e-06,
+      "loss": 0.2635,
+      "step": 17726
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.012885544207207e-06,
+      "loss": 0.1578,
+      "step": 17727
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.011893718710197e-06,
+      "loss": 0.2122,
+      "step": 17728
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.010901903029103e-06,
+      "loss": 0.2891,
+      "step": 17729
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00991009717378e-06,
+      "loss": 0.1714,
+      "step": 17730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.008918301154083e-06,
+      "loss": 0.2019,
+      "step": 17731
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.007926514979864e-06,
+      "loss": 0.21,
+      "step": 17732
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.006934738660977e-06,
+      "loss": 0.2682,
+      "step": 17733
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.005942972207269e-06,
+      "loss": 0.2213,
+      "step": 17734
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.004951215628597e-06,
+      "loss": 0.1987,
+      "step": 17735
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.003959468934809e-06,
+      "loss": 0.2238,
+      "step": 17736
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.002967732135767e-06,
+      "loss": 0.2259,
+      "step": 17737
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00197600524131e-06,
+      "loss": 0.2418,
+      "step": 17738
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.000984288261296e-06,
+      "loss": 0.2128,
+      "step": 17739
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.999992581205577e-06,
+      "loss": 0.2206,
+      "step": 17740
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.999000884084004e-06,
+      "loss": 0.1894,
+      "step": 17741
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.998009196906433e-06,
+      "loss": 0.2162,
+      "step": 17742
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.997017519682708e-06,
+      "loss": 0.1787,
+      "step": 17743
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.996025852422682e-06,
+      "loss": 0.2319,
+      "step": 17744
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.995034195136209e-06,
+      "loss": 0.2376,
+      "step": 17745
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99404254783314e-06,
+      "loss": 0.1955,
+      "step": 17746
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.993050910523329e-06,
+      "loss": 0.2666,
+      "step": 17747
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.992059283216617e-06,
+      "loss": 0.2278,
+      "step": 17748
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.991067665922865e-06,
+      "loss": 0.2738,
+      "step": 17749
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.990076058651917e-06,
+      "loss": 0.2864,
+      "step": 17750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98908446141363e-06,
+      "loss": 0.2105,
+      "step": 17751
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.988092874217852e-06,
+      "loss": 0.2342,
+      "step": 17752
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98710129707443e-06,
+      "loss": 0.24,
+      "step": 17753
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.986109729993216e-06,
+      "loss": 0.2616,
+      "step": 17754
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.985118172984063e-06,
+      "loss": 0.1731,
+      "step": 17755
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.984126626056819e-06,
+      "loss": 0.2056,
+      "step": 17756
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.98313508922134e-06,
+      "loss": 0.207,
+      "step": 17757
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.982143562487466e-06,
+      "loss": 0.2482,
+      "step": 17758
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.981152045865052e-06,
+      "loss": 0.2523,
+      "step": 17759
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.980160539363946e-06,
+      "loss": 0.2495,
+      "step": 17760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.979169042994e-06,
+      "loss": 0.2356,
+      "step": 17761
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.978177556765068e-06,
+      "loss": 0.2532,
+      "step": 17762
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.977186080686987e-06,
+      "loss": 0.2532,
+      "step": 17763
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.976194614769617e-06,
+      "loss": 0.2399,
+      "step": 17764
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.975203159022804e-06,
+      "loss": 0.2417,
+      "step": 17765
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.974211713456398e-06,
+      "loss": 0.2592,
+      "step": 17766
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.973220278080246e-06,
+      "loss": 0.2437,
+      "step": 17767
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.972228852904199e-06,
+      "loss": 0.2178,
+      "step": 17768
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.971237437938105e-06,
+      "loss": 0.244,
+      "step": 17769
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.97024603319181e-06,
+      "loss": 0.2541,
+      "step": 17770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.969254638675171e-06,
+      "loss": 0.255,
+      "step": 17771
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.968263254398027e-06,
+      "loss": 0.1701,
+      "step": 17772
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.967271880370231e-06,
+      "loss": 0.1726,
+      "step": 17773
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.966280516601633e-06,
+      "loss": 0.2557,
+      "step": 17774
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.96528916310208e-06,
+      "loss": 0.2502,
+      "step": 17775
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.96429781988142e-06,
+      "loss": 0.2366,
+      "step": 17776
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.963306486949496e-06,
+      "loss": 0.2249,
+      "step": 17777
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.962315164316164e-06,
+      "loss": 0.2228,
+      "step": 17778
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.961323851991264e-06,
+      "loss": 0.1493,
+      "step": 17779
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.960332549984653e-06,
+      "loss": 0.2283,
+      "step": 17780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.959341258306175e-06,
+      "loss": 0.2197,
+      "step": 17781
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.958349976965673e-06,
+      "loss": 0.2352,
+      "step": 17782
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.957358705973e-06,
+      "loss": 0.2515,
+      "step": 17783
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.956367445338e-06,
+      "loss": 0.2039,
+      "step": 17784
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.95537619507052e-06,
+      "loss": 0.2488,
+      "step": 17785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.954384955180415e-06,
+      "loss": 0.2323,
+      "step": 17786
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.95339372567752e-06,
+      "loss": 0.1941,
+      "step": 17787
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.952402506571686e-06,
+      "loss": 0.2295,
+      "step": 17788
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.951411297872763e-06,
+      "loss": 0.2325,
+      "step": 17789
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.950420099590598e-06,
+      "loss": 0.2071,
+      "step": 17790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.949428911735039e-06,
+      "loss": 0.1911,
+      "step": 17791
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.948437734315925e-06,
+      "loss": 0.298,
+      "step": 17792
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.947446567343108e-06,
+      "loss": 0.1844,
+      "step": 17793
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.946455410826433e-06,
+      "loss": 0.201,
+      "step": 17794
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.945464264775748e-06,
+      "loss": 0.1716,
+      "step": 17795
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.944473129200895e-06,
+      "loss": 0.231,
+      "step": 17796
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.94348200411172e-06,
+      "loss": 0.3077,
+      "step": 17797
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.942490889518076e-06,
+      "loss": 0.2016,
+      "step": 17798
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.941499785429803e-06,
+      "loss": 0.2081,
+      "step": 17799
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.940508691856749e-06,
+      "loss": 0.2408,
+      "step": 17800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.939517608808757e-06,
+      "loss": 0.2078,
+      "step": 17801
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.938526536295674e-06,
+      "loss": 0.2413,
+      "step": 17802
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.937535474327345e-06,
+      "loss": 0.2546,
+      "step": 17803
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.936544422913615e-06,
+      "loss": 0.2128,
+      "step": 17804
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.935553382064336e-06,
+      "loss": 0.1967,
+      "step": 17805
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.93456235178934e-06,
+      "loss": 0.184,
+      "step": 17806
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.93357133209848e-06,
+      "loss": 0.2452,
+      "step": 17807
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.932580323001601e-06,
+      "loss": 0.1694,
+      "step": 17808
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.931589324508547e-06,
+      "loss": 0.1971,
+      "step": 17809
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.930598336629164e-06,
+      "loss": 0.2014,
+      "step": 17810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.929607359373292e-06,
+      "loss": 0.2561,
+      "step": 17811
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.928616392750777e-06,
+      "loss": 0.3134,
+      "step": 17812
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.927625436771464e-06,
+      "loss": 0.1842,
+      "step": 17813
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.926634491445199e-06,
+      "loss": 0.2569,
+      "step": 17814
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.925643556781828e-06,
+      "loss": 0.1852,
+      "step": 17815
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92465263279119e-06,
+      "loss": 0.2376,
+      "step": 17816
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.923661719483128e-06,
+      "loss": 0.2266,
+      "step": 17817
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.922670816867491e-06,
+      "loss": 0.2296,
+      "step": 17818
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.921679924954123e-06,
+      "loss": 0.2162,
+      "step": 17819
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92068904375286e-06,
+      "loss": 0.2185,
+      "step": 17820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.919698173273554e-06,
+      "loss": 0.2217,
+      "step": 17821
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.91870731352604e-06,
+      "loss": 0.248,
+      "step": 17822
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.917716464520168e-06,
+      "loss": 0.2132,
+      "step": 17823
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.916725626265782e-06,
+      "loss": 0.2658,
+      "step": 17824
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.915734798772721e-06,
+      "loss": 0.2482,
+      "step": 17825
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.914743982050828e-06,
+      "loss": 0.2617,
+      "step": 17826
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.913753176109948e-06,
+      "loss": 0.2477,
+      "step": 17827
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.91276238095992e-06,
+      "loss": 0.2667,
+      "step": 17828
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.911771596610595e-06,
+      "loss": 0.2562,
+      "step": 17829
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.910780823071805e-06,
+      "loss": 0.2342,
+      "step": 17830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.909790060353396e-06,
+      "loss": 0.2012,
+      "step": 17831
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.908799308465214e-06,
+      "loss": 0.2398,
+      "step": 17832
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.9078085674171e-06,
+      "loss": 0.2241,
+      "step": 17833
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.906817837218896e-06,
+      "loss": 0.2462,
+      "step": 17834
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.905827117880441e-06,
+      "loss": 0.2212,
+      "step": 17835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.90483640941158e-06,
+      "loss": 0.2277,
+      "step": 17836
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.903845711822151e-06,
+      "loss": 0.2973,
+      "step": 17837
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.902855025121999e-06,
+      "loss": 0.2635,
+      "step": 17838
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.901864349320968e-06,
+      "loss": 0.2004,
+      "step": 17839
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.90087368442889e-06,
+      "loss": 0.1862,
+      "step": 17840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.899883030455618e-06,
+      "loss": 0.2048,
+      "step": 17841
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.898892387410986e-06,
+      "loss": 0.2573,
+      "step": 17842
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.897901755304837e-06,
+      "loss": 0.2077,
+      "step": 17843
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.896911134147013e-06,
+      "loss": 0.1446,
+      "step": 17844
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.895920523947353e-06,
+      "loss": 0.3019,
+      "step": 17845
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.894929924715699e-06,
+      "loss": 0.2565,
+      "step": 17846
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.893939336461888e-06,
+      "loss": 0.1928,
+      "step": 17847
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.89294875919577e-06,
+      "loss": 0.2436,
+      "step": 17848
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.891958192927172e-06,
+      "loss": 0.1911,
+      "step": 17849
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.890967637665945e-06,
+      "loss": 0.283,
+      "step": 17850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.889977093421927e-06,
+      "loss": 0.2909,
+      "step": 17851
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.888986560204958e-06,
+      "loss": 0.2465,
+      "step": 17852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.887996038024877e-06,
+      "loss": 0.2306,
+      "step": 17853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.887005526891522e-06,
+      "loss": 0.2321,
+      "step": 17854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.886015026814736e-06,
+      "loss": 0.2222,
+      "step": 17855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.885024537804354e-06,
+      "loss": 0.2285,
+      "step": 17856
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.884034059870223e-06,
+      "loss": 0.2701,
+      "step": 17857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.88304359302218e-06,
+      "loss": 0.2083,
+      "step": 17858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.88205313727006e-06,
+      "loss": 0.1981,
+      "step": 17859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.881062692623706e-06,
+      "loss": 0.1812,
+      "step": 17860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.880072259092957e-06,
+      "loss": 0.2542,
+      "step": 17861
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.879081836687647e-06,
+      "loss": 0.2704,
+      "step": 17862
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.878091425417628e-06,
+      "loss": 0.2453,
+      "step": 17863
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.877101025292725e-06,
+      "loss": 0.2151,
+      "step": 17864
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87611063632278e-06,
+      "loss": 0.2189,
+      "step": 17865
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.875120258517636e-06,
+      "loss": 0.1863,
+      "step": 17866
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87412989188713e-06,
+      "loss": 0.2401,
+      "step": 17867
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.873139536441098e-06,
+      "loss": 0.2465,
+      "step": 17868
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87214919218938e-06,
+      "loss": 0.2456,
+      "step": 17869
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.871158859141813e-06,
+      "loss": 0.2012,
+      "step": 17870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.870168537308234e-06,
+      "loss": 0.2368,
+      "step": 17871
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.869178226698485e-06,
+      "loss": 0.1804,
+      "step": 17872
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.868187927322406e-06,
+      "loss": 0.1762,
+      "step": 17873
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.867197639189823e-06,
+      "loss": 0.2678,
+      "step": 17874
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.866207362310583e-06,
+      "loss": 0.2638,
+      "step": 17875
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.865217096694521e-06,
+      "loss": 0.2486,
+      "step": 17876
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.86422684235148e-06,
+      "loss": 0.3102,
+      "step": 17877
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.863236599291286e-06,
+      "loss": 0.2146,
+      "step": 17878
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.862246367523783e-06,
+      "loss": 0.2206,
+      "step": 17879
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.861256147058805e-06,
+      "loss": 0.3067,
+      "step": 17880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.860265937906194e-06,
+      "loss": 0.2302,
+      "step": 17881
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.859275740075788e-06,
+      "loss": 0.203,
+      "step": 17882
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.858285553577412e-06,
+      "loss": 0.2217,
+      "step": 17883
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.857295378420914e-06,
+      "loss": 0.2419,
+      "step": 17884
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.856305214616125e-06,
+      "loss": 0.2532,
+      "step": 17885
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.855315062172885e-06,
+      "loss": 0.2496,
+      "step": 17886
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.854324921101028e-06,
+      "loss": 0.2137,
+      "step": 17887
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.853334791410391e-06,
+      "loss": 0.261,
+      "step": 17888
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.852344673110807e-06,
+      "loss": 0.2105,
+      "step": 17889
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.851354566212113e-06,
+      "loss": 0.2368,
+      "step": 17890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.850364470724148e-06,
+      "loss": 0.2478,
+      "step": 17891
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.849374386656748e-06,
+      "loss": 0.2912,
+      "step": 17892
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.848384314019743e-06,
+      "loss": 0.2067,
+      "step": 17893
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.847394252822975e-06,
+      "loss": 0.267,
+      "step": 17894
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.846404203076273e-06,
+      "loss": 0.2126,
+      "step": 17895
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.845414164789474e-06,
+      "loss": 0.1935,
+      "step": 17896
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.844424137972421e-06,
+      "loss": 0.2441,
+      "step": 17897
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.843434122634938e-06,
+      "loss": 0.2165,
+      "step": 17898
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.842444118786863e-06,
+      "loss": 0.2695,
+      "step": 17899
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.841454126438034e-06,
+      "loss": 0.3014,
+      "step": 17900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.840464145598283e-06,
+      "loss": 0.2581,
+      "step": 17901
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.839474176277447e-06,
+      "loss": 0.2747,
+      "step": 17902
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.838484218485359e-06,
+      "loss": 0.2432,
+      "step": 17903
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.83749427223185e-06,
+      "loss": 0.2353,
+      "step": 17904
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.836504337526758e-06,
+      "loss": 0.2905,
+      "step": 17905
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.835514414379921e-06,
+      "loss": 0.2608,
+      "step": 17906
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.834524502801164e-06,
+      "loss": 0.2213,
+      "step": 17907
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.833534602800323e-06,
+      "loss": 0.2274,
+      "step": 17908
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.832544714387236e-06,
+      "loss": 0.2059,
+      "step": 17909
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.831554837571736e-06,
+      "loss": 0.2044,
+      "step": 17910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.830564972363655e-06,
+      "loss": 0.208,
+      "step": 17911
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.829575118772826e-06,
+      "loss": 0.2478,
+      "step": 17912
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.828585276809082e-06,
+      "loss": 0.2097,
+      "step": 17913
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.827595446482255e-06,
+      "loss": 0.237,
+      "step": 17914
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.826605627802182e-06,
+      "loss": 0.2348,
+      "step": 17915
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.825615820778699e-06,
+      "loss": 0.2647,
+      "step": 17916
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.824626025421625e-06,
+      "loss": 0.2797,
+      "step": 17917
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.823636241740804e-06,
+      "loss": 0.3012,
+      "step": 17918
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.822646469746067e-06,
+      "loss": 0.2342,
+      "step": 17919
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.821656709447245e-06,
+      "loss": 0.2526,
+      "step": 17920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.820666960854172e-06,
+      "loss": 0.1998,
+      "step": 17921
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.819677223976679e-06,
+      "loss": 0.2271,
+      "step": 17922
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.818687498824594e-06,
+      "loss": 0.2309,
+      "step": 17923
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.817697785407755e-06,
+      "loss": 0.2944,
+      "step": 17924
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.816708083735993e-06,
+      "loss": 0.2501,
+      "step": 17925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.81571839381914e-06,
+      "loss": 0.2783,
+      "step": 17926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.814728715667026e-06,
+      "loss": 0.2086,
+      "step": 17927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.81373904928948e-06,
+      "loss": 0.241,
+      "step": 17928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.812749394696339e-06,
+      "loss": 0.2544,
+      "step": 17929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.811759751897429e-06,
+      "loss": 0.2293,
+      "step": 17930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.81077012090259e-06,
+      "loss": 0.2987,
+      "step": 17931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80978050172164e-06,
+      "loss": 0.2228,
+      "step": 17932
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80879089436442e-06,
+      "loss": 0.2333,
+      "step": 17933
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.807801298840757e-06,
+      "loss": 0.2288,
+      "step": 17934
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.806811715160485e-06,
+      "loss": 0.2232,
+      "step": 17935
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80582214333343e-06,
+      "loss": 0.2866,
+      "step": 17936
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.804832583369423e-06,
+      "loss": 0.1823,
+      "step": 17937
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.803843035278297e-06,
+      "loss": 0.2039,
+      "step": 17938
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80285349906988e-06,
+      "loss": 0.3011,
+      "step": 17939
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.801863974754006e-06,
+      "loss": 0.2212,
+      "step": 17940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.800874462340498e-06,
+      "loss": 0.2688,
+      "step": 17941
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.799884961839193e-06,
+      "loss": 0.2406,
+      "step": 17942
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.798895473259918e-06,
+      "loss": 0.2527,
+      "step": 17943
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.7979059966125e-06,
+      "loss": 0.2528,
+      "step": 17944
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.796916531906776e-06,
+      "loss": 0.2503,
+      "step": 17945
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.795927079152566e-06,
+      "loss": 0.2267,
+      "step": 17946
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.794937638359707e-06,
+      "loss": 0.1698,
+      "step": 17947
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.79394820953802e-06,
+      "loss": 0.1986,
+      "step": 17948
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.792958792697341e-06,
+      "loss": 0.1922,
+      "step": 17949
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.791969387847503e-06,
+      "loss": 0.1978,
+      "step": 17950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.790979994998321e-06,
+      "loss": 0.1713,
+      "step": 17951
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.789990614159634e-06,
+      "loss": 0.1946,
+      "step": 17952
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.789001245341268e-06,
+      "loss": 0.3145,
+      "step": 17953
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78801188855305e-06,
+      "loss": 0.262,
+      "step": 17954
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.787022543804814e-06,
+      "loss": 0.2414,
+      "step": 17955
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.786033211106381e-06,
+      "loss": 0.2324,
+      "step": 17956
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.785043890467581e-06,
+      "loss": 0.2933,
+      "step": 17957
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.784054581898245e-06,
+      "loss": 0.213,
+      "step": 17958
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.783065285408197e-06,
+      "loss": 0.1571,
+      "step": 17959
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78207600100727e-06,
+      "loss": 0.3006,
+      "step": 17960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.781086728705289e-06,
+      "loss": 0.2521,
+      "step": 17961
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.780097468512078e-06,
+      "loss": 0.213,
+      "step": 17962
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.779108220437467e-06,
+      "loss": 0.2293,
+      "step": 17963
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.778118984491288e-06,
+      "loss": 0.2698,
+      "step": 17964
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.77712976068336e-06,
+      "loss": 0.2392,
+      "step": 17965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.77614054902351e-06,
+      "loss": 0.2094,
+      "step": 17966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.775151349521574e-06,
+      "loss": 0.213,
+      "step": 17967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.77416216218737e-06,
+      "loss": 0.2436,
+      "step": 17968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.773172987030732e-06,
+      "loss": 0.2266,
+      "step": 17969
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.77218382406148e-06,
+      "loss": 0.2648,
+      "step": 17970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.771194673289444e-06,
+      "loss": 0.2305,
+      "step": 17971
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.770205534724449e-06,
+      "loss": 0.2588,
+      "step": 17972
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.769216408376318e-06,
+      "loss": 0.3007,
+      "step": 17973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.768227294254888e-06,
+      "loss": 0.2042,
+      "step": 17974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.76723819236997e-06,
+      "loss": 0.1321,
+      "step": 17975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.7662491027314e-06,
+      "loss": 0.2081,
+      "step": 17976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.765260025349e-06,
+      "loss": 0.2429,
+      "step": 17977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.764270960232598e-06,
+      "loss": 0.2567,
+      "step": 17978
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.763281907392021e-06,
+      "loss": 0.2904,
+      "step": 17979
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.762292866837086e-06,
+      "loss": 0.2267,
+      "step": 17980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.761303838577626e-06,
+      "loss": 0.2915,
+      "step": 17981
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.76031482262346e-06,
+      "loss": 0.2283,
+      "step": 17982
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.75932581898442e-06,
+      "loss": 0.1988,
+      "step": 17983
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.758336827670329e-06,
+      "loss": 0.26,
+      "step": 17984
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.757347848691006e-06,
+      "loss": 0.2983,
+      "step": 17985
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756358882056282e-06,
+      "loss": 0.2239,
+      "step": 17986
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.755369927775978e-06,
+      "loss": 0.1972,
+      "step": 17987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.754380985859922e-06,
+      "loss": 0.2724,
+      "step": 17988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.753392056317933e-06,
+      "loss": 0.2559,
+      "step": 17989
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.752403139159837e-06,
+      "loss": 0.2483,
+      "step": 17990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.751414234395457e-06,
+      "loss": 0.2154,
+      "step": 17991
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.750425342034623e-06,
+      "loss": 0.2452,
+      "step": 17992
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.749436462087154e-06,
+      "loss": 0.2419,
+      "step": 17993
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.748447594562872e-06,
+      "loss": 0.2124,
+      "step": 17994
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.747458739471603e-06,
+      "loss": 0.1664,
+      "step": 17995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.74646989682317e-06,
+      "loss": 0.2841,
+      "step": 17996
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.745481066627396e-06,
+      "loss": 0.2016,
+      "step": 17997
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.744492248894107e-06,
+      "loss": 0.2186,
+      "step": 17998
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.743503443633118e-06,
+      "loss": 0.2545,
+      "step": 17999
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.742514650854258e-06,
+      "loss": 0.2583,
+      "step": 18000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.741525870567349e-06,
+      "loss": 0.2103,
+      "step": 18001
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.740537102782213e-06,
+      "loss": 0.2468,
+      "step": 18002
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.739548347508678e-06,
+      "loss": 0.2706,
+      "step": 18003
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.738559604756556e-06,
+      "loss": 0.2295,
+      "step": 18004
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.737570874535674e-06,
+      "loss": 0.22,
+      "step": 18005
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.736582156855858e-06,
+      "loss": 0.2344,
+      "step": 18006
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73559345172692e-06,
+      "loss": 0.2886,
+      "step": 18007
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.734604759158699e-06,
+      "loss": 0.1926,
+      "step": 18008
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.733616079160997e-06,
+      "loss": 0.1872,
+      "step": 18009
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.732627411743646e-06,
+      "loss": 0.2191,
+      "step": 18010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.731638756916468e-06,
+      "loss": 0.1868,
+      "step": 18011
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.730650114689283e-06,
+      "loss": 0.3553,
+      "step": 18012
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.729661485071914e-06,
+      "loss": 0.2578,
+      "step": 18013
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.728672868074176e-06,
+      "loss": 0.2464,
+      "step": 18014
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.727684263705896e-06,
+      "loss": 0.2521,
+      "step": 18015
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.726695671976889e-06,
+      "loss": 0.2883,
+      "step": 18016
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.725707092896986e-06,
+      "loss": 0.2275,
+      "step": 18017
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.724718526475997e-06,
+      "loss": 0.2107,
+      "step": 18018
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.723729972723746e-06,
+      "loss": 0.2429,
+      "step": 18019
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.722741431650058e-06,
+      "loss": 0.2014,
+      "step": 18020
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.721752903264744e-06,
+      "loss": 0.1873,
+      "step": 18021
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.720764387577634e-06,
+      "loss": 0.224,
+      "step": 18022
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.719775884598542e-06,
+      "loss": 0.2405,
+      "step": 18023
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71878739433729e-06,
+      "loss": 0.2945,
+      "step": 18024
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.717798916803692e-06,
+      "loss": 0.2326,
+      "step": 18025
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.716810452007577e-06,
+      "loss": 0.241,
+      "step": 18026
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.715821999958762e-06,
+      "loss": 0.2362,
+      "step": 18027
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.714833560667063e-06,
+      "loss": 0.2454,
+      "step": 18028
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.713845134142298e-06,
+      "loss": 0.2706,
+      "step": 18029
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71285672039429e-06,
+      "loss": 0.1784,
+      "step": 18030
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.711868319432855e-06,
+      "loss": 0.247,
+      "step": 18031
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71087993126782e-06,
+      "loss": 0.2607,
+      "step": 18032
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.709891555908992e-06,
+      "loss": 0.2188,
+      "step": 18033
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.708903193366194e-06,
+      "loss": 0.209,
+      "step": 18034
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.707914843649247e-06,
+      "loss": 0.1997,
+      "step": 18035
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.706926506767968e-06,
+      "loss": 0.24,
+      "step": 18036
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.705938182732176e-06,
+      "loss": 0.2681,
+      "step": 18037
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.704949871551687e-06,
+      "loss": 0.197,
+      "step": 18038
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.70396157323632e-06,
+      "loss": 0.2909,
+      "step": 18039
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.702973287795889e-06,
+      "loss": 0.235,
+      "step": 18040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.70198501524022e-06,
+      "loss": 0.2939,
+      "step": 18041
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.700996755579128e-06,
+      "loss": 0.2725,
+      "step": 18042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.700008508822422e-06,
+      "loss": 0.3014,
+      "step": 18043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.69902027497993e-06,
+      "loss": 0.2587,
+      "step": 18044
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.698032054061464e-06,
+      "loss": 0.2083,
+      "step": 18045
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.697043846076844e-06,
+      "loss": 0.2355,
+      "step": 18046
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.696055651035883e-06,
+      "loss": 0.2448,
+      "step": 18047
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.6950674689484e-06,
+      "loss": 0.2253,
+      "step": 18048
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.694079299824211e-06,
+      "loss": 0.2412,
+      "step": 18049
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.69309114367313e-06,
+      "loss": 0.1934,
+      "step": 18050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.692103000504985e-06,
+      "loss": 0.198,
+      "step": 18051
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.691114870329576e-06,
+      "loss": 0.3047,
+      "step": 18052
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.690126753156729e-06,
+      "loss": 0.2302,
+      "step": 18053
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.689138648996258e-06,
+      "loss": 0.2774,
+      "step": 18054
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.688150557857979e-06,
+      "loss": 0.2521,
+      "step": 18055
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.687162479751709e-06,
+      "loss": 0.2027,
+      "step": 18056
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.68617441468726e-06,
+      "loss": 0.2501,
+      "step": 18057
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.68518636267445e-06,
+      "loss": 0.1661,
+      "step": 18058
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.684198323723092e-06,
+      "loss": 0.194,
+      "step": 18059
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.683210297843006e-06,
+      "loss": 0.2968,
+      "step": 18060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.682222285044007e-06,
+      "loss": 0.2433,
+      "step": 18061
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.681234285335903e-06,
+      "loss": 0.213,
+      "step": 18062
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.680246298728515e-06,
+      "loss": 0.1887,
+      "step": 18063
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.679258325231655e-06,
+      "loss": 0.2708,
+      "step": 18064
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.678270364855137e-06,
+      "loss": 0.1871,
+      "step": 18065
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.677282417608785e-06,
+      "loss": 0.2389,
+      "step": 18066
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.676294483502399e-06,
+      "loss": 0.2331,
+      "step": 18067
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.675306562545798e-06,
+      "loss": 0.289,
+      "step": 18068
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.6743186547488e-06,
+      "loss": 0.1853,
+      "step": 18069
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.673330760121217e-06,
+      "loss": 0.2099,
+      "step": 18070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.672342878672863e-06,
+      "loss": 0.2148,
+      "step": 18071
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.671355010413552e-06,
+      "loss": 0.1932,
+      "step": 18072
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.670367155353094e-06,
+      "loss": 0.2892,
+      "step": 18073
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.669379313501305e-06,
+      "loss": 0.2396,
+      "step": 18074
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.668391484868004e-06,
+      "loss": 0.2518,
+      "step": 18075
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.667403669462994e-06,
+      "loss": 0.226,
+      "step": 18076
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.666415867296091e-06,
+      "loss": 0.2059,
+      "step": 18077
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.665428078377112e-06,
+      "loss": 0.2238,
+      "step": 18078
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.664440302715866e-06,
+      "loss": 0.262,
+      "step": 18079
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.663452540322171e-06,
+      "loss": 0.2286,
+      "step": 18080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.662464791205831e-06,
+      "loss": 0.3128,
+      "step": 18081
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.661477055376663e-06,
+      "loss": 0.2382,
+      "step": 18082
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.660489332844478e-06,
+      "loss": 0.2114,
+      "step": 18083
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.659501623619092e-06,
+      "loss": 0.2352,
+      "step": 18084
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.658513927710317e-06,
+      "loss": 0.1854,
+      "step": 18085
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.657526245127956e-06,
+      "loss": 0.243,
+      "step": 18086
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.656538575881829e-06,
+      "loss": 0.2638,
+      "step": 18087
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.655550919981746e-06,
+      "loss": 0.2588,
+      "step": 18088
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.654563277437516e-06,
+      "loss": 0.2228,
+      "step": 18089
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.653575648258954e-06,
+      "loss": 0.2735,
+      "step": 18090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.65258803245587e-06,
+      "loss": 0.2692,
+      "step": 18091
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.65160043003807e-06,
+      "loss": 0.2548,
+      "step": 18092
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.65061284101537e-06,
+      "loss": 0.2206,
+      "step": 18093
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.649625265397582e-06,
+      "loss": 0.2548,
+      "step": 18094
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.648637703194515e-06,
+      "loss": 0.2885,
+      "step": 18095
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.647650154415978e-06,
+      "loss": 0.2614,
+      "step": 18096
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.646662619071784e-06,
+      "loss": 0.2726,
+      "step": 18097
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.64567509717174e-06,
+      "loss": 0.2542,
+      "step": 18098
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.644687588725656e-06,
+      "loss": 0.222,
+      "step": 18099
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.64370009374335e-06,
+      "loss": 0.1991,
+      "step": 18100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.642712612234619e-06,
+      "loss": 0.2271,
+      "step": 18101
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.641725144209283e-06,
+      "loss": 0.2433,
+      "step": 18102
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.640737689677146e-06,
+      "loss": 0.2074,
+      "step": 18103
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.639750248648022e-06,
+      "loss": 0.2204,
+      "step": 18104
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.638762821131717e-06,
+      "loss": 0.262,
+      "step": 18105
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.63777540713804e-06,
+      "loss": 0.3012,
+      "step": 18106
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.6367880066768e-06,
+      "loss": 0.206,
+      "step": 18107
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.635800619757807e-06,
+      "loss": 0.2776,
+      "step": 18108
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.634813246390874e-06,
+      "loss": 0.2152,
+      "step": 18109
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.633825886585802e-06,
+      "loss": 0.2786,
+      "step": 18110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.632838540352399e-06,
+      "loss": 0.2525,
+      "step": 18111
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.631851207700482e-06,
+      "loss": 0.2154,
+      "step": 18112
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.63086388863985e-06,
+      "loss": 0.1402,
+      "step": 18113
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.629876583180322e-06,
+      "loss": 0.1977,
+      "step": 18114
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.628889291331696e-06,
+      "loss": 0.2005,
+      "step": 18115
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.627902013103782e-06,
+      "loss": 0.2005,
+      "step": 18116
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.626914748506389e-06,
+      "loss": 0.2346,
+      "step": 18117
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.625927497549325e-06,
+      "loss": 0.2734,
+      "step": 18118
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.624940260242401e-06,
+      "loss": 0.1758,
+      "step": 18119
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.623953036595415e-06,
+      "loss": 0.2058,
+      "step": 18120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.622965826618182e-06,
+      "loss": 0.2673,
+      "step": 18121
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.621978630320504e-06,
+      "loss": 0.2411,
+      "step": 18122
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620991447712191e-06,
+      "loss": 0.1994,
+      "step": 18123
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620004278803053e-06,
+      "loss": 0.1893,
+      "step": 18124
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.619017123602888e-06,
+      "loss": 0.2461,
+      "step": 18125
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.618029982121508e-06,
+      "loss": 0.2328,
+      "step": 18126
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.617042854368717e-06,
+      "loss": 0.2274,
+      "step": 18127
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.616055740354325e-06,
+      "loss": 0.208,
+      "step": 18128
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.615068640088137e-06,
+      "loss": 0.2308,
+      "step": 18129
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.614081553579956e-06,
+      "loss": 0.273,
+      "step": 18130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.613094480839588e-06,
+      "loss": 0.2566,
+      "step": 18131
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.612107421876842e-06,
+      "loss": 0.1853,
+      "step": 18132
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.611120376701523e-06,
+      "loss": 0.2129,
+      "step": 18133
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.610133345323433e-06,
+      "loss": 0.2525,
+      "step": 18134
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.609146327752376e-06,
+      "loss": 0.291,
+      "step": 18135
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.608159323998162e-06,
+      "loss": 0.2503,
+      "step": 18136
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.607172334070595e-06,
+      "loss": 0.2169,
+      "step": 18137
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.606185357979483e-06,
+      "loss": 0.2378,
+      "step": 18138
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.605198395734621e-06,
+      "loss": 0.2196,
+      "step": 18139
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.604211447345823e-06,
+      "loss": 0.1961,
+      "step": 18140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.603224512822887e-06,
+      "loss": 0.2418,
+      "step": 18141
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.602237592175619e-06,
+      "loss": 0.187,
+      "step": 18142
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.601250685413833e-06,
+      "loss": 0.231,
+      "step": 18143
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.600263792547314e-06,
+      "loss": 0.2087,
+      "step": 18144
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.59927691358588e-06,
+      "loss": 0.2359,
+      "step": 18145
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.598290048539331e-06,
+      "loss": 0.1877,
+      "step": 18146
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.597303197417469e-06,
+      "loss": 0.2286,
+      "step": 18147
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.596316360230102e-06,
+      "loss": 0.2937,
+      "step": 18148
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.595329536987026e-06,
+      "loss": 0.2817,
+      "step": 18149
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.594342727698051e-06,
+      "loss": 0.2181,
+      "step": 18150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.593355932372974e-06,
+      "loss": 0.2101,
+      "step": 18151
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.592369151021603e-06,
+      "loss": 0.3209,
+      "step": 18152
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.591382383653742e-06,
+      "loss": 0.2106,
+      "step": 18153
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.59039563027919e-06,
+      "loss": 0.2835,
+      "step": 18154
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.589408890907746e-06,
+      "loss": 0.1942,
+      "step": 18155
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58842216554922e-06,
+      "loss": 0.2149,
+      "step": 18156
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.587435454213411e-06,
+      "loss": 0.2433,
+      "step": 18157
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.586448756910118e-06,
+      "loss": 0.2763,
+      "step": 18158
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.585462073649148e-06,
+      "loss": 0.2354,
+      "step": 18159
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.584475404440296e-06,
+      "loss": 0.2265,
+      "step": 18160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58348874929337e-06,
+      "loss": 0.2378,
+      "step": 18161
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.582502108218173e-06,
+      "loss": 0.2562,
+      "step": 18162
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.5815154812245e-06,
+      "loss": 0.2692,
+      "step": 18163
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.580528868322155e-06,
+      "loss": 0.2336,
+      "step": 18164
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.57954226952094e-06,
+      "loss": 0.2247,
+      "step": 18165
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.578555684830656e-06,
+      "loss": 0.2076,
+      "step": 18166
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.577569114261103e-06,
+      "loss": 0.2575,
+      "step": 18167
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.57658255782208e-06,
+      "loss": 0.2311,
+      "step": 18168
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.575596015523387e-06,
+      "loss": 0.2564,
+      "step": 18169
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.57460948737483e-06,
+      "loss": 0.2454,
+      "step": 18170
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.573622973386204e-06,
+      "loss": 0.2121,
+      "step": 18171
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.572636473567313e-06,
+      "loss": 0.256,
+      "step": 18172
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.571649987927955e-06,
+      "loss": 0.2312,
+      "step": 18173
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.570663516477927e-06,
+      "loss": 0.2053,
+      "step": 18174
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.569677059227033e-06,
+      "loss": 0.2274,
+      "step": 18175
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.568690616185069e-06,
+      "loss": 0.2848,
+      "step": 18176
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.56770418736184e-06,
+      "loss": 0.271,
+      "step": 18177
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.566717772767138e-06,
+      "loss": 0.1974,
+      "step": 18178
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.565731372410766e-06,
+      "loss": 0.2555,
+      "step": 18179
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.564744986302523e-06,
+      "loss": 0.2141,
+      "step": 18180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.563758614452209e-06,
+      "loss": 0.2484,
+      "step": 18181
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.562772256869621e-06,
+      "loss": 0.2372,
+      "step": 18182
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.561785913564557e-06,
+      "loss": 0.2337,
+      "step": 18183
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.560799584546815e-06,
+      "loss": 0.262,
+      "step": 18184
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.559813269826194e-06,
+      "loss": 0.2226,
+      "step": 18185
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.558826969412497e-06,
+      "loss": 0.1989,
+      "step": 18186
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.557840683315511e-06,
+      "loss": 0.2617,
+      "step": 18187
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.556854411545044e-06,
+      "loss": 0.1835,
+      "step": 18188
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.555868154110888e-06,
+      "loss": 0.2524,
+      "step": 18189
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.554881911022844e-06,
+      "loss": 0.205,
+      "step": 18190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55389568229071e-06,
+      "loss": 0.2355,
+      "step": 18191
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55290946792428e-06,
+      "loss": 0.2565,
+      "step": 18192
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55192326793335e-06,
+      "loss": 0.2126,
+      "step": 18193
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.550937082327719e-06,
+      "loss": 0.258,
+      "step": 18194
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.549950911117186e-06,
+      "loss": 0.2053,
+      "step": 18195
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.548964754311546e-06,
+      "loss": 0.2327,
+      "step": 18196
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.547978611920596e-06,
+      "loss": 0.2809,
+      "step": 18197
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.546992483954129e-06,
+      "loss": 0.18,
+      "step": 18198
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.546006370421947e-06,
+      "loss": 0.2199,
+      "step": 18199
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.545020271333838e-06,
+      "loss": 0.2308,
+      "step": 18200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.544034186699612e-06,
+      "loss": 0.2908,
+      "step": 18201
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.543048116529052e-06,
+      "loss": 0.2312,
+      "step": 18202
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.542062060831955e-06,
+      "loss": 0.198,
+      "step": 18203
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.541076019618121e-06,
+      "loss": 0.1968,
+      "step": 18204
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.540089992897343e-06,
+      "loss": 0.2491,
+      "step": 18205
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.539103980679421e-06,
+      "loss": 0.2452,
+      "step": 18206
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.538117982974143e-06,
+      "loss": 0.1817,
+      "step": 18207
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.537131999791305e-06,
+      "loss": 0.2767,
+      "step": 18208
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.536146031140706e-06,
+      "loss": 0.2106,
+      "step": 18209
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.535160077032136e-06,
+      "loss": 0.2178,
+      "step": 18210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.5341741374754e-06,
+      "loss": 0.3145,
+      "step": 18211
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.533188212480275e-06,
+      "loss": 0.2873,
+      "step": 18212
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.532202302056569e-06,
+      "loss": 0.2279,
+      "step": 18213
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.531216406214072e-06,
+      "loss": 0.2509,
+      "step": 18214
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.530230524962578e-06,
+      "loss": 0.278,
+      "step": 18215
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.529244658311879e-06,
+      "loss": 0.2688,
+      "step": 18216
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.528258806271771e-06,
+      "loss": 0.2299,
+      "step": 18217
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.527272968852047e-06,
+      "loss": 0.2279,
+      "step": 18218
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.526287146062498e-06,
+      "loss": 0.227,
+      "step": 18219
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.525301337912926e-06,
+      "loss": 0.2353,
+      "step": 18220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.52431554441311e-06,
+      "loss": 0.1867,
+      "step": 18221
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.523329765572853e-06,
+      "loss": 0.2334,
+      "step": 18222
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.522344001401945e-06,
+      "loss": 0.2069,
+      "step": 18223
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.52135825191018e-06,
+      "loss": 0.2178,
+      "step": 18224
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.520372517107352e-06,
+      "loss": 0.2458,
+      "step": 18225
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.519386797003246e-06,
+      "loss": 0.2769,
+      "step": 18226
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.518401091607661e-06,
+      "loss": 0.3029,
+      "step": 18227
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.517415400930384e-06,
+      "loss": 0.1879,
+      "step": 18228
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.516429724981212e-06,
+      "loss": 0.1731,
+      "step": 18229
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.515444063769938e-06,
+      "loss": 0.1793,
+      "step": 18230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.514458417306348e-06,
+      "loss": 0.2245,
+      "step": 18231
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.513472785600235e-06,
+      "loss": 0.203,
+      "step": 18232
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.512487168661392e-06,
+      "loss": 0.2235,
+      "step": 18233
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.511501566499607e-06,
+      "loss": 0.2321,
+      "step": 18234
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.51051597912468e-06,
+      "loss": 0.2188,
+      "step": 18235
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.50953040654639e-06,
+      "loss": 0.2319,
+      "step": 18236
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.508544848774532e-06,
+      "loss": 0.1934,
+      "step": 18237
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.5075593058189e-06,
+      "loss": 0.2767,
+      "step": 18238
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.50657377768928e-06,
+      "loss": 0.2825,
+      "step": 18239
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.505588264395469e-06,
+      "loss": 0.2361,
+      "step": 18240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.50460276594725e-06,
+      "loss": 0.1916,
+      "step": 18241
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.503617282354416e-06,
+      "loss": 0.228,
+      "step": 18242
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.502631813626753e-06,
+      "loss": 0.2744,
+      "step": 18243
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.50164635977406e-06,
+      "loss": 0.2472,
+      "step": 18244
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.500660920806119e-06,
+      "loss": 0.2221,
+      "step": 18245
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.499675496732718e-06,
+      "loss": 0.1962,
+      "step": 18246
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.498690087563652e-06,
+      "loss": 0.2554,
+      "step": 18247
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.497704693308706e-06,
+      "loss": 0.1821,
+      "step": 18248
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.496719313977675e-06,
+      "loss": 0.2653,
+      "step": 18249
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.495733949580339e-06,
+      "loss": 0.2421,
+      "step": 18250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.494748600126492e-06,
+      "loss": 0.2285,
+      "step": 18251
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.49376326562592e-06,
+      "loss": 0.2687,
+      "step": 18252
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.492777946088414e-06,
+      "loss": 0.2045,
+      "step": 18253
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.491792641523766e-06,
+      "loss": 0.1979,
+      "step": 18254
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.490807351941753e-06,
+      "loss": 0.2386,
+      "step": 18255
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.489822077352172e-06,
+      "loss": 0.274,
+      "step": 18256
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.488836817764807e-06,
+      "loss": 0.2254,
+      "step": 18257
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.487851573189447e-06,
+      "loss": 0.1957,
+      "step": 18258
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.486866343635882e-06,
+      "loss": 0.2055,
+      "step": 18259
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.485881129113894e-06,
+      "loss": 0.2044,
+      "step": 18260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.48489592963327e-06,
+      "loss": 0.1627,
+      "step": 18261
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.483910745203803e-06,
+      "loss": 0.295,
+      "step": 18262
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.482925575835276e-06,
+      "loss": 0.1805,
+      "step": 18263
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.481940421537478e-06,
+      "loss": 0.2872,
+      "step": 18264
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.480955282320192e-06,
+      "loss": 0.2307,
+      "step": 18265
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.479970158193206e-06,
+      "loss": 0.2373,
+      "step": 18266
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.478985049166309e-06,
+      "loss": 0.2282,
+      "step": 18267
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.477999955249282e-06,
+      "loss": 0.29,
+      "step": 18268
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.47701487645192e-06,
+      "loss": 0.2513,
+      "step": 18269
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.476029812783998e-06,
+      "loss": 0.216,
+      "step": 18270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.475044764255304e-06,
+      "loss": 0.2697,
+      "step": 18271
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.474059730875628e-06,
+      "loss": 0.1945,
+      "step": 18272
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.473074712654756e-06,
+      "loss": 0.25,
+      "step": 18273
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.472089709602468e-06,
+      "loss": 0.2989,
+      "step": 18274
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.471104721728551e-06,
+      "loss": 0.229,
+      "step": 18275
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.470119749042793e-06,
+      "loss": 0.2233,
+      "step": 18276
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.469134791554972e-06,
+      "loss": 0.2356,
+      "step": 18277
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.468149849274884e-06,
+      "loss": 0.2562,
+      "step": 18278
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.467164922212302e-06,
+      "loss": 0.2027,
+      "step": 18279
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.466180010377015e-06,
+      "loss": 0.1793,
+      "step": 18280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.465195113778806e-06,
+      "loss": 0.246,
+      "step": 18281
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.464210232427462e-06,
+      "loss": 0.1889,
+      "step": 18282
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.463225366332767e-06,
+      "loss": 0.3503,
+      "step": 18283
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.4622405155045e-06,
+      "loss": 0.221,
+      "step": 18284
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.461255679952448e-06,
+      "loss": 0.2528,
+      "step": 18285
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.460270859686392e-06,
+      "loss": 0.2249,
+      "step": 18286
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.459286054716116e-06,
+      "loss": 0.2197,
+      "step": 18287
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.458301265051411e-06,
+      "loss": 0.1888,
+      "step": 18288
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.457316490702048e-06,
+      "loss": 0.2875,
+      "step": 18289
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.456331731677814e-06,
+      "loss": 0.2445,
+      "step": 18290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.455346987988494e-06,
+      "loss": 0.19,
+      "step": 18291
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.45436225964387e-06,
+      "loss": 0.2218,
+      "step": 18292
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.453377546653724e-06,
+      "loss": 0.2051,
+      "step": 18293
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.452392849027836e-06,
+      "loss": 0.2283,
+      "step": 18294
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.451408166775987e-06,
+      "loss": 0.2274,
+      "step": 18295
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.450423499907965e-06,
+      "loss": 0.2899,
+      "step": 18296
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.449438848433552e-06,
+      "loss": 0.2111,
+      "step": 18297
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.44845421236252e-06,
+      "loss": 0.2029,
+      "step": 18298
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.447469591704657e-06,
+      "loss": 0.2407,
+      "step": 18299
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.446484986469747e-06,
+      "loss": 0.2885,
+      "step": 18300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.445500396667564e-06,
+      "loss": 0.2712,
+      "step": 18301
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.444515822307898e-06,
+      "loss": 0.2026,
+      "step": 18302
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.44353126340052e-06,
+      "loss": 0.227,
+      "step": 18303
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.442546719955215e-06,
+      "loss": 0.24,
+      "step": 18304
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.441562191981766e-06,
+      "loss": 0.25,
+      "step": 18305
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.440577679489951e-06,
+      "loss": 0.2638,
+      "step": 18306
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43959318248955e-06,
+      "loss": 0.2257,
+      "step": 18307
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.438608700990346e-06,
+      "loss": 0.2301,
+      "step": 18308
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.437624235002114e-06,
+      "loss": 0.2295,
+      "step": 18309
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.436639784534636e-06,
+      "loss": 0.222,
+      "step": 18310
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.2444,
+      "step": 18311
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.434670930201065e-06,
+      "loss": 0.2816,
+      "step": 18312
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.433686526354524e-06,
+      "loss": 0.2431,
+      "step": 18313
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.432702138067858e-06,
+      "loss": 0.2312,
+      "step": 18314
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.431717765350842e-06,
+      "loss": 0.2349,
+      "step": 18315
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.430733408213257e-06,
+      "loss": 0.219,
+      "step": 18316
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42974906666488e-06,
+      "loss": 0.2216,
+      "step": 18317
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.428764740715489e-06,
+      "loss": 0.3181,
+      "step": 18318
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.427780430374863e-06,
+      "loss": 0.2386,
+      "step": 18319
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.426796135652777e-06,
+      "loss": 0.2116,
+      "step": 18320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.425811856559017e-06,
+      "loss": 0.2556,
+      "step": 18321
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.424827593103356e-06,
+      "loss": 0.2607,
+      "step": 18322
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.423843345295571e-06,
+      "loss": 0.2365,
+      "step": 18323
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42285911314544e-06,
+      "loss": 0.1678,
+      "step": 18324
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42187489666274e-06,
+      "loss": 0.25,
+      "step": 18325
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.420890695857253e-06,
+      "loss": 0.2084,
+      "step": 18326
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41990651073875e-06,
+      "loss": 0.2174,
+      "step": 18327
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41892234131701e-06,
+      "loss": 0.2718,
+      "step": 18328
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.417938187601808e-06,
+      "loss": 0.1842,
+      "step": 18329
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.416954049602925e-06,
+      "loss": 0.2373,
+      "step": 18330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.415969927330138e-06,
+      "loss": 0.3051,
+      "step": 18331
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.414985820793217e-06,
+      "loss": 0.2045,
+      "step": 18332
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.414001730001944e-06,
+      "loss": 0.3013,
+      "step": 18333
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41301765496609e-06,
+      "loss": 0.238,
+      "step": 18334
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.412033595695436e-06,
+      "loss": 0.27,
+      "step": 18335
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.411049552199759e-06,
+      "loss": 0.2991,
+      "step": 18336
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.410065524488827e-06,
+      "loss": 0.2135,
+      "step": 18337
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.409081512572417e-06,
+      "loss": 0.218,
+      "step": 18338
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.40809751646031e-06,
+      "loss": 0.2174,
+      "step": 18339
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.407113536162276e-06,
+      "loss": 0.1856,
+      "step": 18340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.406129571688095e-06,
+      "loss": 0.2065,
+      "step": 18341
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.405145623047536e-06,
+      "loss": 0.2541,
+      "step": 18342
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.404161690250376e-06,
+      "loss": 0.2345,
+      "step": 18343
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.40317777330639e-06,
+      "loss": 0.231,
+      "step": 18344
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.40219387222535e-06,
+      "loss": 0.2513,
+      "step": 18345
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.401209987017037e-06,
+      "loss": 0.2215,
+      "step": 18346
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.400226117691216e-06,
+      "loss": 0.2748,
+      "step": 18347
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.399242264257665e-06,
+      "loss": 0.2158,
+      "step": 18348
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.398258426726157e-06,
+      "loss": 0.1982,
+      "step": 18349
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.397274605106467e-06,
+      "loss": 0.2393,
+      "step": 18350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39629079940837e-06,
+      "loss": 0.2248,
+      "step": 18351
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.395307009641634e-06,
+      "loss": 0.2281,
+      "step": 18352
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.394323235816032e-06,
+      "loss": 0.1907,
+      "step": 18353
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.393339477941342e-06,
+      "loss": 0.2128,
+      "step": 18354
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.392355736027338e-06,
+      "loss": 0.2066,
+      "step": 18355
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.391372010083782e-06,
+      "loss": 0.221,
+      "step": 18356
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.390388300120455e-06,
+      "loss": 0.2458,
+      "step": 18357
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.389404606147128e-06,
+      "loss": 0.22,
+      "step": 18358
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.388420928173575e-06,
+      "loss": 0.2326,
+      "step": 18359
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.387437266209565e-06,
+      "loss": 0.2221,
+      "step": 18360
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.386453620264868e-06,
+      "loss": 0.261,
+      "step": 18361
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.385469990349258e-06,
+      "loss": 0.209,
+      "step": 18362
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.384486376472505e-06,
+      "loss": 0.2788,
+      "step": 18363
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.383502778644382e-06,
+      "loss": 0.2139,
+      "step": 18364
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.382519196874664e-06,
+      "loss": 0.2647,
+      "step": 18365
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.381535631173114e-06,
+      "loss": 0.2728,
+      "step": 18366
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.380552081549508e-06,
+      "loss": 0.2102,
+      "step": 18367
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.379568548013613e-06,
+      "loss": 0.2621,
+      "step": 18368
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.378585030575201e-06,
+      "loss": 0.2098,
+      "step": 18369
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.377601529244048e-06,
+      "loss": 0.2587,
+      "step": 18370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.376618044029916e-06,
+      "loss": 0.2767,
+      "step": 18371
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.375634574942574e-06,
+      "loss": 0.2447,
+      "step": 18372
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.3746511219918e-06,
+      "loss": 0.2675,
+      "step": 18373
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.37366768518736e-06,
+      "loss": 0.2684,
+      "step": 18374
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.372684264539025e-06,
+      "loss": 0.2917,
+      "step": 18375
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.37170086005656e-06,
+      "loss": 0.2668,
+      "step": 18376
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.370717471749736e-06,
+      "loss": 0.2575,
+      "step": 18377
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.369734099628324e-06,
+      "loss": 0.2352,
+      "step": 18378
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36875074370209e-06,
+      "loss": 0.2337,
+      "step": 18379
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36776740398081e-06,
+      "loss": 0.2136,
+      "step": 18380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.366784080474241e-06,
+      "loss": 0.2223,
+      "step": 18381
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36580077319216e-06,
+      "loss": 0.2204,
+      "step": 18382
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.364817482144331e-06,
+      "loss": 0.2563,
+      "step": 18383
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.363834207340527e-06,
+      "loss": 0.2387,
+      "step": 18384
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36285094879051e-06,
+      "loss": 0.2329,
+      "step": 18385
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.361867706504052e-06,
+      "loss": 0.2488,
+      "step": 18386
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.360884480490918e-06,
+      "loss": 0.242,
+      "step": 18387
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.359901270760874e-06,
+      "loss": 0.2363,
+      "step": 18388
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.358918077323696e-06,
+      "loss": 0.2656,
+      "step": 18389
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.35793490018914e-06,
+      "loss": 0.3008,
+      "step": 18390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.356951739366978e-06,
+      "loss": 0.2496,
+      "step": 18391
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.355968594866978e-06,
+      "loss": 0.2452,
+      "step": 18392
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.354985466698906e-06,
+      "loss": 0.2469,
+      "step": 18393
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.354002354872528e-06,
+      "loss": 0.2036,
+      "step": 18394
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.353019259397609e-06,
+      "loss": 0.214,
+      "step": 18395
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.352036180283916e-06,
+      "loss": 0.2254,
+      "step": 18396
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.351053117541212e-06,
+      "loss": 0.2247,
+      "step": 18397
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.35007007117927e-06,
+      "loss": 0.2412,
+      "step": 18398
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.349087041207854e-06,
+      "loss": 0.2779,
+      "step": 18399
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.348104027636722e-06,
+      "loss": 0.238,
+      "step": 18400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.347121030475645e-06,
+      "loss": 0.234,
+      "step": 18401
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.346138049734388e-06,
+      "loss": 0.2436,
+      "step": 18402
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.345155085422717e-06,
+      "loss": 0.2196,
+      "step": 18403
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.344172137550397e-06,
+      "loss": 0.2352,
+      "step": 18404
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.343189206127189e-06,
+      "loss": 0.237,
+      "step": 18405
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.342206291162857e-06,
+      "loss": 0.2053,
+      "step": 18406
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.341223392667171e-06,
+      "loss": 0.2491,
+      "step": 18407
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.340240510649892e-06,
+      "loss": 0.1964,
+      "step": 18408
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.339257645120786e-06,
+      "loss": 0.2227,
+      "step": 18409
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.338274796089613e-06,
+      "loss": 0.2223,
+      "step": 18410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.337291963566139e-06,
+      "loss": 0.1896,
+      "step": 18411
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.336309147560124e-06,
+      "loss": 0.1824,
+      "step": 18412
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.335326348081341e-06,
+      "loss": 0.2146,
+      "step": 18413
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.334343565139544e-06,
+      "loss": 0.2467,
+      "step": 18414
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.333360798744496e-06,
+      "loss": 0.2227,
+      "step": 18415
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.332378048905964e-06,
+      "loss": 0.2903,
+      "step": 18416
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.331395315633712e-06,
+      "loss": 0.2355,
+      "step": 18417
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.3304125989375e-06,
+      "loss": 0.2192,
+      "step": 18418
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.329429898827089e-06,
+      "loss": 0.1892,
+      "step": 18419
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.328447215312241e-06,
+      "loss": 0.2296,
+      "step": 18420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.32746454840272e-06,
+      "loss": 0.2606,
+      "step": 18421
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.326481898108287e-06,
+      "loss": 0.2299,
+      "step": 18422
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.325499264438711e-06,
+      "loss": 0.2185,
+      "step": 18423
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.32451664740374e-06,
+      "loss": 0.231,
+      "step": 18424
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.323534047013141e-06,
+      "loss": 0.2156,
+      "step": 18425
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.32255146327668e-06,
+      "loss": 0.2255,
+      "step": 18426
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.321568896204114e-06,
+      "loss": 0.2432,
+      "step": 18427
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.320586345805207e-06,
+      "loss": 0.2178,
+      "step": 18428
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.319603812089713e-06,
+      "loss": 0.1951,
+      "step": 18429
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.318621295067396e-06,
+      "loss": 0.28,
+      "step": 18430
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.317638794748018e-06,
+      "loss": 0.2304,
+      "step": 18431
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.316656311141338e-06,
+      "loss": 0.2456,
+      "step": 18432
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.31567384425712e-06,
+      "loss": 0.2684,
+      "step": 18433
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.314691394105119e-06,
+      "loss": 0.1681,
+      "step": 18434
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.313708960695094e-06,
+      "loss": 0.2293,
+      "step": 18435
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.312726544036806e-06,
+      "loss": 0.2012,
+      "step": 18436
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.311744144140014e-06,
+      "loss": 0.161,
+      "step": 18437
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.310761761014485e-06,
+      "loss": 0.1952,
+      "step": 18438
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.309779394669966e-06,
+      "loss": 0.2453,
+      "step": 18439
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.30879704511622e-06,
+      "loss": 0.2335,
+      "step": 18440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.307814712363006e-06,
+      "loss": 0.2218,
+      "step": 18441
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.306832396420089e-06,
+      "loss": 0.228,
+      "step": 18442
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.305850097297219e-06,
+      "loss": 0.2689,
+      "step": 18443
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.304867815004157e-06,
+      "loss": 0.2763,
+      "step": 18444
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.30388554955066e-06,
+      "loss": 0.217,
+      "step": 18445
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.302903300946486e-06,
+      "loss": 0.2262,
+      "step": 18446
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.3019210692014e-06,
+      "loss": 0.2158,
+      "step": 18447
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.300938854325146e-06,
+      "loss": 0.2704,
+      "step": 18448
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.29995665632749e-06,
+      "loss": 0.2351,
+      "step": 18449
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.29897447521819e-06,
+      "loss": 0.2139,
+      "step": 18450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.297992311006999e-06,
+      "loss": 0.2834,
+      "step": 18451
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.29701016370368e-06,
+      "loss": 0.2996,
+      "step": 18452
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.296028033317983e-06,
+      "loss": 0.2538,
+      "step": 18453
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.295045919859667e-06,
+      "loss": 0.2102,
+      "step": 18454
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.294063823338486e-06,
+      "loss": 0.222,
+      "step": 18455
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.293081743764201e-06,
+      "loss": 0.2364,
+      "step": 18456
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.292099681146571e-06,
+      "loss": 0.2481,
+      "step": 18457
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.291117635495338e-06,
+      "loss": 0.2033,
+      "step": 18458
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.290135606820271e-06,
+      "loss": 0.2342,
+      "step": 18459
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.289153595131119e-06,
+      "loss": 0.2284,
+      "step": 18460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.28817160043764e-06,
+      "loss": 0.205,
+      "step": 18461
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.287189622749591e-06,
+      "loss": 0.2724,
+      "step": 18462
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.286207662076723e-06,
+      "loss": 0.2282,
+      "step": 18463
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.285225718428789e-06,
+      "loss": 0.1774,
+      "step": 18464
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.284243791815549e-06,
+      "loss": 0.2571,
+      "step": 18465
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.283261882246762e-06,
+      "loss": 0.2469,
+      "step": 18466
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.282279989732169e-06,
+      "loss": 0.2177,
+      "step": 18467
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.281298114281532e-06,
+      "loss": 0.229,
+      "step": 18468
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.280316255904604e-06,
+      "loss": 0.2442,
+      "step": 18469
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27933441461114e-06,
+      "loss": 0.2553,
+      "step": 18470
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.278352590410894e-06,
+      "loss": 0.2679,
+      "step": 18471
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27737078331362e-06,
+      "loss": 0.227,
+      "step": 18472
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.276388993329062e-06,
+      "loss": 0.2361,
+      "step": 18473
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.275407220466987e-06,
+      "loss": 0.1708,
+      "step": 18474
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.274425464737139e-06,
+      "loss": 0.1678,
+      "step": 18475
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.273443726149278e-06,
+      "loss": 0.213,
+      "step": 18476
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.272462004713149e-06,
+      "loss": 0.2428,
+      "step": 18477
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.271480300438508e-06,
+      "loss": 0.2203,
+      "step": 18478
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.270498613335106e-06,
+      "loss": 0.2192,
+      "step": 18479
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.269516943412696e-06,
+      "loss": 0.2219,
+      "step": 18480
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.268535290681034e-06,
+      "loss": 0.2487,
+      "step": 18481
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.267553655149864e-06,
+      "loss": 0.2213,
+      "step": 18482
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.266572036828943e-06,
+      "loss": 0.3132,
+      "step": 18483
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.26559043572802e-06,
+      "loss": 0.2782,
+      "step": 18484
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.26460885185685e-06,
+      "loss": 0.2364,
+      "step": 18485
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.263627285225183e-06,
+      "loss": 0.3186,
+      "step": 18486
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.262645735842765e-06,
+      "loss": 0.2481,
+      "step": 18487
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.261664203719351e-06,
+      "loss": 0.2594,
+      "step": 18488
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.260682688864689e-06,
+      "loss": 0.2937,
+      "step": 18489
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.259701191288535e-06,
+      "loss": 0.2211,
+      "step": 18490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.258719711000636e-06,
+      "loss": 0.2377,
+      "step": 18491
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.25773824801074e-06,
+      "loss": 0.1969,
+      "step": 18492
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.256756802328598e-06,
+      "loss": 0.198,
+      "step": 18493
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.25577537396396e-06,
+      "loss": 0.2599,
+      "step": 18494
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.254793962926579e-06,
+      "loss": 0.231,
+      "step": 18495
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2538125692262e-06,
+      "loss": 0.2155,
+      "step": 18496
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.252831192872573e-06,
+      "loss": 0.1883,
+      "step": 18497
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.251849833875444e-06,
+      "loss": 0.2438,
+      "step": 18498
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.250868492244571e-06,
+      "loss": 0.2193,
+      "step": 18499
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.249887167989698e-06,
+      "loss": 0.211,
+      "step": 18500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.248905861120569e-06,
+      "loss": 0.1935,
+      "step": 18501
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.247924571646937e-06,
+      "loss": 0.2115,
+      "step": 18502
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.24694329957855e-06,
+      "loss": 0.2446,
+      "step": 18503
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.245962044925155e-06,
+      "loss": 0.2552,
+      "step": 18504
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.244980807696504e-06,
+      "loss": 0.3349,
+      "step": 18505
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.243999587902336e-06,
+      "loss": 0.224,
+      "step": 18506
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.243018385552404e-06,
+      "loss": 0.1991,
+      "step": 18507
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.242037200656455e-06,
+      "loss": 0.2344,
+      "step": 18508
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.241056033224237e-06,
+      "loss": 0.2715,
+      "step": 18509
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.240074883265498e-06,
+      "loss": 0.2488,
+      "step": 18510
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23909375078998e-06,
+      "loss": 0.2892,
+      "step": 18511
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.238112635807434e-06,
+      "loss": 0.2244,
+      "step": 18512
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.237131538327605e-06,
+      "loss": 0.2483,
+      "step": 18513
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.236150458360235e-06,
+      "loss": 0.185,
+      "step": 18514
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23516939591508e-06,
+      "loss": 0.2826,
+      "step": 18515
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.234188351001877e-06,
+      "loss": 0.2523,
+      "step": 18516
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.233207323630376e-06,
+      "loss": 0.1753,
+      "step": 18517
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.232226313810321e-06,
+      "loss": 0.2219,
+      "step": 18518
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.231245321551459e-06,
+      "loss": 0.2312,
+      "step": 18519
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.230264346863536e-06,
+      "loss": 0.2084,
+      "step": 18520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.229283389756293e-06,
+      "loss": 0.2211,
+      "step": 18521
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.228302450239478e-06,
+      "loss": 0.2675,
+      "step": 18522
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.22732152832283e-06,
+      "loss": 0.2134,
+      "step": 18523
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.22634062401611e-06,
+      "loss": 0.1965,
+      "step": 18524
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.22535973732904e-06,
+      "loss": 0.1828,
+      "step": 18525
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.22437886827138e-06,
+      "loss": 0.2326,
+      "step": 18526
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.223398016852868e-06,
+      "loss": 0.2224,
+      "step": 18527
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.222417183083249e-06,
+      "loss": 0.2481,
+      "step": 18528
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.221436366972268e-06,
+      "loss": 0.2146,
+      "step": 18529
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.220455568529667e-06,
+      "loss": 0.271,
+      "step": 18530
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.219474787765188e-06,
+      "loss": 0.2351,
+      "step": 18531
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.218494024688574e-06,
+      "loss": 0.2088,
+      "step": 18532
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.217513279309573e-06,
+      "loss": 0.2226,
+      "step": 18533
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.216532551637924e-06,
+      "loss": 0.2374,
+      "step": 18534
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.215551841683369e-06,
+      "loss": 0.1955,
+      "step": 18535
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.214571149455652e-06,
+      "loss": 0.1758,
+      "step": 18536
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.213590474964514e-06,
+      "loss": 0.2469,
+      "step": 18537
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.212609818219698e-06,
+      "loss": 0.2277,
+      "step": 18538
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.211629179230949e-06,
+      "loss": 0.2482,
+      "step": 18539
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.210648558008003e-06,
+      "loss": 0.2539,
+      "step": 18540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2096679545606e-06,
+      "loss": 0.1823,
+      "step": 18541
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.208687368898489e-06,
+      "loss": 0.2116,
+      "step": 18542
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.207706801031407e-06,
+      "loss": 0.2337,
+      "step": 18543
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2067262509691e-06,
+      "loss": 0.2228,
+      "step": 18544
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2057457187213e-06,
+      "loss": 0.3083,
+      "step": 18545
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.204765204297753e-06,
+      "loss": 0.2092,
+      "step": 18546
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.203784707708198e-06,
+      "loss": 0.2358,
+      "step": 18547
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.202804228962374e-06,
+      "loss": 0.2461,
+      "step": 18548
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.20182376807003e-06,
+      "loss": 0.2709,
+      "step": 18549
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.200843325040893e-06,
+      "loss": 0.2055,
+      "step": 18550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.19986289988471e-06,
+      "loss": 0.232,
+      "step": 18551
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.198882492611222e-06,
+      "loss": 0.2573,
+      "step": 18552
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.197902103230167e-06,
+      "loss": 0.2219,
+      "step": 18553
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.196921731751279e-06,
+      "loss": 0.2241,
+      "step": 18554
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.195941378184303e-06,
+      "loss": 0.2163,
+      "step": 18555
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.194961042538977e-06,
+      "loss": 0.2249,
+      "step": 18556
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.193980724825036e-06,
+      "loss": 0.2928,
+      "step": 18557
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.193000425052228e-06,
+      "loss": 0.2927,
+      "step": 18558
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.192020143230278e-06,
+      "loss": 0.1936,
+      "step": 18559
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.191039879368933e-06,
+      "loss": 0.2727,
+      "step": 18560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.190059633477931e-06,
+      "loss": 0.2223,
+      "step": 18561
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.189079405567007e-06,
+      "loss": 0.2462,
+      "step": 18562
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.188099195645901e-06,
+      "loss": 0.2205,
+      "step": 18563
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.187119003724347e-06,
+      "loss": 0.2378,
+      "step": 18564
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.186138829812085e-06,
+      "loss": 0.2008,
+      "step": 18565
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.185158673918848e-06,
+      "loss": 0.2466,
+      "step": 18566
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18417853605438e-06,
+      "loss": 0.253,
+      "step": 18567
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.183198416228415e-06,
+      "loss": 0.2304,
+      "step": 18568
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.182218314450689e-06,
+      "loss": 0.2402,
+      "step": 18569
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.181238230730936e-06,
+      "loss": 0.2361,
+      "step": 18570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.180258165078894e-06,
+      "loss": 0.2596,
+      "step": 18571
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.1792781175043e-06,
+      "loss": 0.2447,
+      "step": 18572
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.178298088016891e-06,
+      "loss": 0.241,
+      "step": 18573
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.1773180766264e-06,
+      "loss": 0.1786,
+      "step": 18574
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.17633808334256e-06,
+      "loss": 0.2416,
+      "step": 18575
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.175358108175115e-06,
+      "loss": 0.186,
+      "step": 18576
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.174378151133792e-06,
+      "loss": 0.3634,
+      "step": 18577
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.173398212228332e-06,
+      "loss": 0.2177,
+      "step": 18578
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.172418291468465e-06,
+      "loss": 0.2081,
+      "step": 18579
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.171438388863925e-06,
+      "loss": 0.226,
+      "step": 18580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.170458504424448e-06,
+      "loss": 0.2428,
+      "step": 18581
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.169478638159776e-06,
+      "loss": 0.1714,
+      "step": 18582
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16849879007963e-06,
+      "loss": 0.1796,
+      "step": 18583
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16751896019375e-06,
+      "loss": 0.27,
+      "step": 18584
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.166539148511871e-06,
+      "loss": 0.2164,
+      "step": 18585
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.165559355043726e-06,
+      "loss": 0.2158,
+      "step": 18586
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.164579579799047e-06,
+      "loss": 0.2175,
+      "step": 18587
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.163599822787567e-06,
+      "loss": 0.2484,
+      "step": 18588
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16262008401902e-06,
+      "loss": 0.2385,
+      "step": 18589
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.161640363503135e-06,
+      "loss": 0.2232,
+      "step": 18590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16066066124965e-06,
+      "loss": 0.2177,
+      "step": 18591
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.159680977268301e-06,
+      "loss": 0.2205,
+      "step": 18592
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.158701311568808e-06,
+      "loss": 0.2857,
+      "step": 18593
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.157721664160913e-06,
+      "loss": 0.2178,
+      "step": 18594
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.156742035054343e-06,
+      "loss": 0.2515,
+      "step": 18595
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.155762424258832e-06,
+      "loss": 0.2539,
+      "step": 18596
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.154782831784112e-06,
+      "loss": 0.2254,
+      "step": 18597
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.153803257639912e-06,
+      "loss": 0.2501,
+      "step": 18598
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.152823701835964e-06,
+      "loss": 0.2326,
+      "step": 18599
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.151844164381998e-06,
+      "loss": 0.1945,
+      "step": 18600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.150864645287748e-06,
+      "loss": 0.1909,
+      "step": 18601
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.149885144562944e-06,
+      "loss": 0.233,
+      "step": 18602
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.148905662217315e-06,
+      "loss": 0.1849,
+      "step": 18603
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.147926198260589e-06,
+      "loss": 0.2303,
+      "step": 18604
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.1469467527025e-06,
+      "loss": 0.2142,
+      "step": 18605
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.145967325552774e-06,
+      "loss": 0.2539,
+      "step": 18606
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.144987916821149e-06,
+      "loss": 0.1996,
+      "step": 18607
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.144008526517346e-06,
+      "loss": 0.2054,
+      "step": 18608
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.143029154651092e-06,
+      "loss": 0.2644,
+      "step": 18609
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.142049801232126e-06,
+      "loss": 0.2174,
+      "step": 18610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.141070466270172e-06,
+      "loss": 0.2023,
+      "step": 18611
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.140091149774958e-06,
+      "loss": 0.2752,
+      "step": 18612
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.139111851756215e-06,
+      "loss": 0.2375,
+      "step": 18613
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.138132572223667e-06,
+      "loss": 0.2347,
+      "step": 18614
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.137153311187044e-06,
+      "loss": 0.1927,
+      "step": 18615
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.13617406865608e-06,
+      "loss": 0.2297,
+      "step": 18616
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.135194844640498e-06,
+      "loss": 0.2687,
+      "step": 18617
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.134215639150018e-06,
+      "loss": 0.2361,
+      "step": 18618
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.133236452194382e-06,
+      "loss": 0.1902,
+      "step": 18619
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.13225728378331e-06,
+      "loss": 0.2624,
+      "step": 18620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.131278133926528e-06,
+      "loss": 0.2269,
+      "step": 18621
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.130299002633766e-06,
+      "loss": 0.184,
+      "step": 18622
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.129319889914748e-06,
+      "loss": 0.2241,
+      "step": 18623
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.1283407957792e-06,
+      "loss": 0.1952,
+      "step": 18624
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.127361720236854e-06,
+      "loss": 0.2499,
+      "step": 18625
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.126382663297435e-06,
+      "loss": 0.2176,
+      "step": 18626
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.12540362497066e-06,
+      "loss": 0.2394,
+      "step": 18627
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.124424605266265e-06,
+      "loss": 0.221,
+      "step": 18628
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.123445604193972e-06,
+      "loss": 0.1828,
+      "step": 18629
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.122466621763506e-06,
+      "loss": 0.2248,
+      "step": 18630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.121487657984593e-06,
+      "loss": 0.2386,
+      "step": 18631
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.120508712866957e-06,
+      "loss": 0.2094,
+      "step": 18632
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.119529786420323e-06,
+      "loss": 0.2716,
+      "step": 18633
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.118550878654417e-06,
+      "loss": 0.3028,
+      "step": 18634
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.117571989578969e-06,
+      "loss": 0.2226,
+      "step": 18635
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.116593119203689e-06,
+      "loss": 0.2676,
+      "step": 18636
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.115614267538313e-06,
+      "loss": 0.1846,
+      "step": 18637
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.114635434592562e-06,
+      "loss": 0.1896,
+      "step": 18638
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.113656620376158e-06,
+      "loss": 0.2323,
+      "step": 18639
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.11267782489883e-06,
+      "loss": 0.2242,
+      "step": 18640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.111699048170294e-06,
+      "loss": 0.2304,
+      "step": 18641
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.110720290200274e-06,
+      "loss": 0.2355,
+      "step": 18642
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.109741550998498e-06,
+      "loss": 0.2638,
+      "step": 18643
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.108762830574688e-06,
+      "loss": 0.2526,
+      "step": 18644
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.107784128938566e-06,
+      "loss": 0.206,
+      "step": 18645
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.106805446099853e-06,
+      "loss": 0.2247,
+      "step": 18646
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10582678206827e-06,
+      "loss": 0.2117,
+      "step": 18647
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.104848136853543e-06,
+      "loss": 0.1836,
+      "step": 18648
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.103869510465389e-06,
+      "loss": 0.1884,
+      "step": 18649
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10289090291354e-06,
+      "loss": 0.184,
+      "step": 18650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.101912314207703e-06,
+      "loss": 0.2906,
+      "step": 18651
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.100933744357608e-06,
+      "loss": 0.2125,
+      "step": 18652
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.099955193372977e-06,
+      "loss": 0.2379,
+      "step": 18653
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.098976661263529e-06,
+      "loss": 0.2602,
+      "step": 18654
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.097998148038986e-06,
+      "loss": 0.2516,
+      "step": 18655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.097019653709065e-06,
+      "loss": 0.2571,
+      "step": 18656
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.09604117828349e-06,
+      "loss": 0.1725,
+      "step": 18657
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.095062721771978e-06,
+      "loss": 0.2204,
+      "step": 18658
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.094084284184252e-06,
+      "loss": 0.2307,
+      "step": 18659
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.093105865530035e-06,
+      "loss": 0.2297,
+      "step": 18660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.092127465819038e-06,
+      "loss": 0.2507,
+      "step": 18661
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.091149085060985e-06,
+      "loss": 0.2109,
+      "step": 18662
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.090170723265598e-06,
+      "loss": 0.2165,
+      "step": 18663
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.089192380442594e-06,
+      "loss": 0.2676,
+      "step": 18664
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.08821405660169e-06,
+      "loss": 0.2385,
+      "step": 18665
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.087235751752607e-06,
+      "loss": 0.319,
+      "step": 18666
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.08625746590506e-06,
+      "loss": 0.2406,
+      "step": 18667
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.085279199068771e-06,
+      "loss": 0.1913,
+      "step": 18668
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.084300951253463e-06,
+      "loss": 0.2343,
+      "step": 18669
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.083322722468842e-06,
+      "loss": 0.2788,
+      "step": 18670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.082344512724634e-06,
+      "loss": 0.1821,
+      "step": 18671
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.081366322030555e-06,
+      "loss": 0.2082,
+      "step": 18672
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.080388150396323e-06,
+      "loss": 0.2456,
+      "step": 18673
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.079409997831655e-06,
+      "loss": 0.2596,
+      "step": 18674
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.078431864346265e-06,
+      "loss": 0.1634,
+      "step": 18675
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.077453749949871e-06,
+      "loss": 0.2544,
+      "step": 18676
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.076475654652193e-06,
+      "loss": 0.2432,
+      "step": 18677
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.075497578462944e-06,
+      "loss": 0.1926,
+      "step": 18678
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.074519521391846e-06,
+      "loss": 0.2332,
+      "step": 18679
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.073541483448609e-06,
+      "loss": 0.2732,
+      "step": 18680
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.072563464642949e-06,
+      "loss": 0.2572,
+      "step": 18681
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.071585464984584e-06,
+      "loss": 0.2399,
+      "step": 18682
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.070607484483226e-06,
+      "loss": 0.2238,
+      "step": 18683
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.0696295231486e-06,
+      "loss": 0.2598,
+      "step": 18684
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.068651580990407e-06,
+      "loss": 0.2891,
+      "step": 18685
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.067673658018374e-06,
+      "loss": 0.2798,
+      "step": 18686
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.066695754242209e-06,
+      "loss": 0.2184,
+      "step": 18687
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.06571786967163e-06,
+      "loss": 0.2148,
+      "step": 18688
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.064740004316351e-06,
+      "loss": 0.2447,
+      "step": 18689
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.063762158186083e-06,
+      "loss": 0.1732,
+      "step": 18690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.062784331290543e-06,
+      "loss": 0.204,
+      "step": 18691
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.061806523639442e-06,
+      "loss": 0.2151,
+      "step": 18692
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.060828735242502e-06,
+      "loss": 0.2113,
+      "step": 18693
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.059850966109422e-06,
+      "loss": 0.2422,
+      "step": 18694
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.058873216249927e-06,
+      "loss": 0.2095,
+      "step": 18695
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.057895485673726e-06,
+      "loss": 0.2656,
+      "step": 18696
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.056917774390533e-06,
+      "loss": 0.2292,
+      "step": 18697
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.055940082410061e-06,
+      "loss": 0.1941,
+      "step": 18698
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.054962409742018e-06,
+      "loss": 0.1836,
+      "step": 18699
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053984756396121e-06,
+      "loss": 0.2866,
+      "step": 18700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053007122382078e-06,
+      "loss": 0.2321,
+      "step": 18701
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.052029507709607e-06,
+      "loss": 0.2432,
+      "step": 18702
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.051051912388416e-06,
+      "loss": 0.2172,
+      "step": 18703
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.050074336428217e-06,
+      "loss": 0.2568,
+      "step": 18704
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.2416,
+      "step": 18705
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.048119242629635e-06,
+      "loss": 0.2163,
+      "step": 18706
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.047141724810677e-06,
+      "loss": 0.1962,
+      "step": 18707
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.046164226391557e-06,
+      "loss": 0.2014,
+      "step": 18708
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.045186747381981e-06,
+      "loss": 0.2277,
+      "step": 18709
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.04420928779166e-06,
+      "loss": 0.2316,
+      "step": 18710
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.043231847630308e-06,
+      "loss": 0.2022,
+      "step": 18711
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.042254426907631e-06,
+      "loss": 0.3129,
+      "step": 18712
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.041277025633346e-06,
+      "loss": 0.2076,
+      "step": 18713
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040299643817152e-06,
+      "loss": 0.2831,
+      "step": 18714
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.039322281468765e-06,
+      "loss": 0.2505,
+      "step": 18715
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03834493859789e-06,
+      "loss": 0.2144,
+      "step": 18716
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03736761521424e-06,
+      "loss": 0.2455,
+      "step": 18717
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.036390311327528e-06,
+      "loss": 0.2414,
+      "step": 18718
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03541302694745e-06,
+      "loss": 0.174,
+      "step": 18719
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.034435762083723e-06,
+      "loss": 0.2674,
+      "step": 18720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.033458516746055e-06,
+      "loss": 0.2239,
+      "step": 18721
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.032481290944155e-06,
+      "loss": 0.2058,
+      "step": 18722
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.031504084687726e-06,
+      "loss": 0.2047,
+      "step": 18723
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03052689798648e-06,
+      "loss": 0.2084,
+      "step": 18724
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02954973085012e-06,
+      "loss": 0.2476,
+      "step": 18725
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.028572583288354e-06,
+      "loss": 0.2416,
+      "step": 18726
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.0275954553109e-06,
+      "loss": 0.221,
+      "step": 18727
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.026618346927446e-06,
+      "loss": 0.2477,
+      "step": 18728
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.025641258147713e-06,
+      "loss": 0.217,
+      "step": 18729
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.024664188981402e-06,
+      "loss": 0.2123,
+      "step": 18730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.023687139438222e-06,
+      "loss": 0.2362,
+      "step": 18731
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.022710109527876e-06,
+      "loss": 0.2904,
+      "step": 18732
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.021733099260071e-06,
+      "loss": 0.207,
+      "step": 18733
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.020756108644514e-06,
+      "loss": 0.258,
+      "step": 18734
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.019779137690906e-06,
+      "loss": 0.2312,
+      "step": 18735
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.018802186408959e-06,
+      "loss": 0.2442,
+      "step": 18736
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.017825254808376e-06,
+      "loss": 0.2053,
+      "step": 18737
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.016848342898858e-06,
+      "loss": 0.1968,
+      "step": 18738
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.015871450690112e-06,
+      "loss": 0.1922,
+      "step": 18739
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.014894578191846e-06,
+      "loss": 0.261,
+      "step": 18740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.013917725413756e-06,
+      "loss": 0.2617,
+      "step": 18741
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.01294089236556e-06,
+      "loss": 0.2153,
+      "step": 18742
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.011964079056948e-06,
+      "loss": 0.2018,
+      "step": 18743
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.010987285497626e-06,
+      "loss": 0.2538,
+      "step": 18744
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.010010511697303e-06,
+      "loss": 0.2384,
+      "step": 18745
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.009033757665682e-06,
+      "loss": 0.1838,
+      "step": 18746
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.008057023412466e-06,
+      "loss": 0.266,
+      "step": 18747
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.007080308947351e-06,
+      "loss": 0.1956,
+      "step": 18748
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.006103614280048e-06,
+      "loss": 0.2508,
+      "step": 18749
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.005126939420252e-06,
+      "loss": 0.1613,
+      "step": 18750
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.004150284377677e-06,
+      "loss": 0.2438,
+      "step": 18751
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.003173649162013e-06,
+      "loss": 0.2326,
+      "step": 18752
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.002197033782965e-06,
+      "loss": 0.2588,
+      "step": 18753
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.00122043825024e-06,
+      "loss": 0.2163,
+      "step": 18754
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.000243862573536e-06,
+      "loss": 0.1785,
+      "step": 18755
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.999267306762556e-06,
+      "loss": 0.2434,
+      "step": 18756
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.998290770826998e-06,
+      "loss": 0.2098,
+      "step": 18757
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.997314254776564e-06,
+      "loss": 0.224,
+      "step": 18758
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.996337758620956e-06,
+      "loss": 0.2477,
+      "step": 18759
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.995361282369871e-06,
+      "loss": 0.242,
+      "step": 18760
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.994384826033019e-06,
+      "loss": 0.1484,
+      "step": 18761
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.993408389620088e-06,
+      "loss": 0.2509,
+      "step": 18762
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.992431973140786e-06,
+      "loss": 0.1876,
+      "step": 18763
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.991455576604809e-06,
+      "loss": 0.2848,
+      "step": 18764
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.990479200021858e-06,
+      "loss": 0.1962,
+      "step": 18765
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.989502843401638e-06,
+      "loss": 0.2289,
+      "step": 18766
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.988526506753838e-06,
+      "loss": 0.2166,
+      "step": 18767
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.98755019008816e-06,
+      "loss": 0.219,
+      "step": 18768
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.986573893414305e-06,
+      "loss": 0.2012,
+      "step": 18769
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.985597616741972e-06,
+      "loss": 0.1774,
+      "step": 18770
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.98462136008086e-06,
+      "loss": 0.2528,
+      "step": 18771
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.983645123440665e-06,
+      "loss": 0.2458,
+      "step": 18772
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.982668906831084e-06,
+      "loss": 0.197,
+      "step": 18773
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.981692710261817e-06,
+      "loss": 0.2494,
+      "step": 18774
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.980716533742558e-06,
+      "loss": 0.2182,
+      "step": 18775
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.979740377283017e-06,
+      "loss": 0.1812,
+      "step": 18776
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.978764240892874e-06,
+      "loss": 0.2342,
+      "step": 18777
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.977788124581833e-06,
+      "loss": 0.2418,
+      "step": 18778
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.976812028359594e-06,
+      "loss": 0.1514,
+      "step": 18779
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.975835952235854e-06,
+      "loss": 0.2328,
+      "step": 18780
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.974859896220301e-06,
+      "loss": 0.2159,
+      "step": 18781
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.97388386032264e-06,
+      "loss": 0.172,
+      "step": 18782
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.972907844552562e-06,
+      "loss": 0.2141,
+      "step": 18783
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.971931848919763e-06,
+      "loss": 0.2657,
+      "step": 18784
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.970955873433947e-06,
+      "loss": 0.2362,
+      "step": 18785
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.969979918104798e-06,
+      "loss": 0.1946,
+      "step": 18786
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.969003982942012e-06,
+      "loss": 0.2507,
+      "step": 18787
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96802806795529e-06,
+      "loss": 0.3039,
+      "step": 18788
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.967052173154328e-06,
+      "loss": 0.2186,
+      "step": 18789
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.966076298548816e-06,
+      "loss": 0.2429,
+      "step": 18790
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96510044414845e-06,
+      "loss": 0.2347,
+      "step": 18791
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.964124609962923e-06,
+      "loss": 0.2272,
+      "step": 18792
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.963148796001927e-06,
+      "loss": 0.2225,
+      "step": 18793
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96217300227516e-06,
+      "loss": 0.2152,
+      "step": 18794
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96119722879232e-06,
+      "loss": 0.1965,
+      "step": 18795
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.960221475563089e-06,
+      "loss": 0.1808,
+      "step": 18796
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.959245742597166e-06,
+      "loss": 0.2,
+      "step": 18797
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.958270029904245e-06,
+      "loss": 0.2404,
+      "step": 18798
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.957294337494018e-06,
+      "loss": 0.1972,
+      "step": 18799
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.956318665376177e-06,
+      "loss": 0.1741,
+      "step": 18800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.955343013560416e-06,
+      "loss": 0.2031,
+      "step": 18801
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.954367382056421e-06,
+      "loss": 0.2188,
+      "step": 18802
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.95339177087389e-06,
+      "loss": 0.2806,
+      "step": 18803
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.95241618002252e-06,
+      "loss": 0.1644,
+      "step": 18804
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.95144060951199e-06,
+      "loss": 0.2127,
+      "step": 18805
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.950465059351999e-06,
+      "loss": 0.177,
+      "step": 18806
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.949489529552238e-06,
+      "loss": 0.2124,
+      "step": 18807
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.948514020122396e-06,
+      "loss": 0.2215,
+      "step": 18808
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.947538531072165e-06,
+      "loss": 0.2735,
+      "step": 18809
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.946563062411236e-06,
+      "loss": 0.3266,
+      "step": 18810
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.945587614149295e-06,
+      "loss": 0.2443,
+      "step": 18811
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.944612186296036e-06,
+      "loss": 0.2744,
+      "step": 18812
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.943636778861152e-06,
+      "loss": 0.1855,
+      "step": 18813
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.942661391854329e-06,
+      "loss": 0.2357,
+      "step": 18814
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.941686025285257e-06,
+      "loss": 0.2246,
+      "step": 18815
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.940710679163623e-06,
+      "loss": 0.1945,
+      "step": 18816
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.939735353499121e-06,
+      "loss": 0.2334,
+      "step": 18817
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.938760048301432e-06,
+      "loss": 0.276,
+      "step": 18818
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.93778476358026e-06,
+      "loss": 0.2941,
+      "step": 18819
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.936809499345278e-06,
+      "loss": 0.258,
+      "step": 18820
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.935834255606179e-06,
+      "loss": 0.3164,
+      "step": 18821
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.934859032372653e-06,
+      "loss": 0.2282,
+      "step": 18822
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.933883829654388e-06,
+      "loss": 0.1933,
+      "step": 18823
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.932908647461074e-06,
+      "loss": 0.2091,
+      "step": 18824
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.93193348580239e-06,
+      "loss": 0.2304,
+      "step": 18825
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.930958344688031e-06,
+      "loss": 0.2905,
+      "step": 18826
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92998322412768e-06,
+      "loss": 0.2042,
+      "step": 18827
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.929008124131028e-06,
+      "loss": 0.2566,
+      "step": 18828
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.928033044707764e-06,
+      "loss": 0.1696,
+      "step": 18829
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.927057985867563e-06,
+      "loss": 0.2716,
+      "step": 18830
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92608294762012e-06,
+      "loss": 0.2897,
+      "step": 18831
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92510792997512e-06,
+      "loss": 0.2667,
+      "step": 18832
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.924132932942251e-06,
+      "loss": 0.2319,
+      "step": 18833
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.923157956531193e-06,
+      "loss": 0.2461,
+      "step": 18834
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.922183000751635e-06,
+      "loss": 0.2402,
+      "step": 18835
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92120806561326e-06,
+      "loss": 0.2595,
+      "step": 18836
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.920233151125756e-06,
+      "loss": 0.2106,
+      "step": 18837
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.919258257298813e-06,
+      "loss": 0.2708,
+      "step": 18838
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.918283384142102e-06,
+      "loss": 0.1917,
+      "step": 18839
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.917308531665317e-06,
+      "loss": 0.195,
+      "step": 18840
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91633369987814e-06,
+      "loss": 0.2453,
+      "step": 18841
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.915358888790254e-06,
+      "loss": 0.1671,
+      "step": 18842
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.914384098411349e-06,
+      "loss": 0.2513,
+      "step": 18843
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.913409328751099e-06,
+      "loss": 0.2526,
+      "step": 18844
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.912434579819191e-06,
+      "loss": 0.2375,
+      "step": 18845
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91145985162531e-06,
+      "loss": 0.2173,
+      "step": 18846
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91048514417914e-06,
+      "loss": 0.2194,
+      "step": 18847
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.909510457490366e-06,
+      "loss": 0.2824,
+      "step": 18848
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90853579156866e-06,
+      "loss": 0.2347,
+      "step": 18849
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.907561146423716e-06,
+      "loss": 0.2785,
+      "step": 18850
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90658652206521e-06,
+      "loss": 0.2648,
+      "step": 18851
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.905611918502822e-06,
+      "loss": 0.2073,
+      "step": 18852
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.904637335746244e-06,
+      "loss": 0.2244,
+      "step": 18853
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.903662773805143e-06,
+      "loss": 0.2073,
+      "step": 18854
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.902688232689212e-06,
+      "loss": 0.1791,
+      "step": 18855
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.901713712408127e-06,
+      "loss": 0.2378,
+      "step": 18856
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.900739212971571e-06,
+      "loss": 0.2888,
+      "step": 18857
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.899764734389227e-06,
+      "loss": 0.2427,
+      "step": 18858
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.898790276670768e-06,
+      "loss": 0.2179,
+      "step": 18859
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.897815839825881e-06,
+      "loss": 0.2176,
+      "step": 18860
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.896841423864242e-06,
+      "loss": 0.2191,
+      "step": 18861
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.895867028795539e-06,
+      "loss": 0.1985,
+      "step": 18862
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.894892654629438e-06,
+      "loss": 0.153,
+      "step": 18863
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.893918301375629e-06,
+      "loss": 0.1969,
+      "step": 18864
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.892943969043787e-06,
+      "loss": 0.201,
+      "step": 18865
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.891969657643594e-06,
+      "loss": 0.2692,
+      "step": 18866
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.890995367184731e-06,
+      "loss": 0.2325,
+      "step": 18867
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.890021097676868e-06,
+      "loss": 0.2766,
+      "step": 18868
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.88904684912969e-06,
+      "loss": 0.2827,
+      "step": 18869
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.888072621552872e-06,
+      "loss": 0.2738,
+      "step": 18870
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.887098414956093e-06,
+      "loss": 0.205,
+      "step": 18871
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.886124229349037e-06,
+      "loss": 0.245,
+      "step": 18872
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.885150064741375e-06,
+      "loss": 0.303,
+      "step": 18873
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.884175921142782e-06,
+      "loss": 0.2739,
+      "step": 18874
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.88320179856294e-06,
+      "loss": 0.2626,
+      "step": 18875
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.882227697011526e-06,
+      "loss": 0.2284,
+      "step": 18876
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.881253616498218e-06,
+      "loss": 0.2178,
+      "step": 18877
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.880279557032687e-06,
+      "loss": 0.2263,
+      "step": 18878
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.879305518624612e-06,
+      "loss": 0.298,
+      "step": 18879
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.87833150128367e-06,
+      "loss": 0.2245,
+      "step": 18880
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.877357505019538e-06,
+      "loss": 0.2082,
+      "step": 18881
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.876383529841892e-06,
+      "loss": 0.2389,
+      "step": 18882
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.875409575760406e-06,
+      "loss": 0.2136,
+      "step": 18883
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.874435642784753e-06,
+      "loss": 0.2378,
+      "step": 18884
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.873461730924612e-06,
+      "loss": 0.2277,
+      "step": 18885
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.872487840189654e-06,
+      "loss": 0.2522,
+      "step": 18886
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.871513970589562e-06,
+      "loss": 0.2133,
+      "step": 18887
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.870540122133999e-06,
+      "loss": 0.2178,
+      "step": 18888
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.869566294832645e-06,
+      "loss": 0.171,
+      "step": 18889
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.868592488695175e-06,
+      "loss": 0.1837,
+      "step": 18890
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.867618703731266e-06,
+      "loss": 0.2359,
+      "step": 18891
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.866644939950583e-06,
+      "loss": 0.2174,
+      "step": 18892
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.865671197362806e-06,
+      "loss": 0.2034,
+      "step": 18893
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.864697475977604e-06,
+      "loss": 0.2142,
+      "step": 18894
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.863723775804651e-06,
+      "loss": 0.262,
+      "step": 18895
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.862750096853629e-06,
+      "loss": 0.1945,
+      "step": 18896
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.861776439134193e-06,
+      "loss": 0.2408,
+      "step": 18897
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.860802802656029e-06,
+      "loss": 0.295,
+      "step": 18898
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.859829187428806e-06,
+      "loss": 0.2276,
+      "step": 18899
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.858855593462193e-06,
+      "loss": 0.2107,
+      "step": 18900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.857882020765869e-06,
+      "loss": 0.2198,
+      "step": 18901
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.856908469349495e-06,
+      "loss": 0.2462,
+      "step": 18902
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.85593493922275e-06,
+      "loss": 0.2299,
+      "step": 18903
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.8549614303953e-06,
+      "loss": 0.2557,
+      "step": 18904
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.85398794287682e-06,
+      "loss": 0.2067,
+      "step": 18905
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.853014476676982e-06,
+      "loss": 0.196,
+      "step": 18906
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.852041031805451e-06,
+      "loss": 0.3071,
+      "step": 18907
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.851067608271903e-06,
+      "loss": 0.2772,
+      "step": 18908
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.850094206086003e-06,
+      "loss": 0.2647,
+      "step": 18909
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.849120825257421e-06,
+      "loss": 0.2556,
+      "step": 18910
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.848147465795837e-06,
+      "loss": 0.1973,
+      "step": 18911
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.847174127710905e-06,
+      "loss": 0.2576,
+      "step": 18912
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.846200811012301e-06,
+      "loss": 0.228,
+      "step": 18913
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.845227515709697e-06,
+      "loss": 0.2251,
+      "step": 18914
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.84425424181276e-06,
+      "loss": 0.1986,
+      "step": 18915
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.843280989331159e-06,
+      "loss": 0.2487,
+      "step": 18916
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.842307758274559e-06,
+      "loss": 0.2408,
+      "step": 18917
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.841334548652629e-06,
+      "loss": 0.2323,
+      "step": 18918
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.840361360475039e-06,
+      "loss": 0.2006,
+      "step": 18919
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.839388193751459e-06,
+      "loss": 0.1937,
+      "step": 18920
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.83841504849155e-06,
+      "loss": 0.2302,
+      "step": 18921
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.83744192470498e-06,
+      "loss": 0.2444,
+      "step": 18922
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.836468822401423e-06,
+      "loss": 0.2299,
+      "step": 18923
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.835495741590542e-06,
+      "loss": 0.2043,
+      "step": 18924
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.834522682282003e-06,
+      "loss": 0.1532,
+      "step": 18925
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.833549644485472e-06,
+      "loss": 0.237,
+      "step": 18926
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.832576628210614e-06,
+      "loss": 0.1773,
+      "step": 18927
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.831603633467101e-06,
+      "loss": 0.1813,
+      "step": 18928
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.83063066026459e-06,
+      "loss": 0.2427,
+      "step": 18929
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.829657708612759e-06,
+      "loss": 0.1959,
+      "step": 18930
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.828684778521258e-06,
+      "loss": 0.2709,
+      "step": 18931
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.827711869999764e-06,
+      "loss": 0.2071,
+      "step": 18932
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.826738983057936e-06,
+      "loss": 0.2086,
+      "step": 18933
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82576611770544e-06,
+      "loss": 0.2843,
+      "step": 18934
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.824793273951945e-06,
+      "loss": 0.2561,
+      "step": 18935
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.823820451807109e-06,
+      "loss": 0.1987,
+      "step": 18936
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.822847651280598e-06,
+      "loss": 0.2659,
+      "step": 18937
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.821874872382076e-06,
+      "loss": 0.2759,
+      "step": 18938
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.820902115121207e-06,
+      "loss": 0.2724,
+      "step": 18939
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.819929379507658e-06,
+      "loss": 0.2265,
+      "step": 18940
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.818956665551085e-06,
+      "loss": 0.2035,
+      "step": 18941
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.817983973261157e-06,
+      "loss": 0.21,
+      "step": 18942
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.817011302647532e-06,
+      "loss": 0.2548,
+      "step": 18943
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.816038653719874e-06,
+      "loss": 0.2804,
+      "step": 18944
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.815066026487855e-06,
+      "loss": 0.1931,
+      "step": 18945
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.814093420961122e-06,
+      "loss": 0.2666,
+      "step": 18946
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.813120837149341e-06,
+      "loss": 0.2186,
+      "step": 18947
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.81214827506218e-06,
+      "loss": 0.3023,
+      "step": 18948
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.811175734709298e-06,
+      "loss": 0.2384,
+      "step": 18949
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.810203216100353e-06,
+      "loss": 0.2362,
+      "step": 18950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.809230719245008e-06,
+      "loss": 0.2645,
+      "step": 18951
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.808258244152925e-06,
+      "loss": 0.2515,
+      "step": 18952
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.80728579083376e-06,
+      "loss": 0.2345,
+      "step": 18953
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.806313359297184e-06,
+      "loss": 0.3236,
+      "step": 18954
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.805340949552846e-06,
+      "loss": 0.2485,
+      "step": 18955
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.804368561610407e-06,
+      "loss": 0.2576,
+      "step": 18956
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.803396195479534e-06,
+      "loss": 0.1516,
+      "step": 18957
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.802423851169881e-06,
+      "loss": 0.1971,
+      "step": 18958
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.801451528691114e-06,
+      "loss": 0.1995,
+      "step": 18959
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.800479228052883e-06,
+      "loss": 0.2308,
+      "step": 18960
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.799506949264849e-06,
+      "loss": 0.2457,
+      "step": 18961
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.798534692336672e-06,
+      "loss": 0.2942,
+      "step": 18962
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.797562457278013e-06,
+      "loss": 0.2032,
+      "step": 18963
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.796590244098533e-06,
+      "loss": 0.1623,
+      "step": 18964
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.795618052807879e-06,
+      "loss": 0.1933,
+      "step": 18965
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.794645883415717e-06,
+      "loss": 0.2126,
+      "step": 18966
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.793673735931704e-06,
+      "loss": 0.2145,
+      "step": 18967
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.792701610365494e-06,
+      "loss": 0.208,
+      "step": 18968
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.79172950672675e-06,
+      "loss": 0.2591,
+      "step": 18969
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.790757425025123e-06,
+      "loss": 0.1462,
+      "step": 18970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.78978536527027e-06,
+      "loss": 0.2395,
+      "step": 18971
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.788813327471851e-06,
+      "loss": 0.2331,
+      "step": 18972
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.787841311639526e-06,
+      "loss": 0.23,
+      "step": 18973
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.786869317782938e-06,
+      "loss": 0.2358,
+      "step": 18974
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.785897345911755e-06,
+      "loss": 0.2137,
+      "step": 18975
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.784925396035627e-06,
+      "loss": 0.2183,
+      "step": 18976
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.783953468164212e-06,
+      "loss": 0.2247,
+      "step": 18977
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.782981562307163e-06,
+      "loss": 0.2366,
+      "step": 18978
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.782009678474137e-06,
+      "loss": 0.2379,
+      "step": 18979
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.781037816674787e-06,
+      "loss": 0.2286,
+      "step": 18980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.780065976918766e-06,
+      "loss": 0.2435,
+      "step": 18981
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.779094159215733e-06,
+      "loss": 0.2478,
+      "step": 18982
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.778122363575343e-06,
+      "loss": 0.2003,
+      "step": 18983
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.777150590007243e-06,
+      "loss": 0.2269,
+      "step": 18984
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.77617883852109e-06,
+      "loss": 0.2142,
+      "step": 18985
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.775207109126538e-06,
+      "loss": 0.2478,
+      "step": 18986
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.774235401833238e-06,
+      "loss": 0.1864,
+      "step": 18987
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.773263716650852e-06,
+      "loss": 0.2282,
+      "step": 18988
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.772292053589022e-06,
+      "loss": 0.2184,
+      "step": 18989
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.7713204126574e-06,
+      "loss": 0.2471,
+      "step": 18990
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.770348793865647e-06,
+      "loss": 0.2221,
+      "step": 18991
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.769377197223408e-06,
+      "loss": 0.1734,
+      "step": 18992
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.768405622740343e-06,
+      "loss": 0.2349,
+      "step": 18993
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.767434070426095e-06,
+      "loss": 0.2416,
+      "step": 18994
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76646254029032e-06,
+      "loss": 0.1837,
+      "step": 18995
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.765491032342665e-06,
+      "loss": 0.2432,
+      "step": 18996
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.764519546592785e-06,
+      "loss": 0.1725,
+      "step": 18997
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.763548083050336e-06,
+      "loss": 0.2306,
+      "step": 18998
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.762576641724956e-06,
+      "loss": 0.3086,
+      "step": 18999
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.761605222626306e-06,
+      "loss": 0.2038,
+      "step": 19000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76063382576403e-06,
+      "loss": 0.1952,
+      "step": 19001
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.759662451147782e-06,
+      "loss": 0.2266,
+      "step": 19002
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.758691098787207e-06,
+      "loss": 0.2406,
+      "step": 19003
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.757719768691958e-06,
+      "loss": 0.2288,
+      "step": 19004
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.756748460871681e-06,
+      "loss": 0.2449,
+      "step": 19005
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.755777175336032e-06,
+      "loss": 0.2137,
+      "step": 19006
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.754805912094657e-06,
+      "loss": 0.2346,
+      "step": 19007
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.753834671157198e-06,
+      "loss": 0.305,
+      "step": 19008
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.752863452533309e-06,
+      "loss": 0.2374,
+      "step": 19009
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.751892256232638e-06,
+      "loss": 0.1947,
+      "step": 19010
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.750921082264832e-06,
+      "loss": 0.2351,
+      "step": 19011
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.749949930639544e-06,
+      "loss": 0.2227,
+      "step": 19012
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.748978801366412e-06,
+      "loss": 0.2682,
+      "step": 19013
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.748007694455087e-06,
+      "loss": 0.2217,
+      "step": 19014
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.747036609915217e-06,
+      "loss": 0.2195,
+      "step": 19015
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.746065547756452e-06,
+      "loss": 0.3047,
+      "step": 19016
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.745094507988435e-06,
+      "loss": 0.2601,
+      "step": 19017
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.744123490620811e-06,
+      "loss": 0.2027,
+      "step": 19018
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74315249566323e-06,
+      "loss": 0.2297,
+      "step": 19019
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.742181523125332e-06,
+      "loss": 0.2663,
+      "step": 19020
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.741210573016766e-06,
+      "loss": 0.2801,
+      "step": 19021
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.740239645347185e-06,
+      "loss": 0.312,
+      "step": 19022
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.739268740126222e-06,
+      "loss": 0.2309,
+      "step": 19023
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.738297857363527e-06,
+      "loss": 0.2,
+      "step": 19024
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.737326997068748e-06,
+      "loss": 0.2766,
+      "step": 19025
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.736356159251524e-06,
+      "loss": 0.281,
+      "step": 19026
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.735385343921507e-06,
+      "loss": 0.2581,
+      "step": 19027
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.734414551088331e-06,
+      "loss": 0.2221,
+      "step": 19028
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.733443780761647e-06,
+      "loss": 0.221,
+      "step": 19029
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.732473032951095e-06,
+      "loss": 0.2149,
+      "step": 19030
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.731502307666328e-06,
+      "loss": 0.2313,
+      "step": 19031
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.730531604916973e-06,
+      "loss": 0.2472,
+      "step": 19032
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.729560924712684e-06,
+      "loss": 0.1719,
+      "step": 19033
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.728590267063102e-06,
+      "loss": 0.18,
+      "step": 19034
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.727619631977869e-06,
+      "loss": 0.2567,
+      "step": 19035
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72664901946663e-06,
+      "loss": 0.2426,
+      "step": 19036
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.725678429539021e-06,
+      "loss": 0.2125,
+      "step": 19037
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72470786220469e-06,
+      "loss": 0.2922,
+      "step": 19038
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.723737317473273e-06,
+      "loss": 0.2339,
+      "step": 19039
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.722766795354415e-06,
+      "loss": 0.2608,
+      "step": 19040
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.721796295857762e-06,
+      "loss": 0.2483,
+      "step": 19041
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.720825818992945e-06,
+      "loss": 0.1927,
+      "step": 19042
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.719855364769611e-06,
+      "loss": 0.2625,
+      "step": 19043
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.718884933197399e-06,
+      "loss": 0.2196,
+      "step": 19044
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71791452428595e-06,
+      "loss": 0.2821,
+      "step": 19045
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.716944138044906e-06,
+      "loss": 0.2076,
+      "step": 19046
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.715973774483901e-06,
+      "loss": 0.2906,
+      "step": 19047
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.715003433612578e-06,
+      "loss": 0.219,
+      "step": 19048
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.714033115440578e-06,
+      "loss": 0.2039,
+      "step": 19049
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.713062819977538e-06,
+      "loss": 0.267,
+      "step": 19050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.712092547233101e-06,
+      "loss": 0.2077,
+      "step": 19051
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.7111222972169e-06,
+      "loss": 0.196,
+      "step": 19052
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.710152069938577e-06,
+      "loss": 0.1346,
+      "step": 19053
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.709181865407767e-06,
+      "loss": 0.2176,
+      "step": 19054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.708211683634112e-06,
+      "loss": 0.2319,
+      "step": 19055
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.707241524627252e-06,
+      "loss": 0.1667,
+      "step": 19056
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.706271388396816e-06,
+      "loss": 0.1979,
+      "step": 19057
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.705301274952448e-06,
+      "loss": 0.2207,
+      "step": 19058
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.704331184303783e-06,
+      "loss": 0.1742,
+      "step": 19059
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.703361116460462e-06,
+      "loss": 0.2473,
+      "step": 19060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.702391071432116e-06,
+      "loss": 0.2413,
+      "step": 19061
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.701421049228384e-06,
+      "loss": 0.274,
+      "step": 19062
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.700451049858902e-06,
+      "loss": 0.2144,
+      "step": 19063
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.699481073333303e-06,
+      "loss": 0.2448,
+      "step": 19064
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.698511119661233e-06,
+      "loss": 0.2716,
+      "step": 19065
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.697541188852314e-06,
+      "loss": 0.3165,
+      "step": 19066
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.696571280916189e-06,
+      "loss": 0.215,
+      "step": 19067
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.695601395862494e-06,
+      "loss": 0.252,
+      "step": 19068
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.69463153370086e-06,
+      "loss": 0.2712,
+      "step": 19069
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.693661694440926e-06,
+      "loss": 0.2194,
+      "step": 19070
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.692691878092323e-06,
+      "loss": 0.2422,
+      "step": 19071
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.691722084664685e-06,
+      "loss": 0.2114,
+      "step": 19072
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.690752314167647e-06,
+      "loss": 0.2155,
+      "step": 19073
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.689782566610845e-06,
+      "loss": 0.2138,
+      "step": 19074
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.68881284200391e-06,
+      "loss": 0.1927,
+      "step": 19075
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.687843140356477e-06,
+      "loss": 0.1918,
+      "step": 19076
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.686873461678176e-06,
+      "loss": 0.1899,
+      "step": 19077
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.685903805978643e-06,
+      "loss": 0.2298,
+      "step": 19078
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.68493417326751e-06,
+      "loss": 0.1878,
+      "step": 19079
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.68396456355441e-06,
+      "loss": 0.2442,
+      "step": 19080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.68299497684897e-06,
+      "loss": 0.2701,
+      "step": 19081
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.682025413160827e-06,
+      "loss": 0.3033,
+      "step": 19082
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.681055872499611e-06,
+      "loss": 0.2082,
+      "step": 19083
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.680086354874956e-06,
+      "loss": 0.2376,
+      "step": 19084
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.679116860296493e-06,
+      "loss": 0.2159,
+      "step": 19085
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67814738877385e-06,
+      "loss": 0.2426,
+      "step": 19086
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67717794031666e-06,
+      "loss": 0.2509,
+      "step": 19087
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67620851493455e-06,
+      "loss": 0.2118,
+      "step": 19088
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.675239112637158e-06,
+      "loss": 0.2628,
+      "step": 19089
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.674269733434108e-06,
+      "loss": 0.1776,
+      "step": 19090
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.673300377335026e-06,
+      "loss": 0.2005,
+      "step": 19091
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.672331044349552e-06,
+      "loss": 0.1559,
+      "step": 19092
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.671361734487308e-06,
+      "loss": 0.213,
+      "step": 19093
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.670392447757928e-06,
+      "loss": 0.2292,
+      "step": 19094
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.669423184171037e-06,
+      "loss": 0.221,
+      "step": 19095
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.668453943736265e-06,
+      "loss": 0.2339,
+      "step": 19096
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66748472646324e-06,
+      "loss": 0.2434,
+      "step": 19097
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66651553236159e-06,
+      "loss": 0.3159,
+      "step": 19098
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66554636144095e-06,
+      "loss": 0.2512,
+      "step": 19099
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.664577213710934e-06,
+      "loss": 0.2326,
+      "step": 19100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.663608089181181e-06,
+      "loss": 0.2699,
+      "step": 19101
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.662638987861315e-06,
+      "loss": 0.2279,
+      "step": 19102
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.661669909760962e-06,
+      "loss": 0.2586,
+      "step": 19103
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.660700854889751e-06,
+      "loss": 0.2371,
+      "step": 19104
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.659731823257309e-06,
+      "loss": 0.1893,
+      "step": 19105
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.658762814873257e-06,
+      "loss": 0.2299,
+      "step": 19106
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.657793829747225e-06,
+      "loss": 0.3377,
+      "step": 19107
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.65682486788884e-06,
+      "loss": 0.2405,
+      "step": 19108
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.655855929307728e-06,
+      "loss": 0.2284,
+      "step": 19109
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.654887014013513e-06,
+      "loss": 0.2563,
+      "step": 19110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.65391812201582e-06,
+      "loss": 0.1876,
+      "step": 19111
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.652949253324275e-06,
+      "loss": 0.2503,
+      "step": 19112
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6519804079485e-06,
+      "loss": 0.2095,
+      "step": 19113
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.651011585898128e-06,
+      "loss": 0.2588,
+      "step": 19114
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.650042787182773e-06,
+      "loss": 0.254,
+      "step": 19115
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64907401181206e-06,
+      "loss": 0.2383,
+      "step": 19116
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.648105259795623e-06,
+      "loss": 0.2431,
+      "step": 19117
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.647136531143077e-06,
+      "loss": 0.1988,
+      "step": 19118
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.646167825864047e-06,
+      "loss": 0.1819,
+      "step": 19119
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.645199143968155e-06,
+      "loss": 0.2062,
+      "step": 19120
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.644230485465027e-06,
+      "loss": 0.2091,
+      "step": 19121
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64326185036428e-06,
+      "loss": 0.2402,
+      "step": 19122
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64229323867555e-06,
+      "loss": 0.2202,
+      "step": 19123
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.641324650408445e-06,
+      "loss": 0.1775,
+      "step": 19124
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64035608557259e-06,
+      "loss": 0.1547,
+      "step": 19125
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.63938754417761e-06,
+      "loss": 0.3198,
+      "step": 19126
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.638419026233125e-06,
+      "loss": 0.2229,
+      "step": 19127
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.63745053174876e-06,
+      "loss": 0.2248,
+      "step": 19128
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.63648206073413e-06,
+      "loss": 0.2297,
+      "step": 19129
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.635513613198859e-06,
+      "loss": 0.1522,
+      "step": 19130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.634545189152565e-06,
+      "loss": 0.2275,
+      "step": 19131
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.633576788604874e-06,
+      "loss": 0.31,
+      "step": 19132
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.632608411565407e-06,
+      "loss": 0.2089,
+      "step": 19133
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.631640058043772e-06,
+      "loss": 0.2711,
+      "step": 19134
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6306717280496e-06,
+      "loss": 0.2413,
+      "step": 19135
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.629703421592505e-06,
+      "loss": 0.2242,
+      "step": 19136
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.628735138682109e-06,
+      "loss": 0.1833,
+      "step": 19137
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.627766879328034e-06,
+      "loss": 0.2502,
+      "step": 19138
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.626798643539891e-06,
+      "loss": 0.2255,
+      "step": 19139
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6258304313273e-06,
+      "loss": 0.2606,
+      "step": 19140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.624862242699885e-06,
+      "loss": 0.2291,
+      "step": 19141
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.623894077667264e-06,
+      "loss": 0.2199,
+      "step": 19142
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6229259362390475e-06,
+      "loss": 0.2488,
+      "step": 19143
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.621957818424857e-06,
+      "loss": 0.2005,
+      "step": 19144
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.620989724234311e-06,
+      "loss": 0.1994,
+      "step": 19145
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.620021653677025e-06,
+      "loss": 0.2491,
+      "step": 19146
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6190536067626194e-06,
+      "loss": 0.19,
+      "step": 19147
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6180855835007064e-06,
+      "loss": 0.225,
+      "step": 19148
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.617117583900903e-06,
+      "loss": 0.2411,
+      "step": 19149
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.616149607972825e-06,
+      "loss": 0.232,
+      "step": 19150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.615181655726091e-06,
+      "loss": 0.2249,
+      "step": 19151
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.614213727170317e-06,
+      "loss": 0.1775,
+      "step": 19152
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.613245822315116e-06,
+      "loss": 0.1703,
+      "step": 19153
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.612277941170106e-06,
+      "loss": 0.266,
+      "step": 19154
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.611310083744897e-06,
+      "loss": 0.2443,
+      "step": 19155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.610342250049105e-06,
+      "loss": 0.2329,
+      "step": 19156
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6093744400923545e-06,
+      "loss": 0.1799,
+      "step": 19157
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6084066538842475e-06,
+      "loss": 0.2309,
+      "step": 19158
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.607438891434399e-06,
+      "loss": 0.2154,
+      "step": 19159
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.606471152752428e-06,
+      "loss": 0.1734,
+      "step": 19160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.605503437847948e-06,
+      "loss": 0.2572,
+      "step": 19161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6045357467305715e-06,
+      "loss": 0.1986,
+      "step": 19162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.603568079409908e-06,
+      "loss": 0.2326,
+      "step": 19163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.602600435895574e-06,
+      "loss": 0.2027,
+      "step": 19164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.601632816197179e-06,
+      "loss": 0.1855,
+      "step": 19165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6006652203243395e-06,
+      "loss": 0.2267,
+      "step": 19166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.59969764828667e-06,
+      "loss": 0.174,
+      "step": 19167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.598730100093772e-06,
+      "loss": 0.2371,
+      "step": 19168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.597762575755265e-06,
+      "loss": 0.2816,
+      "step": 19169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.59679507528076e-06,
+      "loss": 0.2729,
+      "step": 19170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.595827598679868e-06,
+      "loss": 0.2316,
+      "step": 19171
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.594860145962198e-06,
+      "loss": 0.196,
+      "step": 19172
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.593892717137361e-06,
+      "loss": 0.1782,
+      "step": 19173
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.592925312214967e-06,
+      "loss": 0.2261,
+      "step": 19174
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.591957931204629e-06,
+      "loss": 0.2051,
+      "step": 19175
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.590990574115959e-06,
+      "loss": 0.2194,
+      "step": 19176
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.590023240958558e-06,
+      "loss": 0.2303,
+      "step": 19177
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.589055931742044e-06,
+      "loss": 0.2793,
+      "step": 19178
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5880886464760214e-06,
+      "loss": 0.2389,
+      "step": 19179
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.587121385170103e-06,
+      "loss": 0.2275,
+      "step": 19180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.586154147833897e-06,
+      "loss": 0.2215,
+      "step": 19181
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.585186934477009e-06,
+      "loss": 0.251,
+      "step": 19182
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.584219745109047e-06,
+      "loss": 0.2958,
+      "step": 19183
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.583252579739623e-06,
+      "loss": 0.2847,
+      "step": 19184
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.582285438378344e-06,
+      "loss": 0.2493,
+      "step": 19185
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.581318321034819e-06,
+      "loss": 0.2315,
+      "step": 19186
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5803512277186495e-06,
+      "loss": 0.2209,
+      "step": 19187
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.579384158439448e-06,
+      "loss": 0.2453,
+      "step": 19188
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57841711320682e-06,
+      "loss": 0.2915,
+      "step": 19189
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.577450092030369e-06,
+      "loss": 0.1785,
+      "step": 19190
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.576483094919711e-06,
+      "loss": 0.1644,
+      "step": 19191
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57551612188444e-06,
+      "loss": 0.1921,
+      "step": 19192
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.574549172934169e-06,
+      "loss": 0.3097,
+      "step": 19193
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.573582248078504e-06,
+      "loss": 0.2397,
+      "step": 19194
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.572615347327048e-06,
+      "loss": 0.2305,
+      "step": 19195
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57164847068941e-06,
+      "loss": 0.2184,
+      "step": 19196
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.570681618175189e-06,
+      "loss": 0.2046,
+      "step": 19197
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.569714789793994e-06,
+      "loss": 0.1951,
+      "step": 19198
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.568747985555426e-06,
+      "loss": 0.2068,
+      "step": 19199
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.567781205469098e-06,
+      "loss": 0.2418,
+      "step": 19200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.566814449544602e-06,
+      "loss": 0.1836,
+      "step": 19201
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.56584771779155e-06,
+      "loss": 0.2225,
+      "step": 19202
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.564881010219544e-06,
+      "loss": 0.2329,
+      "step": 19203
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.563914326838185e-06,
+      "loss": 0.2337,
+      "step": 19204
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.562947667657081e-06,
+      "loss": 0.2444,
+      "step": 19205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.56198103268583e-06,
+      "loss": 0.219,
+      "step": 19206
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.561014421934036e-06,
+      "loss": 0.2174,
+      "step": 19207
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.560047835411299e-06,
+      "loss": 0.2417,
+      "step": 19208
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.559081273127226e-06,
+      "loss": 0.2834,
+      "step": 19209
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.558114735091422e-06,
+      "loss": 0.2022,
+      "step": 19210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.557148221313476e-06,
+      "loss": 0.2656,
+      "step": 19211
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.556181731802999e-06,
+      "loss": 0.1938,
+      "step": 19212
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.55521526656959e-06,
+      "loss": 0.2567,
+      "step": 19213
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.55424882562285e-06,
+      "loss": 0.2024,
+      "step": 19214
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.553282408972382e-06,
+      "loss": 0.1855,
+      "step": 19215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.552316016627782e-06,
+      "loss": 0.216,
+      "step": 19216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5513496485986496e-06,
+      "loss": 0.1696,
+      "step": 19217
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.55038330489459e-06,
+      "loss": 0.2329,
+      "step": 19218
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5494169855252e-06,
+      "loss": 0.2664,
+      "step": 19219
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.548450690500082e-06,
+      "loss": 0.2481,
+      "step": 19220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.54748441982883e-06,
+      "loss": 0.2655,
+      "step": 19221
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.546518173521047e-06,
+      "loss": 0.1636,
+      "step": 19222
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.54555195158633e-06,
+      "loss": 0.2505,
+      "step": 19223
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.544585754034275e-06,
+      "loss": 0.2063,
+      "step": 19224
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.543619580874489e-06,
+      "loss": 0.2577,
+      "step": 19225
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5426534321165605e-06,
+      "loss": 0.2339,
+      "step": 19226
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.54168730777009e-06,
+      "loss": 0.2814,
+      "step": 19227
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.540721207844679e-06,
+      "loss": 0.184,
+      "step": 19228
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.539755132349922e-06,
+      "loss": 0.2065,
+      "step": 19229
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.538789081295415e-06,
+      "loss": 0.2122,
+      "step": 19230
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.537823054690756e-06,
+      "loss": 0.2432,
+      "step": 19231
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.536857052545541e-06,
+      "loss": 0.2506,
+      "step": 19232
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.535891074869364e-06,
+      "loss": 0.2071,
+      "step": 19233
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.534925121671831e-06,
+      "loss": 0.2468,
+      "step": 19234
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.533959192962525e-06,
+      "loss": 0.2044,
+      "step": 19235
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.532993288751048e-06,
+      "loss": 0.2294,
+      "step": 19236
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.532027409046995e-06,
+      "loss": 0.2715,
+      "step": 19237
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.531061553859961e-06,
+      "loss": 0.2857,
+      "step": 19238
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.530095723199543e-06,
+      "loss": 0.2474,
+      "step": 19239
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5291299170753305e-06,
+      "loss": 0.2269,
+      "step": 19240
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.528164135496921e-06,
+      "loss": 0.1798,
+      "step": 19241
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.527198378473907e-06,
+      "loss": 0.3062,
+      "step": 19242
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5262326460158845e-06,
+      "loss": 0.227,
+      "step": 19243
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.52526693813245e-06,
+      "loss": 0.203,
+      "step": 19244
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.52430125483319e-06,
+      "loss": 0.3172,
+      "step": 19245
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.523335596127701e-06,
+      "loss": 0.2329,
+      "step": 19246
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.522369962025576e-06,
+      "loss": 0.1886,
+      "step": 19247
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.521404352536407e-06,
+      "loss": 0.1969,
+      "step": 19248
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.52043876766979e-06,
+      "loss": 0.2245,
+      "step": 19249
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.519473207435312e-06,
+      "loss": 0.2288,
+      "step": 19250
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5185076718425645e-06,
+      "loss": 0.2767,
+      "step": 19251
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.517542160901143e-06,
+      "loss": 0.2038,
+      "step": 19252
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.516576674620639e-06,
+      "loss": 0.2311,
+      "step": 19253
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.515611213010645e-06,
+      "loss": 0.2304,
+      "step": 19254
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.514645776080747e-06,
+      "loss": 0.2722,
+      "step": 19255
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5136803638405385e-06,
+      "loss": 0.2488,
+      "step": 19256
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5127149762996085e-06,
+      "loss": 0.2214,
+      "step": 19257
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.511749613467552e-06,
+      "loss": 0.2436,
+      "step": 19258
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.510784275353953e-06,
+      "loss": 0.2039,
+      "step": 19259
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5098189619684e-06,
+      "loss": 0.2804,
+      "step": 19260
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.508853673320489e-06,
+      "loss": 0.2513,
+      "step": 19261
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.507888409419807e-06,
+      "loss": 0.2389,
+      "step": 19262
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.506923170275944e-06,
+      "loss": 0.2589,
+      "step": 19263
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.505957955898484e-06,
+      "loss": 0.2408,
+      "step": 19264
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.50499276629702e-06,
+      "loss": 0.2282,
+      "step": 19265
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5040276014811386e-06,
+      "loss": 0.2845,
+      "step": 19266
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.503062461460426e-06,
+      "loss": 0.2509,
+      "step": 19267
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.502097346244478e-06,
+      "loss": 0.1873,
+      "step": 19268
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5011322558428705e-06,
+      "loss": 0.2581,
+      "step": 19269
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.500167190265197e-06,
+      "loss": 0.3392,
+      "step": 19270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.499202149521046e-06,
+      "loss": 0.2299,
+      "step": 19271
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.498237133620001e-06,
+      "loss": 0.1889,
+      "step": 19272
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4972721425716525e-06,
+      "loss": 0.2687,
+      "step": 19273
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.496307176385582e-06,
+      "loss": 0.2077,
+      "step": 19274
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.495342235071377e-06,
+      "loss": 0.2252,
+      "step": 19275
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.494377318638623e-06,
+      "loss": 0.2364,
+      "step": 19276
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.493412427096908e-06,
+      "loss": 0.2869,
+      "step": 19277
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.492447560455817e-06,
+      "loss": 0.2369,
+      "step": 19278
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4914827187249335e-06,
+      "loss": 0.2738,
+      "step": 19279
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.490517901913841e-06,
+      "loss": 0.2556,
+      "step": 19280
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.489553110032126e-06,
+      "loss": 0.2229,
+      "step": 19281
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.488588343089376e-06,
+      "loss": 0.2,
+      "step": 19282
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.487623601095169e-06,
+      "loss": 0.227,
+      "step": 19283
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.48665888405909e-06,
+      "loss": 0.2226,
+      "step": 19284
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.485694191990724e-06,
+      "loss": 0.2467,
+      "step": 19285
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4847295248996545e-06,
+      "loss": 0.2465,
+      "step": 19286
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.483764882795468e-06,
+      "loss": 0.2445,
+      "step": 19287
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.482800265687741e-06,
+      "loss": 0.2511,
+      "step": 19288
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.481835673586059e-06,
+      "loss": 0.1744,
+      "step": 19289
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4808711065000025e-06,
+      "loss": 0.2577,
+      "step": 19290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.479906564439154e-06,
+      "loss": 0.2631,
+      "step": 19291
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.478942047413105e-06,
+      "loss": 0.2439,
+      "step": 19292
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.477977555431421e-06,
+      "loss": 0.1803,
+      "step": 19293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.477013088503692e-06,
+      "loss": 0.2805,
+      "step": 19294
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.476048646639498e-06,
+      "loss": 0.2216,
+      "step": 19295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4750842298484205e-06,
+      "loss": 0.1969,
+      "step": 19296
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.474119838140041e-06,
+      "loss": 0.2112,
+      "step": 19297
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.473155471523937e-06,
+      "loss": 0.1845,
+      "step": 19298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.472191130009689e-06,
+      "loss": 0.2425,
+      "step": 19299
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.471226813606877e-06,
+      "loss": 0.2441,
+      "step": 19300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4702625223250824e-06,
+      "loss": 0.2243,
+      "step": 19301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.469298256173888e-06,
+      "loss": 0.2182,
+      "step": 19302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.468334015162861e-06,
+      "loss": 0.2157,
+      "step": 19303
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.467369799301591e-06,
+      "loss": 0.2352,
+      "step": 19304
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.466405608599653e-06,
+      "loss": 0.2129,
+      "step": 19305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.465441443066625e-06,
+      "loss": 0.2241,
+      "step": 19306
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.464477302712087e-06,
+      "loss": 0.2224,
+      "step": 19307
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.463513187545615e-06,
+      "loss": 0.2093,
+      "step": 19308
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.462549097576785e-06,
+      "loss": 0.2217,
+      "step": 19309
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4615850328151775e-06,
+      "loss": 0.2533,
+      "step": 19310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4606209932703725e-06,
+      "loss": 0.2861,
+      "step": 19311
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.459656978951937e-06,
+      "loss": 0.1988,
+      "step": 19312
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.458692989869454e-06,
+      "loss": 0.172,
+      "step": 19313
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4577290260325005e-06,
+      "loss": 0.2158,
+      "step": 19314
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.456765087450652e-06,
+      "loss": 0.2282,
+      "step": 19315
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.455801174133483e-06,
+      "loss": 0.2282,
+      "step": 19316
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.454837286090571e-06,
+      "loss": 0.201,
+      "step": 19317
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.453873423331488e-06,
+      "loss": 0.2679,
+      "step": 19318
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.45290958586581e-06,
+      "loss": 0.1764,
+      "step": 19319
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.451945773703114e-06,
+      "loss": 0.2203,
+      "step": 19320
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4509819868529766e-06,
+      "loss": 0.2475,
+      "step": 19321
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.450018225324966e-06,
+      "loss": 0.2044,
+      "step": 19322
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.449054489128661e-06,
+      "loss": 0.3885,
+      "step": 19323
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.448090778273633e-06,
+      "loss": 0.2545,
+      "step": 19324
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.447127092769453e-06,
+      "loss": 0.2404,
+      "step": 19325
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.446163432625706e-06,
+      "loss": 0.2608,
+      "step": 19326
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.445199797851952e-06,
+      "loss": 0.1906,
+      "step": 19327
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.444236188457765e-06,
+      "loss": 0.3139,
+      "step": 19328
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.443272604452725e-06,
+      "loss": 0.2045,
+      "step": 19329
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.442309045846399e-06,
+      "loss": 0.2106,
+      "step": 19330
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.441345512648363e-06,
+      "loss": 0.2643,
+      "step": 19331
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.440382004868185e-06,
+      "loss": 0.2492,
+      "step": 19332
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.439418522515438e-06,
+      "loss": 0.2268,
+      "step": 19333
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.43845506559969e-06,
+      "loss": 0.2418,
+      "step": 19334
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.437491634130518e-06,
+      "loss": 0.1936,
+      "step": 19335
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.436528228117493e-06,
+      "loss": 0.2193,
+      "step": 19336
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.435564847570178e-06,
+      "loss": 0.2623,
+      "step": 19337
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.434601492498148e-06,
+      "loss": 0.2136,
+      "step": 19338
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.433638162910974e-06,
+      "loss": 0.2801,
+      "step": 19339
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.432674858818227e-06,
+      "loss": 0.2309,
+      "step": 19340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.431711580229473e-06,
+      "loss": 0.2073,
+      "step": 19341
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4307483271542815e-06,
+      "loss": 0.197,
+      "step": 19342
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.42978509960222e-06,
+      "loss": 0.188,
+      "step": 19343
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.428821897582863e-06,
+      "loss": 0.2281,
+      "step": 19344
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4278587211057785e-06,
+      "loss": 0.2218,
+      "step": 19345
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.426895570180527e-06,
+      "loss": 0.2314,
+      "step": 19346
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.425932444816683e-06,
+      "loss": 0.2811,
+      "step": 19347
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.424969345023814e-06,
+      "loss": 0.2421,
+      "step": 19348
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4240062708114835e-06,
+      "loss": 0.2839,
+      "step": 19349
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4230432221892655e-06,
+      "loss": 0.2448,
+      "step": 19350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.422080199166721e-06,
+      "loss": 0.1833,
+      "step": 19351
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.421117201753418e-06,
+      "loss": 0.2153,
+      "step": 19352
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.420154229958925e-06,
+      "loss": 0.2732,
+      "step": 19353
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.419191283792806e-06,
+      "loss": 0.2132,
+      "step": 19354
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.418228363264631e-06,
+      "loss": 0.296,
+      "step": 19355
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.417265468383961e-06,
+      "loss": 0.2467,
+      "step": 19356
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.416302599160363e-06,
+      "loss": 0.2124,
+      "step": 19357
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.415339755603401e-06,
+      "loss": 0.3407,
+      "step": 19358
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4143769377226404e-06,
+      "loss": 0.2302,
+      "step": 19359
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.413414145527654e-06,
+      "loss": 0.2336,
+      "step": 19360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4124513790279916e-06,
+      "loss": 0.2702,
+      "step": 19361
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.411488638233229e-06,
+      "loss": 0.2072,
+      "step": 19362
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.410525923152924e-06,
+      "loss": 0.1791,
+      "step": 19363
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.409563233796643e-06,
+      "loss": 0.2275,
+      "step": 19364
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.408600570173951e-06,
+      "loss": 0.2431,
+      "step": 19365
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.407637932294407e-06,
+      "loss": 0.2457,
+      "step": 19366
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4066753201675745e-06,
+      "loss": 0.1582,
+      "step": 19367
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4057127338030165e-06,
+      "loss": 0.1905,
+      "step": 19368
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.404750173210302e-06,
+      "loss": 0.1817,
+      "step": 19369
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.403787638398985e-06,
+      "loss": 0.2828,
+      "step": 19370
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.402825129378627e-06,
+      "loss": 0.2201,
+      "step": 19371
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4018626461587954e-06,
+      "loss": 0.2283,
+      "step": 19372
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.400900188749047e-06,
+      "loss": 0.1803,
+      "step": 19373
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.399937757158948e-06,
+      "loss": 0.2148,
+      "step": 19374
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.398975351398054e-06,
+      "loss": 0.2499,
+      "step": 19375
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3980129714759265e-06,
+      "loss": 0.2987,
+      "step": 19376
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.397050617402125e-06,
+      "loss": 0.2517,
+      "step": 19377
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.396088289186214e-06,
+      "loss": 0.2845,
+      "step": 19378
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.395125986837756e-06,
+      "loss": 0.2238,
+      "step": 19379
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.394163710366298e-06,
+      "loss": 0.2041,
+      "step": 19380
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.393201459781409e-06,
+      "loss": 0.2688,
+      "step": 19381
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.392239235092646e-06,
+      "loss": 0.1763,
+      "step": 19382
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.391277036309568e-06,
+      "loss": 0.2697,
+      "step": 19383
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.390314863441734e-06,
+      "loss": 0.1976,
+      "step": 19384
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.389352716498701e-06,
+      "loss": 0.253,
+      "step": 19385
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.388390595490025e-06,
+      "loss": 0.1801,
+      "step": 19386
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.387428500425267e-06,
+      "loss": 0.2807,
+      "step": 19387
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.386466431313985e-06,
+      "loss": 0.2252,
+      "step": 19388
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3855043881657365e-06,
+      "loss": 0.2141,
+      "step": 19389
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.384542370990075e-06,
+      "loss": 0.2427,
+      "step": 19390
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.383580379796561e-06,
+      "loss": 0.2135,
+      "step": 19391
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.382618414594749e-06,
+      "loss": 0.2553,
+      "step": 19392
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.381656475394192e-06,
+      "loss": 0.2361,
+      "step": 19393
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.380694562204456e-06,
+      "loss": 0.2481,
+      "step": 19394
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.379732675035085e-06,
+      "loss": 0.2567,
+      "step": 19395
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.378770813895642e-06,
+      "loss": 0.2061,
+      "step": 19396
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.377808978795677e-06,
+      "loss": 0.271,
+      "step": 19397
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.376847169744753e-06,
+      "loss": 0.2192,
+      "step": 19398
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.375885386752417e-06,
+      "loss": 0.2536,
+      "step": 19399
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.374923629828226e-06,
+      "loss": 0.213,
+      "step": 19400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.373961898981733e-06,
+      "loss": 0.2544,
+      "step": 19401
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.37300019422249e-06,
+      "loss": 0.1939,
+      "step": 19402
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.372038515560062e-06,
+      "loss": 0.2347,
+      "step": 19403
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.371076863003988e-06,
+      "loss": 0.2213,
+      "step": 19404
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.370115236563828e-06,
+      "loss": 0.2136,
+      "step": 19405
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.369153636249136e-06,
+      "loss": 0.2336,
+      "step": 19406
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.36819206206946e-06,
+      "loss": 0.2305,
+      "step": 19407
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.367230514034358e-06,
+      "loss": 0.2274,
+      "step": 19408
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3662689921533785e-06,
+      "loss": 0.2441,
+      "step": 19409
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.365307496436072e-06,
+      "loss": 0.2249,
+      "step": 19410
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.364346026891991e-06,
+      "loss": 0.2014,
+      "step": 19411
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.363384583530688e-06,
+      "loss": 0.2981,
+      "step": 19412
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.362423166361718e-06,
+      "loss": 0.2366,
+      "step": 19413
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.361461775394625e-06,
+      "loss": 0.2615,
+      "step": 19414
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.360500410638962e-06,
+      "loss": 0.2517,
+      "step": 19415
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.359539072104279e-06,
+      "loss": 0.2381,
+      "step": 19416
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.358577759800127e-06,
+      "loss": 0.1792,
+      "step": 19417
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.357616473736058e-06,
+      "loss": 0.2079,
+      "step": 19418
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.356655213921615e-06,
+      "loss": 0.2238,
+      "step": 19419
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.35569398036635e-06,
+      "loss": 0.2312,
+      "step": 19420
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.354732773079814e-06,
+      "loss": 0.1577,
+      "step": 19421
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.353771592071555e-06,
+      "loss": 0.2359,
+      "step": 19422
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.352810437351124e-06,
+      "loss": 0.1714,
+      "step": 19423
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.351849308928062e-06,
+      "loss": 0.2492,
+      "step": 19424
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.350888206811923e-06,
+      "loss": 0.1643,
+      "step": 19425
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3499271310122515e-06,
+      "loss": 0.2453,
+      "step": 19426
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.348966081538597e-06,
+      "loss": 0.2442,
+      "step": 19427
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.348005058400507e-06,
+      "loss": 0.2669,
+      "step": 19428
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.347044061607523e-06,
+      "loss": 0.2239,
+      "step": 19429
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.346083091169199e-06,
+      "loss": 0.2132,
+      "step": 19430
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.345122147095076e-06,
+      "loss": 0.2049,
+      "step": 19431
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.344161229394706e-06,
+      "loss": 0.2347,
+      "step": 19432
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.343200338077629e-06,
+      "loss": 0.1922,
+      "step": 19433
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.342239473153392e-06,
+      "loss": 0.2465,
+      "step": 19434
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3412786346315415e-06,
+      "loss": 0.2251,
+      "step": 19435
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3403178225216185e-06,
+      "loss": 0.247,
+      "step": 19436
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.339357036833179e-06,
+      "loss": 0.2012,
+      "step": 19437
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3383962775757536e-06,
+      "loss": 0.2373,
+      "step": 19438
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.337435544758894e-06,
+      "loss": 0.2829,
+      "step": 19439
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.336474838392143e-06,
+      "loss": 0.2083,
+      "step": 19440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.335514158485044e-06,
+      "loss": 0.2409,
+      "step": 19441
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.334553505047143e-06,
+      "loss": 0.2229,
+      "step": 19442
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.333592878087978e-06,
+      "loss": 0.23,
+      "step": 19443
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.332632277617096e-06,
+      "loss": 0.2906,
+      "step": 19444
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3316717036440356e-06,
+      "loss": 0.1857,
+      "step": 19445
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.330711156178345e-06,
+      "loss": 0.205,
+      "step": 19446
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.329750635229564e-06,
+      "loss": 0.2158,
+      "step": 19447
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.328790140807234e-06,
+      "loss": 0.2205,
+      "step": 19448
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.327829672920894e-06,
+      "loss": 0.2747,
+      "step": 19449
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.32686923158009e-06,
+      "loss": 0.1837,
+      "step": 19450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.325908816794363e-06,
+      "loss": 0.2436,
+      "step": 19451
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.324948428573249e-06,
+      "loss": 0.2485,
+      "step": 19452
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.323988066926291e-06,
+      "loss": 0.1897,
+      "step": 19453
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.323027731863027e-06,
+      "loss": 0.187,
+      "step": 19454
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.322067423393002e-06,
+      "loss": 0.2265,
+      "step": 19455
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.321107141525757e-06,
+      "loss": 0.3058,
+      "step": 19456
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.320146886270826e-06,
+      "loss": 0.2581,
+      "step": 19457
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.319186657637749e-06,
+      "loss": 0.1917,
+      "step": 19458
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.318226455636067e-06,
+      "loss": 0.2151,
+      "step": 19459
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.317266280275316e-06,
+      "loss": 0.1898,
+      "step": 19460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.316306131565042e-06,
+      "loss": 0.1974,
+      "step": 19461
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3153460095147745e-06,
+      "loss": 0.284,
+      "step": 19462
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.314385914134052e-06,
+      "loss": 0.2309,
+      "step": 19463
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.313425845432416e-06,
+      "loss": 0.2545,
+      "step": 19464
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.312465803419402e-06,
+      "loss": 0.2319,
+      "step": 19465
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3115057881045514e-06,
+      "loss": 0.2354,
+      "step": 19466
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.310545799497395e-06,
+      "loss": 0.2191,
+      "step": 19467
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.309585837607471e-06,
+      "loss": 0.2243,
+      "step": 19468
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.308625902444317e-06,
+      "loss": 0.2342,
+      "step": 19469
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3076659940174654e-06,
+      "loss": 0.2796,
+      "step": 19470
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3067061123364635e-06,
+      "loss": 0.2228,
+      "step": 19471
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.305746257410831e-06,
+      "loss": 0.2142,
+      "step": 19472
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3047864292501125e-06,
+      "loss": 0.2802,
+      "step": 19473
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.303826627863841e-06,
+      "loss": 0.2692,
+      "step": 19474
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.302866853261552e-06,
+      "loss": 0.2875,
+      "step": 19475
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.301907105452782e-06,
+      "loss": 0.2631,
+      "step": 19476
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.300947384447059e-06,
+      "loss": 0.2273,
+      "step": 19477
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2999876902539214e-06,
+      "loss": 0.2297,
+      "step": 19478
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.299028022882899e-06,
+      "loss": 0.1988,
+      "step": 19479
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.298068382343535e-06,
+      "loss": 0.2367,
+      "step": 19480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.29710876864535e-06,
+      "loss": 0.1884,
+      "step": 19481
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.296149181797885e-06,
+      "loss": 0.2308,
+      "step": 19482
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2951896218106685e-06,
+      "loss": 0.2385,
+      "step": 19483
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.294230088693236e-06,
+      "loss": 0.1989,
+      "step": 19484
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.29327058245512e-06,
+      "loss": 0.2679,
+      "step": 19485
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.292311103105848e-06,
+      "loss": 0.2817,
+      "step": 19486
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.291351650654955e-06,
+      "loss": 0.2342,
+      "step": 19487
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2903922251119665e-06,
+      "loss": 0.2079,
+      "step": 19488
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2894328264864225e-06,
+      "loss": 0.2495,
+      "step": 19489
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.28847345478785e-06,
+      "loss": 0.2021,
+      "step": 19490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2875141100257784e-06,
+      "loss": 0.2545,
+      "step": 19491
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.286554792209738e-06,
+      "loss": 0.2161,
+      "step": 19492
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.285595501349259e-06,
+      "loss": 0.247,
+      "step": 19493
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.284636237453868e-06,
+      "loss": 0.2006,
+      "step": 19494
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.283677000533106e-06,
+      "loss": 0.163,
+      "step": 19495
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.282717790596488e-06,
+      "loss": 0.1883,
+      "step": 19496
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2817586076535475e-06,
+      "loss": 0.2686,
+      "step": 19497
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2807994517138155e-06,
+      "loss": 0.1388,
+      "step": 19498
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.27984032278682e-06,
+      "loss": 0.14,
+      "step": 19499
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.278881220882089e-06,
+      "loss": 0.1193,
+      "step": 19500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.277922146009149e-06,
+      "loss": 0.112,
+      "step": 19501
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.276963098177525e-06,
+      "loss": 0.13,
+      "step": 19502
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2760040773967474e-06,
+      "loss": 0.1212,
+      "step": 19503
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.275045083676344e-06,
+      "loss": 0.1422,
+      "step": 19504
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2740861170258445e-06,
+      "loss": 0.1585,
+      "step": 19505
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.273127177454767e-06,
+      "loss": 0.1404,
+      "step": 19506
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.272168264972641e-06,
+      "loss": 0.143,
+      "step": 19507
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2712093795889956e-06,
+      "loss": 0.1282,
+      "step": 19508
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.270250521313356e-06,
+      "loss": 0.0833,
+      "step": 19509
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.269291690155243e-06,
+      "loss": 0.1372,
+      "step": 19510
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.268332886124186e-06,
+      "loss": 0.1084,
+      "step": 19511
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2673741092297036e-06,
+      "loss": 0.1329,
+      "step": 19512
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.266415359481329e-06,
+      "loss": 0.1217,
+      "step": 19513
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.265456636888586e-06,
+      "loss": 0.1177,
+      "step": 19514
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.264497941460988e-06,
+      "loss": 0.1081,
+      "step": 19515
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.26353927320807e-06,
+      "loss": 0.0933,
+      "step": 19516
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.262580632139351e-06,
+      "loss": 0.1029,
+      "step": 19517
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.261622018264354e-06,
+      "loss": 0.1264,
+      "step": 19518
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2606634315926035e-06,
+      "loss": 0.1348,
+      "step": 19519
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.259704872133622e-06,
+      "loss": 0.0915,
+      "step": 19520
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.258746339896926e-06,
+      "loss": 0.0937,
+      "step": 19521
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.257787834892046e-06,
+      "loss": 0.1015,
+      "step": 19522
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.256829357128501e-06,
+      "loss": 0.1219,
+      "step": 19523
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.255870906615814e-06,
+      "loss": 0.1026,
+      "step": 19524
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.254912483363503e-06,
+      "loss": 0.0994,
+      "step": 19525
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.25395408738109e-06,
+      "loss": 0.114,
+      "step": 19526
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.252995718678095e-06,
+      "loss": 0.1343,
+      "step": 19527
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.252037377264041e-06,
+      "loss": 0.1213,
+      "step": 19528
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.251079063148449e-06,
+      "loss": 0.1012,
+      "step": 19529
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.250120776340836e-06,
+      "loss": 0.1228,
+      "step": 19530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.249162516850721e-06,
+      "loss": 0.0908,
+      "step": 19531
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.248204284687626e-06,
+      "loss": 0.1316,
+      "step": 19532
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.247246079861069e-06,
+      "loss": 0.0884,
+      "step": 19533
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.246287902380571e-06,
+      "loss": 0.1176,
+      "step": 19534
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2453297522556475e-06,
+      "loss": 0.1166,
+      "step": 19535
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2443716294958175e-06,
+      "loss": 0.0841,
+      "step": 19536
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.243413534110598e-06,
+      "loss": 0.0937,
+      "step": 19537
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.242455466109515e-06,
+      "loss": 0.1043,
+      "step": 19538
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.241497425502076e-06,
+      "loss": 0.0924,
+      "step": 19539
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.240539412297798e-06,
+      "loss": 0.1338,
+      "step": 19540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.239581426506203e-06,
+      "loss": 0.111,
+      "step": 19541
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.238623468136808e-06,
+      "loss": 0.0975,
+      "step": 19542
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.23766553719913e-06,
+      "loss": 0.1061,
+      "step": 19543
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.23670763370268e-06,
+      "loss": 0.1298,
+      "step": 19544
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.235749757656978e-06,
+      "loss": 0.1073,
+      "step": 19545
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.234791909071534e-06,
+      "loss": 0.082,
+      "step": 19546
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.233834087955873e-06,
+      "loss": 0.1381,
+      "step": 19547
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.232876294319507e-06,
+      "loss": 0.1425,
+      "step": 19548
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.231918528171944e-06,
+      "loss": 0.1101,
+      "step": 19549
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.230960789522703e-06,
+      "loss": 0.0994,
+      "step": 19550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2300030783813e-06,
+      "loss": 0.0983,
+      "step": 19551
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.229045394757247e-06,
+      "loss": 0.0911,
+      "step": 19552
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.22808773866006e-06,
+      "loss": 0.1275,
+      "step": 19553
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2271301100992485e-06,
+      "loss": 0.0704,
+      "step": 19554
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.226172509084326e-06,
+      "loss": 0.1327,
+      "step": 19555
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.225214935624809e-06,
+      "loss": 0.1057,
+      "step": 19556
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.224257389730207e-06,
+      "loss": 0.1432,
+      "step": 19557
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.223299871410037e-06,
+      "loss": 0.1144,
+      "step": 19558
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.222342380673806e-06,
+      "loss": 0.129,
+      "step": 19559
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.221384917531025e-06,
+      "loss": 0.085,
+      "step": 19560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.220427481991209e-06,
+      "loss": 0.1029,
+      "step": 19561
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.219470074063866e-06,
+      "loss": 0.1364,
+      "step": 19562
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.218512693758515e-06,
+      "loss": 0.1085,
+      "step": 19563
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.217555341084654e-06,
+      "loss": 0.1161,
+      "step": 19564
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.216598016051804e-06,
+      "loss": 0.108,
+      "step": 19565
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.21564071866947e-06,
+      "loss": 0.1151,
+      "step": 19566
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2146834489471665e-06,
+      "loss": 0.1248,
+      "step": 19567
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.213726206894396e-06,
+      "loss": 0.121,
+      "step": 19568
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.212768992520671e-06,
+      "loss": 0.1162,
+      "step": 19569
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.211811805835503e-06,
+      "loss": 0.1311,
+      "step": 19570
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.210854646848396e-06,
+      "loss": 0.1298,
+      "step": 19571
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.209897515568866e-06,
+      "loss": 0.1192,
+      "step": 19572
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.208940412006412e-06,
+      "loss": 0.1241,
+      "step": 19573
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2079833361705475e-06,
+      "loss": 0.1216,
+      "step": 19574
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.20702628807078e-06,
+      "loss": 0.1135,
+      "step": 19575
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.206069267716615e-06,
+      "loss": 0.1103,
+      "step": 19576
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.205112275117564e-06,
+      "loss": 0.139,
+      "step": 19577
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.204155310283128e-06,
+      "loss": 0.0818,
+      "step": 19578
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.203198373222815e-06,
+      "loss": 0.1533,
+      "step": 19579
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.202241463946131e-06,
+      "loss": 0.0975,
+      "step": 19580
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.201284582462585e-06,
+      "loss": 0.1254,
+      "step": 19581
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2003277287816845e-06,
+      "loss": 0.1029,
+      "step": 19582
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.199370902912928e-06,
+      "loss": 0.1334,
+      "step": 19583
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.1984141048658226e-06,
+      "loss": 0.1102,
+      "step": 19584
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.197457334649878e-06,
+      "loss": 0.0884,
+      "step": 19585
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.196500592274594e-06,
+      "loss": 0.0959,
+      "step": 19586
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.195543877749478e-06,
+      "loss": 0.1315,
+      "step": 19587
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.19458719108403e-06,
+      "loss": 0.093,
+      "step": 19588
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.193630532287757e-06,
+      "loss": 0.0982,
+      "step": 19589
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.192673901370162e-06,
+      "loss": 0.1317,
+      "step": 19590
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.191717298340748e-06,
+      "loss": 0.1694,
+      "step": 19591
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.19076072320902e-06,
+      "loss": 0.1486,
+      "step": 19592
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.189804175984478e-06,
+      "loss": 0.1156,
+      "step": 19593
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.188847656676625e-06,
+      "loss": 0.0769,
+      "step": 19594
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1878911652949626e-06,
+      "loss": 0.1278,
+      "step": 19595
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.186934701848996e-06,
+      "loss": 0.1174,
+      "step": 19596
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.185978266348221e-06,
+      "loss": 0.1495,
+      "step": 19597
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.185021858802141e-06,
+      "loss": 0.1198,
+      "step": 19598
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.18406547922026e-06,
+      "loss": 0.0674,
+      "step": 19599
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.183109127612077e-06,
+      "loss": 0.0817,
+      "step": 19600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1821528039870945e-06,
+      "loss": 0.1109,
+      "step": 19601
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.181196508354808e-06,
+      "loss": 0.1612,
+      "step": 19602
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1802402407247194e-06,
+      "loss": 0.1025,
+      "step": 19603
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1792840011063305e-06,
+      "loss": 0.1012,
+      "step": 19604
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.178327789509134e-06,
+      "loss": 0.1567,
+      "step": 19605
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.177371605942642e-06,
+      "loss": 0.1312,
+      "step": 19606
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.17641545041634e-06,
+      "loss": 0.1295,
+      "step": 19607
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.175459322939733e-06,
+      "loss": 0.1439,
+      "step": 19608
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.174503223522318e-06,
+      "loss": 0.1302,
+      "step": 19609
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.173547152173593e-06,
+      "loss": 0.1061,
+      "step": 19610
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1725911089030575e-06,
+      "loss": 0.1186,
+      "step": 19611
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1716350937202035e-06,
+      "loss": 0.1267,
+      "step": 19612
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.170679106634534e-06,
+      "loss": 0.1041,
+      "step": 19613
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.16972314765554e-06,
+      "loss": 0.1179,
+      "step": 19614
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.168767216792725e-06,
+      "loss": 0.1344,
+      "step": 19615
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1678113140555835e-06,
+      "loss": 0.1175,
+      "step": 19616
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.166855439453607e-06,
+      "loss": 0.1718,
+      "step": 19617
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.165899592996295e-06,
+      "loss": 0.1516,
+      "step": 19618
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.164943774693142e-06,
+      "loss": 0.1346,
+      "step": 19619
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.163987984553644e-06,
+      "loss": 0.1153,
+      "step": 19620
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.163032222587294e-06,
+      "loss": 0.1155,
+      "step": 19621
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1620764888035885e-06,
+      "loss": 0.1033,
+      "step": 19622
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.161120783212019e-06,
+      "loss": 0.0966,
+      "step": 19623
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.160165105822082e-06,
+      "loss": 0.104,
+      "step": 19624
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1592094566432745e-06,
+      "loss": 0.1363,
+      "step": 19625
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.158253835685083e-06,
+      "loss": 0.0661,
+      "step": 19626
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.157298242957005e-06,
+      "loss": 0.1306,
+      "step": 19627
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.156342678468532e-06,
+      "loss": 0.1308,
+      "step": 19628
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1553871422291574e-06,
+      "loss": 0.0911,
+      "step": 19629
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.154431634248375e-06,
+      "loss": 0.0977,
+      "step": 19630
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1534761545356745e-06,
+      "loss": 0.1257,
+      "step": 19631
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.152520703100545e-06,
+      "loss": 0.0966,
+      "step": 19632
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.151565279952484e-06,
+      "loss": 0.1137,
+      "step": 19633
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.150609885100979e-06,
+      "loss": 0.1274,
+      "step": 19634
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.149654518555526e-06,
+      "loss": 0.112,
+      "step": 19635
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.148699180325609e-06,
+      "loss": 0.1215,
+      "step": 19636
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.14774387042072e-06,
+      "loss": 0.1084,
+      "step": 19637
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.146788588850352e-06,
+      "loss": 0.0902,
+      "step": 19638
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.14583333562399e-06,
+      "loss": 0.1241,
+      "step": 19639
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.144878110751135e-06,
+      "loss": 0.0999,
+      "step": 19640
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.143922914241259e-06,
+      "loss": 0.1027,
+      "step": 19641
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.142967746103864e-06,
+      "loss": 0.0875,
+      "step": 19642
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.142012606348433e-06,
+      "loss": 0.1257,
+      "step": 19643
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.141057494984457e-06,
+      "loss": 0.1213,
+      "step": 19644
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.140102412021426e-06,
+      "loss": 0.1175,
+      "step": 19645
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.139147357468822e-06,
+      "loss": 0.0831,
+      "step": 19646
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.138192331336136e-06,
+      "loss": 0.1208,
+      "step": 19647
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.137237333632854e-06,
+      "loss": 0.1149,
+      "step": 19648
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1362823643684675e-06,
+      "loss": 0.102,
+      "step": 19649
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.135327423552456e-06,
+      "loss": 0.1258,
+      "step": 19650
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.134372511194312e-06,
+      "loss": 0.1002,
+      "step": 19651
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.133417627303518e-06,
+      "loss": 0.0923,
+      "step": 19652
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.132462771889561e-06,
+      "loss": 0.0954,
+      "step": 19653
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.131507944961931e-06,
+      "loss": 0.0892,
+      "step": 19654
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.130553146530105e-06,
+      "loss": 0.1218,
+      "step": 19655
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.129598376603572e-06,
+      "loss": 0.1098,
+      "step": 19656
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.128643635191816e-06,
+      "loss": 0.1419,
+      "step": 19657
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.127688922304325e-06,
+      "loss": 0.1144,
+      "step": 19658
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.126734237950583e-06,
+      "loss": 0.1909,
+      "step": 19659
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.125779582140069e-06,
+      "loss": 0.1227,
+      "step": 19660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.124824954882268e-06,
+      "loss": 0.1093,
+      "step": 19661
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.123870356186665e-06,
+      "loss": 0.1038,
+      "step": 19662
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.122915786062741e-06,
+      "loss": 0.1077,
+      "step": 19663
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.121961244519986e-06,
+      "loss": 0.1286,
+      "step": 19664
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.121006731567873e-06,
+      "loss": 0.1135,
+      "step": 19665
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.120052247215885e-06,
+      "loss": 0.1104,
+      "step": 19666
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.11909779147351e-06,
+      "loss": 0.1089,
+      "step": 19667
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.118143364350228e-06,
+      "loss": 0.1384,
+      "step": 19668
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1171889658555184e-06,
+      "loss": 0.0847,
+      "step": 19669
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.116234595998861e-06,
+      "loss": 0.1587,
+      "step": 19670
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1152802547897405e-06,
+      "loss": 0.0985,
+      "step": 19671
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.114325942237631e-06,
+      "loss": 0.0982,
+      "step": 19672
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.11337165835202e-06,
+      "loss": 0.1111,
+      "step": 19673
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1124174031423875e-06,
+      "loss": 0.1004,
+      "step": 19674
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.111463176618205e-06,
+      "loss": 0.1016,
+      "step": 19675
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.110508978788958e-06,
+      "loss": 0.1067,
+      "step": 19676
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.109554809664125e-06,
+      "loss": 0.0849,
+      "step": 19677
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.108600669253186e-06,
+      "loss": 0.0996,
+      "step": 19678
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.107646557565615e-06,
+      "loss": 0.1361,
+      "step": 19679
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.106692474610894e-06,
+      "loss": 0.1378,
+      "step": 19680
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.105738420398498e-06,
+      "loss": 0.1348,
+      "step": 19681
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.104784394937908e-06,
+      "loss": 0.1376,
+      "step": 19682
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.103830398238605e-06,
+      "loss": 0.109,
+      "step": 19683
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.102876430310053e-06,
+      "loss": 0.113,
+      "step": 19684
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.101922491161739e-06,
+      "loss": 0.0911,
+      "step": 19685
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.100968580803138e-06,
+      "loss": 0.0865,
+      "step": 19686
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.100014699243727e-06,
+      "loss": 0.1007,
+      "step": 19687
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.09906084649298e-06,
+      "loss": 0.1257,
+      "step": 19688
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.098107022560372e-06,
+      "loss": 0.1185,
+      "step": 19689
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.097153227455379e-06,
+      "loss": 0.1237,
+      "step": 19690
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.096199461187474e-06,
+      "loss": 0.148,
+      "step": 19691
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.095245723766137e-06,
+      "loss": 0.1056,
+      "step": 19692
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.094292015200841e-06,
+      "loss": 0.0814,
+      "step": 19693
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0933383355010574e-06,
+      "loss": 0.1057,
+      "step": 19694
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.092384684676263e-06,
+      "loss": 0.1144,
+      "step": 19695
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0914310627359274e-06,
+      "loss": 0.129,
+      "step": 19696
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.090477469689526e-06,
+      "loss": 0.1109,
+      "step": 19697
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.089523905546538e-06,
+      "loss": 0.0918,
+      "step": 19698
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.088570370316428e-06,
+      "loss": 0.1337,
+      "step": 19699
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.087616864008668e-06,
+      "loss": 0.1077,
+      "step": 19700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.086663386632736e-06,
+      "loss": 0.0797,
+      "step": 19701
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.085709938198101e-06,
+      "loss": 0.0871,
+      "step": 19702
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.084756518714237e-06,
+      "loss": 0.1405,
+      "step": 19703
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.083803128190611e-06,
+      "loss": 0.0897,
+      "step": 19704
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0828497666366965e-06,
+      "loss": 0.101,
+      "step": 19705
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.081896434061961e-06,
+      "loss": 0.1024,
+      "step": 19706
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.080943130475886e-06,
+      "loss": 0.1044,
+      "step": 19707
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.079989855887929e-06,
+      "loss": 0.1381,
+      "step": 19708
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.079036610307561e-06,
+      "loss": 0.0845,
+      "step": 19709
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.078083393744259e-06,
+      "loss": 0.0766,
+      "step": 19710
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.077130206207488e-06,
+      "loss": 0.0793,
+      "step": 19711
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.076177047706721e-06,
+      "loss": 0.108,
+      "step": 19712
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.075223918251419e-06,
+      "loss": 0.0908,
+      "step": 19713
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.074270817851057e-06,
+      "loss": 0.1242,
+      "step": 19714
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.073317746515097e-06,
+      "loss": 0.1077,
+      "step": 19715
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.072364704253014e-06,
+      "loss": 0.0961,
+      "step": 19716
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.071411691074276e-06,
+      "loss": 0.13,
+      "step": 19717
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.070458706988341e-06,
+      "loss": 0.1219,
+      "step": 19718
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.069505752004684e-06,
+      "loss": 0.1386,
+      "step": 19719
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.068552826132768e-06,
+      "loss": 0.1097,
+      "step": 19720
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.067599929382063e-06,
+      "loss": 0.1727,
+      "step": 19721
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.066647061762033e-06,
+      "loss": 0.1017,
+      "step": 19722
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.065694223282142e-06,
+      "loss": 0.1218,
+      "step": 19723
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.064741413951856e-06,
+      "loss": 0.1158,
+      "step": 19724
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.063788633780646e-06,
+      "loss": 0.1356,
+      "step": 19725
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.06283588277797e-06,
+      "loss": 0.1124,
+      "step": 19726
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0618831609532975e-06,
+      "loss": 0.106,
+      "step": 19727
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.06093046831609e-06,
+      "loss": 0.1077,
+      "step": 19728
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.059977804875813e-06,
+      "loss": 0.1212,
+      "step": 19729
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.059025170641929e-06,
+      "loss": 0.1421,
+      "step": 19730
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.058072565623899e-06,
+      "loss": 0.1243,
+      "step": 19731
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.057119989831198e-06,
+      "loss": 0.0871,
+      "step": 19732
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0561674432732715e-06,
+      "loss": 0.118,
+      "step": 19733
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.055214925959596e-06,
+      "loss": 0.0866,
+      "step": 19734
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.054262437899628e-06,
+      "loss": 0.1268,
+      "step": 19735
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.053309979102833e-06,
+      "loss": 0.1558,
+      "step": 19736
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.052357549578668e-06,
+      "loss": 0.122,
+      "step": 19737
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.051405149336596e-06,
+      "loss": 0.1282,
+      "step": 19738
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.050452778386081e-06,
+      "loss": 0.121,
+      "step": 19739
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.049500436736579e-06,
+      "loss": 0.1059,
+      "step": 19740
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.048548124397561e-06,
+      "loss": 0.1026,
+      "step": 19741
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.047595841378472e-06,
+      "loss": 0.0961,
+      "step": 19742
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.046643587688784e-06,
+      "loss": 0.1142,
+      "step": 19743
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.045691363337952e-06,
+      "loss": 0.1214,
+      "step": 19744
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.044739168335436e-06,
+      "loss": 0.1183,
+      "step": 19745
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.043787002690698e-06,
+      "loss": 0.1148,
+      "step": 19746
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.042834866413193e-06,
+      "loss": 0.0701,
+      "step": 19747
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.04188275951238e-06,
+      "loss": 0.0886,
+      "step": 19748
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.040930681997716e-06,
+      "loss": 0.1066,
+      "step": 19749
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.039978633878664e-06,
+      "loss": 0.1202,
+      "step": 19750
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.039026615164681e-06,
+      "loss": 0.0868,
+      "step": 19751
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0380746258652215e-06,
+      "loss": 0.1806,
+      "step": 19752
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.037122665989742e-06,
+      "loss": 0.1235,
+      "step": 19753
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.036170735547701e-06,
+      "loss": 0.1124,
+      "step": 19754
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.035218834548554e-06,
+      "loss": 0.1326,
+      "step": 19755
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.034266963001762e-06,
+      "loss": 0.1426,
+      "step": 19756
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.033315120916775e-06,
+      "loss": 0.1716,
+      "step": 19757
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.032363308303047e-06,
+      "loss": 0.1472,
+      "step": 19758
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.031411525170042e-06,
+      "loss": 0.0991,
+      "step": 19759
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.030459771527208e-06,
+      "loss": 0.1157,
+      "step": 19760
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.029508047384005e-06,
+      "loss": 0.1351,
+      "step": 19761
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0285563527498825e-06,
+      "loss": 0.0957,
+      "step": 19762
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.027604687634297e-06,
+      "loss": 0.1543,
+      "step": 19763
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0266530520467015e-06,
+      "loss": 0.0926,
+      "step": 19764
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.025701445996553e-06,
+      "loss": 0.1079,
+      "step": 19765
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0247498694932995e-06,
+      "loss": 0.1054,
+      "step": 19766
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.023798322546396e-06,
+      "loss": 0.1204,
+      "step": 19767
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.022846805165295e-06,
+      "loss": 0.113,
+      "step": 19768
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.021895317359453e-06,
+      "loss": 0.1156,
+      "step": 19769
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.02094385913832e-06,
+      "loss": 0.1291,
+      "step": 19770
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.019992430511344e-06,
+      "loss": 0.1049,
+      "step": 19771
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.019041031487979e-06,
+      "loss": 0.0733,
+      "step": 19772
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.018089662077679e-06,
+      "loss": 0.1229,
+      "step": 19773
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.017138322289889e-06,
+      "loss": 0.1154,
+      "step": 19774
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.016187012134069e-06,
+      "loss": 0.101,
+      "step": 19775
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.015235731619658e-06,
+      "loss": 0.1105,
+      "step": 19776
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.014284480756115e-06,
+      "loss": 0.1586,
+      "step": 19777
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0133332595528845e-06,
+      "loss": 0.0837,
+      "step": 19778
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.012382068019419e-06,
+      "loss": 0.097,
+      "step": 19779
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.01143090616517e-06,
+      "loss": 0.0857,
+      "step": 19780
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.01047977399958e-06,
+      "loss": 0.1142,
+      "step": 19781
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.009528671532101e-06,
+      "loss": 0.1107,
+      "step": 19782
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.008577598772179e-06,
+      "loss": 0.1085,
+      "step": 19783
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.007626555729265e-06,
+      "loss": 0.1519,
+      "step": 19784
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.006675542412811e-06,
+      "loss": 0.1069,
+      "step": 19785
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.005724558832254e-06,
+      "loss": 0.1149,
+      "step": 19786
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.004773604997048e-06,
+      "loss": 0.1371,
+      "step": 19787
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.003822680916639e-06,
+      "loss": 0.1101,
+      "step": 19788
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.002871786600474e-06,
+      "loss": 0.1012,
+      "step": 19789
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.0019209220579965e-06,
+      "loss": 0.1009,
+      "step": 19790
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.000970087298653e-06,
+      "loss": 0.1232,
+      "step": 19791
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.000019282331889e-06,
+      "loss": 0.1121,
+      "step": 19792
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.999068507167154e-06,
+      "loss": 0.0844,
+      "step": 19793
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9981177618138914e-06,
+      "loss": 0.1353,
+      "step": 19794
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9971670462815425e-06,
+      "loss": 0.1417,
+      "step": 19795
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9962163605795544e-06,
+      "loss": 0.1223,
+      "step": 19796
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.995265704717371e-06,
+      "loss": 0.0992,
+      "step": 19797
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.994315078704435e-06,
+      "loss": 0.1195,
+      "step": 19798
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.993364482550193e-06,
+      "loss": 0.1412,
+      "step": 19799
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.992413916264087e-06,
+      "loss": 0.096,
+      "step": 19800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.991463379855554e-06,
+      "loss": 0.0917,
+      "step": 19801
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.990512873334046e-06,
+      "loss": 0.1262,
+      "step": 19802
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.989562396709001e-06,
+      "loss": 0.1034,
+      "step": 19803
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.988611949989864e-06,
+      "loss": 0.1025,
+      "step": 19804
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.987661533186072e-06,
+      "loss": 0.1079,
+      "step": 19805
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.986711146307068e-06,
+      "loss": 0.1001,
+      "step": 19806
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9857607893622955e-06,
+      "loss": 0.1251,
+      "step": 19807
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9848104623611914e-06,
+      "loss": 0.1077,
+      "step": 19808
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.983860165313205e-06,
+      "loss": 0.1257,
+      "step": 19809
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.982909898227764e-06,
+      "loss": 0.1126,
+      "step": 19810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9819596611143184e-06,
+      "loss": 0.1196,
+      "step": 19811
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.981009453982303e-06,
+      "loss": 0.1194,
+      "step": 19812
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.98005927684116e-06,
+      "loss": 0.0827,
+      "step": 19813
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.979109129700328e-06,
+      "loss": 0.1066,
+      "step": 19814
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.978159012569243e-06,
+      "loss": 0.1259,
+      "step": 19815
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.977208925457346e-06,
+      "loss": 0.121,
+      "step": 19816
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9762588683740725e-06,
+      "loss": 0.0986,
+      "step": 19817
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.975308841328871e-06,
+      "loss": 0.0955,
+      "step": 19818
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.974358844331163e-06,
+      "loss": 0.121,
+      "step": 19819
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.973408877390395e-06,
+      "loss": 0.1023,
+      "step": 19820
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.972458940516005e-06,
+      "loss": 0.1172,
+      "step": 19821
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.971509033717425e-06,
+      "loss": 0.1388,
+      "step": 19822
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.970559157004098e-06,
+      "loss": 0.083,
+      "step": 19823
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.969609310385454e-06,
+      "loss": 0.0898,
+      "step": 19824
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.96865949387093e-06,
+      "loss": 0.1192,
+      "step": 19825
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.967709707469961e-06,
+      "loss": 0.0907,
+      "step": 19826
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.966759951191985e-06,
+      "loss": 0.1012,
+      "step": 19827
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.965810225046439e-06,
+      "loss": 0.1348,
+      "step": 19828
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.964860529042752e-06,
+      "loss": 0.0979,
+      "step": 19829
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.963910863190359e-06,
+      "loss": 0.0995,
+      "step": 19830
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.962961227498697e-06,
+      "loss": 0.134,
+      "step": 19831
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9620116219771975e-06,
+      "loss": 0.1627,
+      "step": 19832
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9610620466353e-06,
+      "loss": 0.1033,
+      "step": 19833
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.960112501482428e-06,
+      "loss": 0.0945,
+      "step": 19834
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.959162986528019e-06,
+      "loss": 0.0929,
+      "step": 19835
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.958213501781506e-06,
+      "loss": 0.1157,
+      "step": 19836
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.95726404725232e-06,
+      "loss": 0.1219,
+      "step": 19837
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.956314622949897e-06,
+      "loss": 0.0978,
+      "step": 19838
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.955365228883663e-06,
+      "loss": 0.1133,
+      "step": 19839
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.954415865063051e-06,
+      "loss": 0.0997,
+      "step": 19840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.953466531497492e-06,
+      "loss": 0.101,
+      "step": 19841
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.952517228196418e-06,
+      "loss": 0.127,
+      "step": 19842
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.951567955169265e-06,
+      "loss": 0.1209,
+      "step": 19843
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9506187124254495e-06,
+      "loss": 0.1166,
+      "step": 19844
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.949669499974412e-06,
+      "loss": 0.1223,
+      "step": 19845
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9487203178255776e-06,
+      "loss": 0.0913,
+      "step": 19846
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.947771165988379e-06,
+      "loss": 0.0855,
+      "step": 19847
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.946822044472244e-06,
+      "loss": 0.1186,
+      "step": 19848
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9458729532865965e-06,
+      "loss": 0.1181,
+      "step": 19849
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.944923892440869e-06,
+      "loss": 0.0836,
+      "step": 19850
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.94397486194449e-06,
+      "loss": 0.0722,
+      "step": 19851
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.943025861806891e-06,
+      "loss": 0.0932,
+      "step": 19852
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.942076892037489e-06,
+      "loss": 0.1066,
+      "step": 19853
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.941127952645719e-06,
+      "loss": 0.1488,
+      "step": 19854
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.940179043641005e-06,
+      "loss": 0.0893,
+      "step": 19855
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9392301650327756e-06,
+      "loss": 0.1057,
+      "step": 19856
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.938281316830457e-06,
+      "loss": 0.0986,
+      "step": 19857
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.937332499043472e-06,
+      "loss": 0.107,
+      "step": 19858
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.936383711681249e-06,
+      "loss": 0.0911,
+      "step": 19859
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.93543495475321e-06,
+      "loss": 0.1228,
+      "step": 19860
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.934486228268784e-06,
+      "loss": 0.087,
+      "step": 19861
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.933537532237395e-06,
+      "loss": 0.1129,
+      "step": 19862
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.932588866668467e-06,
+      "loss": 0.1227,
+      "step": 19863
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9316402315714234e-06,
+      "loss": 0.1537,
+      "step": 19864
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.930691626955688e-06,
+      "loss": 0.0915,
+      "step": 19865
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.929743052830682e-06,
+      "loss": 0.142,
+      "step": 19866
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.928794509205838e-06,
+      "loss": 0.11,
+      "step": 19867
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.927845996090568e-06,
+      "loss": 0.116,
+      "step": 19868
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9268975134942966e-06,
+      "loss": 0.0825,
+      "step": 19869
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.925949061426449e-06,
+      "loss": 0.1125,
+      "step": 19870
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.925000639896449e-06,
+      "loss": 0.167,
+      "step": 19871
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.924052248913716e-06,
+      "loss": 0.1106,
+      "step": 19872
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.92310388848767e-06,
+      "loss": 0.1571,
+      "step": 19873
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.922155558627732e-06,
+      "loss": 0.1004,
+      "step": 19874
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.921207259343323e-06,
+      "loss": 0.0894,
+      "step": 19875
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.920258990643871e-06,
+      "loss": 0.1024,
+      "step": 19876
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.919310752538784e-06,
+      "loss": 0.145,
+      "step": 19877
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.918362545037487e-06,
+      "loss": 0.1161,
+      "step": 19878
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.917414368149401e-06,
+      "loss": 0.1201,
+      "step": 19879
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.916466221883944e-06,
+      "loss": 0.1161,
+      "step": 19880
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.915518106250538e-06,
+      "loss": 0.095,
+      "step": 19881
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.914570021258598e-06,
+      "loss": 0.1098,
+      "step": 19882
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.913621966917542e-06,
+      "loss": 0.0849,
+      "step": 19883
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.912673943236787e-06,
+      "loss": 0.0824,
+      "step": 19884
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.911725950225756e-06,
+      "loss": 0.1359,
+      "step": 19885
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.910777987893866e-06,
+      "loss": 0.1228,
+      "step": 19886
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.1406,
+      "step": 19887
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.908882155305162e-06,
+      "loss": 0.1369,
+      "step": 19888
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.907934285067186e-06,
+      "loss": 0.1166,
+      "step": 19889
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.906986445546015e-06,
+      "loss": 0.1341,
+      "step": 19890
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.906038636751067e-06,
+      "loss": 0.1358,
+      "step": 19891
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.905090858691754e-06,
+      "loss": 0.1057,
+      "step": 19892
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.90414311137749e-06,
+      "loss": 0.1179,
+      "step": 19893
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.903195394817695e-06,
+      "loss": 0.0946,
+      "step": 19894
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.902247709021783e-06,
+      "loss": 0.1252,
+      "step": 19895
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.901300053999168e-06,
+      "loss": 0.1045,
+      "step": 19896
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.900352429759262e-06,
+      "loss": 0.0965,
+      "step": 19897
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.899404836311478e-06,
+      "loss": 0.1409,
+      "step": 19898
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.898457273665233e-06,
+      "loss": 0.1762,
+      "step": 19899
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.897509741829935e-06,
+      "loss": 0.1485,
+      "step": 19900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.896562240815008e-06,
+      "loss": 0.1249,
+      "step": 19901
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.89561477062985e-06,
+      "loss": 0.1058,
+      "step": 19902
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.894667331283882e-06,
+      "loss": 0.0994,
+      "step": 19903
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.893719922786515e-06,
+      "loss": 0.1011,
+      "step": 19904
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.89277254514716e-06,
+      "loss": 0.1098,
+      "step": 19905
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.891825198375227e-06,
+      "loss": 0.1021,
+      "step": 19906
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.890877882480128e-06,
+      "loss": 0.121,
+      "step": 19907
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.889930597471273e-06,
+      "loss": 0.1166,
+      "step": 19908
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.888983343358071e-06,
+      "loss": 0.136,
+      "step": 19909
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.8880361201499415e-06,
+      "loss": 0.1344,
+      "step": 19910
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.88708892785628e-06,
+      "loss": 0.1103,
+      "step": 19911
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.8861417664865045e-06,
+      "loss": 0.1364,
+      "step": 19912
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.885194636050023e-06,
+      "loss": 0.0845,
+      "step": 19913
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.884247536556245e-06,
+      "loss": 0.1564,
+      "step": 19914
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.88330046801458e-06,
+      "loss": 0.0969,
+      "step": 19915
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.882353430434431e-06,
+      "loss": 0.118,
+      "step": 19916
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.88140642382521e-06,
+      "loss": 0.1277,
+      "step": 19917
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.88045944819632e-06,
+      "loss": 0.1128,
+      "step": 19918
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.879512503557177e-06,
+      "loss": 0.1338,
+      "step": 19919
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.878565589917185e-06,
+      "loss": 0.1099,
+      "step": 19920
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.877618707285745e-06,
+      "loss": 0.1491,
+      "step": 19921
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.876671855672268e-06,
+      "loss": 0.081,
+      "step": 19922
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.875725035086158e-06,
+      "loss": 0.1301,
+      "step": 19923
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.874778245536824e-06,
+      "loss": 0.0994,
+      "step": 19924
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.873831487033671e-06,
+      "loss": 0.0997,
+      "step": 19925
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.872884759586101e-06,
+      "loss": 0.1128,
+      "step": 19926
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.871938063203518e-06,
+      "loss": 0.1016,
+      "step": 19927
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8709913978953324e-06,
+      "loss": 0.0978,
+      "step": 19928
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.870044763670944e-06,
+      "loss": 0.1121,
+      "step": 19929
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8690981605397625e-06,
+      "loss": 0.1134,
+      "step": 19930
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8681515885111826e-06,
+      "loss": 0.0962,
+      "step": 19931
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.867205047594613e-06,
+      "loss": 0.1429,
+      "step": 19932
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.866258537799455e-06,
+      "loss": 0.1413,
+      "step": 19933
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.865312059135115e-06,
+      "loss": 0.1206,
+      "step": 19934
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.86436561161099e-06,
+      "loss": 0.1282,
+      "step": 19935
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.863419195236483e-06,
+      "loss": 0.1051,
+      "step": 19936
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.862472810020998e-06,
+      "loss": 0.0968,
+      "step": 19937
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.861526455973937e-06,
+      "loss": 0.093,
+      "step": 19938
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.860580133104701e-06,
+      "loss": 0.0876,
+      "step": 19939
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.859633841422688e-06,
+      "loss": 0.1296,
+      "step": 19940
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8586875809373e-06,
+      "loss": 0.1246,
+      "step": 19941
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.857741351657938e-06,
+      "loss": 0.1158,
+      "step": 19942
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8567951535939995e-06,
+      "loss": 0.0985,
+      "step": 19943
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.855848986754892e-06,
+      "loss": 0.0853,
+      "step": 19944
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8549028511500016e-06,
+      "loss": 0.1317,
+      "step": 19945
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.853956746788737e-06,
+      "loss": 0.1611,
+      "step": 19946
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.853010673680496e-06,
+      "loss": 0.0917,
+      "step": 19947
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.852064631834673e-06,
+      "loss": 0.1352,
+      "step": 19948
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.851118621260671e-06,
+      "loss": 0.1061,
+      "step": 19949
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.850172641967883e-06,
+      "loss": 0.0995,
+      "step": 19950
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.84922669396571e-06,
+      "loss": 0.1004,
+      "step": 19951
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.848280777263543e-06,
+      "loss": 0.141,
+      "step": 19952
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.847334891870787e-06,
+      "loss": 0.0883,
+      "step": 19953
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.846389037796837e-06,
+      "loss": 0.1139,
+      "step": 19954
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.845443215051084e-06,
+      "loss": 0.1226,
+      "step": 19955
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8444974236429284e-06,
+      "loss": 0.1662,
+      "step": 19956
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.843551663581765e-06,
+      "loss": 0.1286,
+      "step": 19957
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.842605934876988e-06,
+      "loss": 0.1611,
+      "step": 19958
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.841660237537993e-06,
+      "loss": 0.1159,
+      "step": 19959
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8407145715741715e-06,
+      "loss": 0.1135,
+      "step": 19960
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.839768936994921e-06,
+      "loss": 0.1589,
+      "step": 19961
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.838823333809637e-06,
+      "loss": 0.0879,
+      "step": 19962
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.837877762027711e-06,
+      "loss": 0.0873,
+      "step": 19963
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.836932221658538e-06,
+      "loss": 0.1166,
+      "step": 19964
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.835986712711507e-06,
+      "loss": 0.1272,
+      "step": 19965
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.835041235196016e-06,
+      "loss": 0.118,
+      "step": 19966
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.834095789121453e-06,
+      "loss": 0.0976,
+      "step": 19967
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.833150374497215e-06,
+      "loss": 0.1901,
+      "step": 19968
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8322049913326895e-06,
+      "loss": 0.1038,
+      "step": 19969
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.831259639637267e-06,
+      "loss": 0.1008,
+      "step": 19970
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.830314319420342e-06,
+      "loss": 0.1096,
+      "step": 19971
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.829369030691307e-06,
+      "loss": 0.107,
+      "step": 19972
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8284237734595495e-06,
+      "loss": 0.096,
+      "step": 19973
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.827478547734459e-06,
+      "loss": 0.0972,
+      "step": 19974
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.826533353525427e-06,
+      "loss": 0.1451,
+      "step": 19975
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.825588190841843e-06,
+      "loss": 0.1489,
+      "step": 19976
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.824643059693094e-06,
+      "loss": 0.0867,
+      "step": 19977
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.823697960088577e-06,
+      "loss": 0.0889,
+      "step": 19978
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.822752892037671e-06,
+      "loss": 0.1073,
+      "step": 19979
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.821807855549769e-06,
+      "loss": 0.1044,
+      "step": 19980
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.820862850634258e-06,
+      "loss": 0.1107,
+      "step": 19981
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.819917877300526e-06,
+      "loss": 0.1218,
+      "step": 19982
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.818972935557962e-06,
+      "loss": 0.1091,
+      "step": 19983
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.81802802541595e-06,
+      "loss": 0.1203,
+      "step": 19984
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.81708314688388e-06,
+      "loss": 0.0885,
+      "step": 19985
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.816138299971133e-06,
+      "loss": 0.0996,
+      "step": 19986
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.8151934846871056e-06,
+      "loss": 0.0961,
+      "step": 19987
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.81424870104117e-06,
+      "loss": 0.1004,
+      "step": 19988
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.813303949042725e-06,
+      "loss": 0.124,
+      "step": 19989
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.812359228701146e-06,
+      "loss": 0.1161,
+      "step": 19990
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.811414540025822e-06,
+      "loss": 0.13,
+      "step": 19991
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.810469883026141e-06,
+      "loss": 0.1091,
+      "step": 19992
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.80952525771148e-06,
+      "loss": 0.1027,
+      "step": 19993
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.808580664091227e-06,
+      "loss": 0.1371,
+      "step": 19994
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.807636102174764e-06,
+      "loss": 0.1036,
+      "step": 19995
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.806691571971477e-06,
+      "loss": 0.1203,
+      "step": 19996
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.80574707349075e-06,
+      "loss": 0.1019,
+      "step": 19997
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.80480260674196e-06,
+      "loss": 0.0919,
+      "step": 19998
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.803858171734495e-06,
+      "loss": 0.1116,
+      "step": 19999
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.802913768477732e-06,
+      "loss": 0.0665,
+      "step": 20000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.801969396981054e-06,
+      "loss": 0.1297,
+      "step": 20001
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.801025057253852e-06,
+      "loss": 0.1133,
+      "step": 20002
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.800080749305494e-06,
+      "loss": 0.0876,
+      "step": 20003
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.799136473145365e-06,
+      "loss": 0.1241,
+      "step": 20004
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.798192228782847e-06,
+      "loss": 0.115,
+      "step": 20005
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.79724801622732e-06,
+      "loss": 0.1041,
+      "step": 20006
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.796303835488166e-06,
+      "loss": 0.1194,
+      "step": 20007
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.795359686574761e-06,
+      "loss": 0.0962,
+      "step": 20008
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.794415569496484e-06,
+      "loss": 0.1332,
+      "step": 20009
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.793471484262713e-06,
+      "loss": 0.1211,
+      "step": 20010
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7925274308828315e-06,
+      "loss": 0.1243,
+      "step": 20011
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.79158340936622e-06,
+      "loss": 0.0874,
+      "step": 20012
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7906394197222444e-06,
+      "loss": 0.1102,
+      "step": 20013
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.789695461960293e-06,
+      "loss": 0.1177,
+      "step": 20014
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.788751536089739e-06,
+      "loss": 0.1002,
+      "step": 20015
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.787807642119961e-06,
+      "loss": 0.0922,
+      "step": 20016
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7868637800603355e-06,
+      "loss": 0.1043,
+      "step": 20017
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.785919949920237e-06,
+      "loss": 0.1055,
+      "step": 20018
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.784976151709042e-06,
+      "loss": 0.1015,
+      "step": 20019
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.784032385436125e-06,
+      "loss": 0.1288,
+      "step": 20020
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7830886511108715e-06,
+      "loss": 0.1275,
+      "step": 20021
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.782144948742643e-06,
+      "loss": 0.1264,
+      "step": 20022
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.781201278340821e-06,
+      "loss": 0.1049,
+      "step": 20023
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.780257639914778e-06,
+      "loss": 0.0922,
+      "step": 20024
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.77931403347389e-06,
+      "loss": 0.1043,
+      "step": 20025
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.778370459027533e-06,
+      "loss": 0.1207,
+      "step": 20026
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.777426916585075e-06,
+      "loss": 0.1152,
+      "step": 20027
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.776483406155892e-06,
+      "loss": 0.0839,
+      "step": 20028
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7755399277493525e-06,
+      "loss": 0.1224,
+      "step": 20029
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.774596481374838e-06,
+      "loss": 0.0796,
+      "step": 20030
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7736530670417165e-06,
+      "loss": 0.0961,
+      "step": 20031
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.772709684759359e-06,
+      "loss": 0.1049,
+      "step": 20032
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.771766334537137e-06,
+      "loss": 0.096,
+      "step": 20033
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.770823016384423e-06,
+      "loss": 0.0764,
+      "step": 20034
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.769879730310583e-06,
+      "loss": 0.1272,
+      "step": 20035
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.768936476325001e-06,
+      "loss": 0.1304,
+      "step": 20036
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.767993254437033e-06,
+      "loss": 0.1445,
+      "step": 20037
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7670500646560525e-06,
+      "loss": 0.0891,
+      "step": 20038
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.766106906991433e-06,
+      "loss": 0.0945,
+      "step": 20039
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7651637814525424e-06,
+      "loss": 0.0944,
+      "step": 20040
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.764220688048753e-06,
+      "loss": 0.1036,
+      "step": 20041
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.763277626789426e-06,
+      "loss": 0.078,
+      "step": 20042
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.762334597683935e-06,
+      "loss": 0.0895,
+      "step": 20043
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.761391600741644e-06,
+      "loss": 0.1241,
+      "step": 20044
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7604486359719305e-06,
+      "loss": 0.1284,
+      "step": 20045
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.75950570338415e-06,
+      "loss": 0.1011,
+      "step": 20046
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.758562802987675e-06,
+      "loss": 0.0796,
+      "step": 20047
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7576199347918725e-06,
+      "loss": 0.1552,
+      "step": 20048
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.75667709880611e-06,
+      "loss": 0.1115,
+      "step": 20049
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.755734295039755e-06,
+      "loss": 0.1346,
+      "step": 20050
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.754791523502168e-06,
+      "loss": 0.1405,
+      "step": 20051
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7538487842027166e-06,
+      "loss": 0.0992,
+      "step": 20052
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.752906077150766e-06,
+      "loss": 0.1002,
+      "step": 20053
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7519634023556834e-06,
+      "loss": 0.1212,
+      "step": 20054
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.751020759826836e-06,
+      "loss": 0.1499,
+      "step": 20055
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7500781495735804e-06,
+      "loss": 0.1163,
+      "step": 20056
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.749135571605283e-06,
+      "loss": 0.0845,
+      "step": 20057
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.74819302593131e-06,
+      "loss": 0.108,
+      "step": 20058
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.747250512561024e-06,
+      "loss": 0.0996,
+      "step": 20059
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.746308031503787e-06,
+      "loss": 0.1015,
+      "step": 20060
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.745365582768963e-06,
+      "loss": 0.1017,
+      "step": 20061
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7444231663659095e-06,
+      "loss": 0.1122,
+      "step": 20062
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.743480782303996e-06,
+      "loss": 0.105,
+      "step": 20063
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.74253843059258e-06,
+      "loss": 0.0706,
+      "step": 20064
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.741596111241026e-06,
+      "loss": 0.1065,
+      "step": 20065
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.74065382425869e-06,
+      "loss": 0.1034,
+      "step": 20066
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.739711569654935e-06,
+      "loss": 0.1347,
+      "step": 20067
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.738769347439123e-06,
+      "loss": 0.0866,
+      "step": 20068
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.737827157620609e-06,
+      "loss": 0.1155,
+      "step": 20069
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.736885000208765e-06,
+      "loss": 0.1222,
+      "step": 20070
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7359428752129356e-06,
+      "loss": 0.0914,
+      "step": 20071
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.735000782642487e-06,
+      "loss": 0.085,
+      "step": 20072
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.734058722506778e-06,
+      "loss": 0.1558,
+      "step": 20073
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733116694815169e-06,
+      "loss": 0.1168,
+      "step": 20074
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7321746995770144e-06,
+      "loss": 0.1036,
+      "step": 20075
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731232736801672e-06,
+      "loss": 0.0632,
+      "step": 20076
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.730290806498501e-06,
+      "loss": 0.1142,
+      "step": 20077
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.729348908676858e-06,
+      "loss": 0.1,
+      "step": 20078
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.728407043346105e-06,
+      "loss": 0.1046,
+      "step": 20079
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.727465210515589e-06,
+      "loss": 0.1086,
+      "step": 20080
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.726523410194671e-06,
+      "loss": 0.0962,
+      "step": 20081
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.725581642392707e-06,
+      "loss": 0.1055,
+      "step": 20082
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.724639907119054e-06,
+      "loss": 0.0886,
+      "step": 20083
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.723698204383067e-06,
+      "loss": 0.1168,
+      "step": 20084
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.722756534194098e-06,
+      "loss": 0.1258,
+      "step": 20085
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.721814896561503e-06,
+      "loss": 0.1236,
+      "step": 20086
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.720873291494635e-06,
+      "loss": 0.1059,
+      "step": 20087
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.719931719002852e-06,
+      "loss": 0.1366,
+      "step": 20088
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7189901790955085e-06,
+      "loss": 0.1114,
+      "step": 20089
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.718048671781949e-06,
+      "loss": 0.0955,
+      "step": 20090
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.717107197071534e-06,
+      "loss": 0.1143,
+      "step": 20091
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.716165754973614e-06,
+      "loss": 0.1275,
+      "step": 20092
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.715224345497542e-06,
+      "loss": 0.1525,
+      "step": 20093
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.714282968652672e-06,
+      "loss": 0.1263,
+      "step": 20094
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.713341624448351e-06,
+      "loss": 0.085,
+      "step": 20095
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7124003128939295e-06,
+      "loss": 0.1046,
+      "step": 20096
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.711459033998766e-06,
+      "loss": 0.1612,
+      "step": 20097
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.710517787772206e-06,
+      "loss": 0.1179,
+      "step": 20098
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7095765742236015e-06,
+      "loss": 0.1005,
+      "step": 20099
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.708635393362303e-06,
+      "loss": 0.0901,
+      "step": 20100
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.707694245197658e-06,
+      "loss": 0.1096,
+      "step": 20101
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.706753129739016e-06,
+      "loss": 0.1336,
+      "step": 20102
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.705812046995731e-06,
+      "loss": 0.123,
+      "step": 20103
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.704870996977145e-06,
+      "loss": 0.1218,
+      "step": 20104
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7039299796926086e-06,
+      "loss": 0.1026,
+      "step": 20105
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7029889951514715e-06,
+      "loss": 0.1015,
+      "step": 20106
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.702048043363083e-06,
+      "loss": 0.1115,
+      "step": 20107
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.701107124336789e-06,
+      "loss": 0.1352,
+      "step": 20108
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.700166238081935e-06,
+      "loss": 0.0983,
+      "step": 20109
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.699225384607868e-06,
+      "loss": 0.1134,
+      "step": 20110
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.698284563923936e-06,
+      "loss": 0.1126,
+      "step": 20111
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.6973437760394825e-06,
+      "loss": 0.1381,
+      "step": 20112
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.696403020963862e-06,
+      "loss": 0.1245,
+      "step": 20113
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.695462298706407e-06,
+      "loss": 0.1104,
+      "step": 20114
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.694521609276471e-06,
+      "loss": 0.1154,
+      "step": 20115
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.693580952683398e-06,
+      "loss": 0.1101,
+      "step": 20116
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.69264032893653e-06,
+      "loss": 0.0882,
+      "step": 20117
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6916997380452175e-06,
+      "loss": 0.1002,
+      "step": 20118
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.690759180018797e-06,
+      "loss": 0.1345,
+      "step": 20119
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.689818654866613e-06,
+      "loss": 0.0792,
+      "step": 20120
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.688878162598012e-06,
+      "loss": 0.1357,
+      "step": 20121
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.687937703222334e-06,
+      "loss": 0.1092,
+      "step": 20122
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6869972767489274e-06,
+      "loss": 0.0943,
+      "step": 20123
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.686056883187127e-06,
+      "loss": 0.1077,
+      "step": 20124
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.68511652254628e-06,
+      "loss": 0.0955,
+      "step": 20125
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6841761948357235e-06,
+      "loss": 0.1267,
+      "step": 20126
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.683235900064805e-06,
+      "loss": 0.0936,
+      "step": 20127
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.68229563824286e-06,
+      "loss": 0.0712,
+      "step": 20128
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.681355409379229e-06,
+      "loss": 0.1435,
+      "step": 20129
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.680415213483252e-06,
+      "loss": 0.1178,
+      "step": 20130
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679475050564275e-06,
+      "loss": 0.1244,
+      "step": 20131
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.678534920631634e-06,
+      "loss": 0.1231,
+      "step": 20132
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.677594823694668e-06,
+      "loss": 0.1229,
+      "step": 20133
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.676654759762714e-06,
+      "loss": 0.132,
+      "step": 20134
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.675714728845114e-06,
+      "loss": 0.0903,
+      "step": 20135
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.674774730951205e-06,
+      "loss": 0.1382,
+      "step": 20136
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6738347660903255e-06,
+      "loss": 0.1121,
+      "step": 20137
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.672894834271813e-06,
+      "loss": 0.1445,
+      "step": 20138
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.671954935505001e-06,
+      "loss": 0.1156,
+      "step": 20139
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.671015069799234e-06,
+      "loss": 0.0924,
+      "step": 20140
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.670075237163842e-06,
+      "loss": 0.141,
+      "step": 20141
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.669135437608169e-06,
+      "loss": 0.1001,
+      "step": 20142
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.668195671141542e-06,
+      "loss": 0.0841,
+      "step": 20143
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.667255937773302e-06,
+      "loss": 0.1293,
+      "step": 20144
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.666316237512784e-06,
+      "loss": 0.1061,
+      "step": 20145
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.66537657036932e-06,
+      "loss": 0.1136,
+      "step": 20146
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6644369363522525e-06,
+      "loss": 0.0956,
+      "step": 20147
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.663497335470903e-06,
+      "loss": 0.1004,
+      "step": 20148
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6625577677346175e-06,
+      "loss": 0.1184,
+      "step": 20149
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6616182331527245e-06,
+      "loss": 0.1176,
+      "step": 20150
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.660678731734558e-06,
+      "loss": 0.1076,
+      "step": 20151
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6597392634894535e-06,
+      "loss": 0.1023,
+      "step": 20152
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.658799828426739e-06,
+      "loss": 0.1074,
+      "step": 20153
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.65786042655575e-06,
+      "loss": 0.0991,
+      "step": 20154
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.656921057885815e-06,
+      "loss": 0.1076,
+      "step": 20155
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.655981722426277e-06,
+      "loss": 0.1145,
+      "step": 20156
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6550424201864505e-06,
+      "loss": 0.0969,
+      "step": 20157
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.654103151175679e-06,
+      "loss": 0.0897,
+      "step": 20158
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.653163915403289e-06,
+      "loss": 0.0886,
+      "step": 20159
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.652224712878611e-06,
+      "loss": 0.0748,
+      "step": 20160
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.651285543610979e-06,
+      "loss": 0.112,
+      "step": 20161
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6503464076097155e-06,
+      "loss": 0.1239,
+      "step": 20162
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.649407304884155e-06,
+      "loss": 0.0997,
+      "step": 20163
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.648468235443623e-06,
+      "loss": 0.0947,
+      "step": 20164
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6475291992974534e-06,
+      "loss": 0.104,
+      "step": 20165
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.646590196454973e-06,
+      "loss": 0.1638,
+      "step": 20166
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.645651226925507e-06,
+      "loss": 0.1278,
+      "step": 20167
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.644712290718385e-06,
+      "loss": 0.1103,
+      "step": 20168
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.643773387842936e-06,
+      "loss": 0.1062,
+      "step": 20169
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.642834518308481e-06,
+      "loss": 0.1028,
+      "step": 20170
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.64189568212436e-06,
+      "loss": 0.1065,
+      "step": 20171
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.640956879299886e-06,
+      "loss": 0.1067,
+      "step": 20172
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.640018109844387e-06,
+      "loss": 0.1157,
+      "step": 20173
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.639079373767195e-06,
+      "loss": 0.0852,
+      "step": 20174
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.638140671077633e-06,
+      "loss": 0.1132,
+      "step": 20175
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.637202001785026e-06,
+      "loss": 0.1203,
+      "step": 20176
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.636263365898696e-06,
+      "loss": 0.1238,
+      "step": 20177
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.635324763427972e-06,
+      "loss": 0.0907,
+      "step": 20178
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.634386194382174e-06,
+      "loss": 0.1019,
+      "step": 20179
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.633447658770626e-06,
+      "loss": 0.0926,
+      "step": 20180
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.63250915660266e-06,
+      "loss": 0.1234,
+      "step": 20181
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.631570687887587e-06,
+      "loss": 0.1134,
+      "step": 20182
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6306322526347345e-06,
+      "loss": 0.1175,
+      "step": 20183
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.629693850853427e-06,
+      "loss": 0.1433,
+      "step": 20184
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.628755482552989e-06,
+      "loss": 0.1059,
+      "step": 20185
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.627817147742734e-06,
+      "loss": 0.0771,
+      "step": 20186
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.62687884643199e-06,
+      "loss": 0.1623,
+      "step": 20187
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.625940578630075e-06,
+      "loss": 0.1004,
+      "step": 20188
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6250023443463096e-06,
+      "loss": 0.1609,
+      "step": 20189
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.624064143590021e-06,
+      "loss": 0.1115,
+      "step": 20190
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.623125976370519e-06,
+      "loss": 0.1419,
+      "step": 20191
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.622187842697131e-06,
+      "loss": 0.1001,
+      "step": 20192
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.621249742579174e-06,
+      "loss": 0.1263,
+      "step": 20193
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.620311676025965e-06,
+      "loss": 0.1175,
+      "step": 20194
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.619373643046829e-06,
+      "loss": 0.099,
+      "step": 20195
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6184356436510785e-06,
+      "loss": 0.1128,
+      "step": 20196
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.617497677848033e-06,
+      "loss": 0.0973,
+      "step": 20197
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.616559745647009e-06,
+      "loss": 0.1107,
+      "step": 20198
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.615621847057329e-06,
+      "loss": 0.1221,
+      "step": 20199
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.614683982088308e-06,
+      "loss": 0.1037,
+      "step": 20200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6137461507492606e-06,
+      "loss": 0.1083,
+      "step": 20201
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6128083530495046e-06,
+      "loss": 0.1337,
+      "step": 20202
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.611870588998355e-06,
+      "loss": 0.132,
+      "step": 20203
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.610932858605128e-06,
+      "loss": 0.1204,
+      "step": 20204
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.609995161879146e-06,
+      "loss": 0.1208,
+      "step": 20205
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.609057498829712e-06,
+      "loss": 0.0977,
+      "step": 20206
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.608119869466147e-06,
+      "loss": 0.0949,
+      "step": 20207
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6071822737977665e-06,
+      "loss": 0.0834,
+      "step": 20208
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.606244711833883e-06,
+      "loss": 0.1129,
+      "step": 20209
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.605307183583814e-06,
+      "loss": 0.0963,
+      "step": 20210
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.604369689056867e-06,
+      "loss": 0.1419,
+      "step": 20211
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.603432228262357e-06,
+      "loss": 0.0951,
+      "step": 20212
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.602494801209597e-06,
+      "loss": 0.1089,
+      "step": 20213
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.601557407907906e-06,
+      "loss": 0.1313,
+      "step": 20214
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.600620048366586e-06,
+      "loss": 0.1027,
+      "step": 20215
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.599682722594951e-06,
+      "loss": 0.1224,
+      "step": 20216
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.598745430602316e-06,
+      "loss": 0.0864,
+      "step": 20217
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5978081723979905e-06,
+      "loss": 0.1061,
+      "step": 20218
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.596870947991289e-06,
+      "loss": 0.1307,
+      "step": 20219
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.595933757391515e-06,
+      "loss": 0.1179,
+      "step": 20220
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.594996600607982e-06,
+      "loss": 0.1033,
+      "step": 20221
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.594059477649999e-06,
+      "loss": 0.0935,
+      "step": 20222
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.593122388526878e-06,
+      "loss": 0.1412,
+      "step": 20223
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.592185333247929e-06,
+      "loss": 0.0933,
+      "step": 20224
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.591248311822454e-06,
+      "loss": 0.1486,
+      "step": 20225
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.590311324259766e-06,
+      "loss": 0.1095,
+      "step": 20226
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.589374370569172e-06,
+      "loss": 0.0994,
+      "step": 20227
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.588437450759984e-06,
+      "loss": 0.0838,
+      "step": 20228
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.587500564841504e-06,
+      "loss": 0.1545,
+      "step": 20229
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.586563712823042e-06,
+      "loss": 0.0839,
+      "step": 20230
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.585626894713899e-06,
+      "loss": 0.0937,
+      "step": 20231
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.584690110523391e-06,
+      "loss": 0.121,
+      "step": 20232
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5837533602608165e-06,
+      "loss": 0.129,
+      "step": 20233
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.582816643935488e-06,
+      "loss": 0.1168,
+      "step": 20234
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.581879961556703e-06,
+      "loss": 0.1481,
+      "step": 20235
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.580943313133772e-06,
+      "loss": 0.1052,
+      "step": 20236
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.580006698675998e-06,
+      "loss": 0.1407,
+      "step": 20237
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.579070118192683e-06,
+      "loss": 0.1319,
+      "step": 20238
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5781335716931394e-06,
+      "loss": 0.1316,
+      "step": 20239
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.577197059186662e-06,
+      "loss": 0.1742,
+      "step": 20240
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.576260580682554e-06,
+      "loss": 0.1131,
+      "step": 20241
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.575324136190126e-06,
+      "loss": 0.118,
+      "step": 20242
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5743877257186765e-06,
+      "loss": 0.1042,
+      "step": 20243
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.573451349277507e-06,
+      "loss": 0.1374,
+      "step": 20244
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.572515006875919e-06,
+      "loss": 0.0927,
+      "step": 20245
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.571578698523218e-06,
+      "loss": 0.0827,
+      "step": 20246
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.570642424228699e-06,
+      "loss": 0.0734,
+      "step": 20247
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.569706184001674e-06,
+      "loss": 0.1202,
+      "step": 20248
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.568769977851433e-06,
+      "loss": 0.1062,
+      "step": 20249
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.567833805787276e-06,
+      "loss": 0.1027,
+      "step": 20250
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.566897667818511e-06,
+      "loss": 0.0856,
+      "step": 20251
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5659615639544316e-06,
+      "loss": 0.1105,
+      "step": 20252
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.565025494204343e-06,
+      "loss": 0.0702,
+      "step": 20253
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.564089458577536e-06,
+      "loss": 0.0864,
+      "step": 20254
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.563153457083315e-06,
+      "loss": 0.1337,
+      "step": 20255
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.562217489730976e-06,
+      "loss": 0.1216,
+      "step": 20256
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.56128155652982e-06,
+      "loss": 0.0965,
+      "step": 20257
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.560345657489144e-06,
+      "loss": 0.1374,
+      "step": 20258
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.55940979261824e-06,
+      "loss": 0.1118,
+      "step": 20259
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.558473961926411e-06,
+      "loss": 0.0856,
+      "step": 20260
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5575381654229496e-06,
+      "loss": 0.146,
+      "step": 20261
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5566024031171535e-06,
+      "loss": 0.1031,
+      "step": 20262
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.555666675018323e-06,
+      "loss": 0.1073,
+      "step": 20263
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.554730981135746e-06,
+      "loss": 0.1093,
+      "step": 20264
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.553795321478719e-06,
+      "loss": 0.1126,
+      "step": 20265
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5528596960565436e-06,
+      "loss": 0.1037,
+      "step": 20266
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.551924104878513e-06,
+      "loss": 0.1107,
+      "step": 20267
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.550988547953911e-06,
+      "loss": 0.1206,
+      "step": 20268
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.550053025292044e-06,
+      "loss": 0.1084,
+      "step": 20269
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.549117536902198e-06,
+      "loss": 0.0921,
+      "step": 20270
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.54818208279367e-06,
+      "loss": 0.1278,
+      "step": 20271
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.547246662975752e-06,
+      "loss": 0.1153,
+      "step": 20272
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.546311277457736e-06,
+      "loss": 0.1057,
+      "step": 20273
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.545375926248911e-06,
+      "loss": 0.0848,
+      "step": 20274
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.544440609358575e-06,
+      "loss": 0.08,
+      "step": 20275
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.543505326796018e-06,
+      "loss": 0.1226,
+      "step": 20276
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.542570078570529e-06,
+      "loss": 0.1178,
+      "step": 20277
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5416348646913996e-06,
+      "loss": 0.1143,
+      "step": 20278
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.54069968516792e-06,
+      "loss": 0.1237,
+      "step": 20279
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.539764540009381e-06,
+      "loss": 0.1114,
+      "step": 20280
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.538829429225068e-06,
+      "loss": 0.0832,
+      "step": 20281
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.537894352824284e-06,
+      "loss": 0.0974,
+      "step": 20282
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5369593108162995e-06,
+      "loss": 0.1461,
+      "step": 20283
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.536024303210418e-06,
+      "loss": 0.1569,
+      "step": 20284
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.535089330015919e-06,
+      "loss": 0.1216,
+      "step": 20285
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.534154391242096e-06,
+      "loss": 0.155,
+      "step": 20286
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5332194868982345e-06,
+      "loss": 0.1135,
+      "step": 20287
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.532284616993622e-06,
+      "loss": 0.0859,
+      "step": 20288
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.531349781537546e-06,
+      "loss": 0.1028,
+      "step": 20289
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5304149805392905e-06,
+      "loss": 0.1109,
+      "step": 20290
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5294802140081446e-06,
+      "loss": 0.1177,
+      "step": 20291
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.528545481953398e-06,
+      "loss": 0.1078,
+      "step": 20292
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.52761078438433e-06,
+      "loss": 0.1165,
+      "step": 20293
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.526676121310228e-06,
+      "loss": 0.1147,
+      "step": 20294
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.525741492740378e-06,
+      "loss": 0.1229,
+      "step": 20295
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.524806898684066e-06,
+      "loss": 0.1466,
+      "step": 20296
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.523872339150572e-06,
+      "loss": 0.1262,
+      "step": 20297
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.522937814149182e-06,
+      "loss": 0.1241,
+      "step": 20298
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.522003323689179e-06,
+      "loss": 0.1294,
+      "step": 20299
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.521068867779849e-06,
+      "loss": 0.1082,
+      "step": 20300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.520134446430475e-06,
+      "loss": 0.0892,
+      "step": 20301
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.519200059650335e-06,
+      "loss": 0.1318,
+      "step": 20302
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.518265707448716e-06,
+      "loss": 0.147,
+      "step": 20303
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.517331389834896e-06,
+      "loss": 0.1197,
+      "step": 20304
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5163971068181596e-06,
+      "loss": 0.0921,
+      "step": 20305
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.515462858407791e-06,
+      "loss": 0.1057,
+      "step": 20306
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.514528644613063e-06,
+      "loss": 0.0736,
+      "step": 20307
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.513594465443258e-06,
+      "loss": 0.1124,
+      "step": 20308
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5126603209076614e-06,
+      "loss": 0.127,
+      "step": 20309
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.51172621101555e-06,
+      "loss": 0.0892,
+      "step": 20310
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.510792135776205e-06,
+      "loss": 0.1242,
+      "step": 20311
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.509858095198903e-06,
+      "loss": 0.1207,
+      "step": 20312
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5089240892929236e-06,
+      "loss": 0.152,
+      "step": 20313
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.507990118067545e-06,
+      "loss": 0.1009,
+      "step": 20314
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.507056181532045e-06,
+      "loss": 0.1081,
+      "step": 20315
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.506122279695708e-06,
+      "loss": 0.1015,
+      "step": 20316
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.505188412567799e-06,
+      "loss": 0.0948,
+      "step": 20317
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5042545801576054e-06,
+      "loss": 0.1497,
+      "step": 20318
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.503320782474398e-06,
+      "loss": 0.1281,
+      "step": 20319
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.502387019527458e-06,
+      "loss": 0.1168,
+      "step": 20320
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.501453291326059e-06,
+      "loss": 0.0844,
+      "step": 20321
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.500519597879476e-06,
+      "loss": 0.1128,
+      "step": 20322
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4995859391969855e-06,
+      "loss": 0.1264,
+      "step": 20323
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.498652315287859e-06,
+      "loss": 0.1095,
+      "step": 20324
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.497718726161382e-06,
+      "loss": 0.0824,
+      "step": 20325
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.496785171826814e-06,
+      "loss": 0.1087,
+      "step": 20326
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.495851652293439e-06,
+      "loss": 0.1125,
+      "step": 20327
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.494918167570527e-06,
+      "loss": 0.1088,
+      "step": 20328
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.493984717667354e-06,
+      "loss": 0.1398,
+      "step": 20329
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.493051302593192e-06,
+      "loss": 0.1255,
+      "step": 20330
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.49211792235731e-06,
+      "loss": 0.1127,
+      "step": 20331
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.491184576968985e-06,
+      "loss": 0.1562,
+      "step": 20332
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.490251266437484e-06,
+      "loss": 0.1165,
+      "step": 20333
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4893179907720825e-06,
+      "loss": 0.0912,
+      "step": 20334
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.488384749982054e-06,
+      "loss": 0.0917,
+      "step": 20335
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.487451544076664e-06,
+      "loss": 0.1145,
+      "step": 20336
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.486518373065185e-06,
+      "loss": 0.1295,
+      "step": 20337
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4855852369568885e-06,
+      "loss": 0.1061,
+      "step": 20338
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.484652135761042e-06,
+      "loss": 0.101,
+      "step": 20339
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.483719069486918e-06,
+      "loss": 0.104,
+      "step": 20340
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4827860381437825e-06,
+      "loss": 0.0977,
+      "step": 20341
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.481853041740903e-06,
+      "loss": 0.0919,
+      "step": 20342
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.480920080287554e-06,
+      "loss": 0.1196,
+      "step": 20343
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.479987153792998e-06,
+      "loss": 0.0918,
+      "step": 20344
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.479054262266509e-06,
+      "loss": 0.1335,
+      "step": 20345
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.478121405717348e-06,
+      "loss": 0.0973,
+      "step": 20346
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.477188584154783e-06,
+      "loss": 0.0937,
+      "step": 20347
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.476255797588085e-06,
+      "loss": 0.1275,
+      "step": 20348
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.475323046026513e-06,
+      "loss": 0.1071,
+      "step": 20349
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4743903294793454e-06,
+      "loss": 0.1182,
+      "step": 20350
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.473457647955832e-06,
+      "loss": 0.089,
+      "step": 20351
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.47252500146525e-06,
+      "loss": 0.1136,
+      "step": 20352
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.471592390016859e-06,
+      "loss": 0.1151,
+      "step": 20353
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4706598136199285e-06,
+      "loss": 0.0846,
+      "step": 20354
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.469727272283717e-06,
+      "loss": 0.1102,
+      "step": 20355
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.468794766017491e-06,
+      "loss": 0.1586,
+      "step": 20356
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4678622948305134e-06,
+      "loss": 0.1505,
+      "step": 20357
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.466929858732047e-06,
+      "loss": 0.074,
+      "step": 20358
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.465997457731361e-06,
+      "loss": 0.1063,
+      "step": 20359
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.465065091837708e-06,
+      "loss": 0.1189,
+      "step": 20360
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4641327610603556e-06,
+      "loss": 0.097,
+      "step": 20361
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.463200465408564e-06,
+      "loss": 0.1071,
+      "step": 20362
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.462268204891597e-06,
+      "loss": 0.1621,
+      "step": 20363
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.461335979518716e-06,
+      "loss": 0.089,
+      "step": 20364
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.460403789299178e-06,
+      "loss": 0.0884,
+      "step": 20365
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.459471634242245e-06,
+      "loss": 0.0969,
+      "step": 20366
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.458539514357176e-06,
+      "loss": 0.0954,
+      "step": 20367
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.457607429653236e-06,
+      "loss": 0.119,
+      "step": 20368
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.456675380139681e-06,
+      "loss": 0.0987,
+      "step": 20369
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4557433658257686e-06,
+      "loss": 0.0974,
+      "step": 20370
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.454811386720758e-06,
+      "loss": 0.1159,
+      "step": 20371
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.453879442833908e-06,
+      "loss": 0.1074,
+      "step": 20372
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4529475341744764e-06,
+      "loss": 0.1038,
+      "step": 20373
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4520156607517255e-06,
+      "loss": 0.1209,
+      "step": 20374
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.451083822574907e-06,
+      "loss": 0.1131,
+      "step": 20375
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.450152019653275e-06,
+      "loss": 0.1046,
+      "step": 20376
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.449220251996094e-06,
+      "loss": 0.1116,
+      "step": 20377
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.448288519612614e-06,
+      "loss": 0.119,
+      "step": 20378
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.447356822512097e-06,
+      "loss": 0.1053,
+      "step": 20379
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.446425160703794e-06,
+      "loss": 0.1557,
+      "step": 20380
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4454935341969606e-06,
+      "loss": 0.1054,
+      "step": 20381
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.44456194300085e-06,
+      "loss": 0.1346,
+      "step": 20382
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.443630387124726e-06,
+      "loss": 0.0997,
+      "step": 20383
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.44269886657783e-06,
+      "loss": 0.0858,
+      "step": 20384
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.441767381369422e-06,
+      "loss": 0.147,
+      "step": 20385
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.440835931508755e-06,
+      "loss": 0.1061,
+      "step": 20386
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.439904517005082e-06,
+      "loss": 0.0857,
+      "step": 20387
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.438973137867659e-06,
+      "loss": 0.1017,
+      "step": 20388
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.438041794105732e-06,
+      "loss": 0.096,
+      "step": 20389
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.437110485728557e-06,
+      "loss": 0.123,
+      "step": 20390
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.436179212745384e-06,
+      "loss": 0.1066,
+      "step": 20391
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.435247975165465e-06,
+      "loss": 0.0914,
+      "step": 20392
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.434316772998057e-06,
+      "loss": 0.1773,
+      "step": 20393
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.433385606252398e-06,
+      "loss": 0.0886,
+      "step": 20394
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.432454474937747e-06,
+      "loss": 0.1616,
+      "step": 20395
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4315233790633525e-06,
+      "loss": 0.1029,
+      "step": 20396
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.430592318638463e-06,
+      "loss": 0.1106,
+      "step": 20397
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.429661293672331e-06,
+      "loss": 0.1772,
+      "step": 20398
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.428730304174201e-06,
+      "loss": 0.0994,
+      "step": 20399
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.427799350153321e-06,
+      "loss": 0.1594,
+      "step": 20400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.426868431618944e-06,
+      "loss": 0.0931,
+      "step": 20401
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.425937548580314e-06,
+      "loss": 0.1035,
+      "step": 20402
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4250067010466825e-06,
+      "loss": 0.0978,
+      "step": 20403
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4240758890272915e-06,
+      "loss": 0.1087,
+      "step": 20404
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.42314511253139e-06,
+      "loss": 0.0805,
+      "step": 20405
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.422214371568225e-06,
+      "loss": 0.1103,
+      "step": 20406
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.421283666147039e-06,
+      "loss": 0.1022,
+      "step": 20407
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.420352996277087e-06,
+      "loss": 0.1433,
+      "step": 20408
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.419422361967605e-06,
+      "loss": 0.1041,
+      "step": 20409
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.418491763227838e-06,
+      "loss": 0.1219,
+      "step": 20410
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.417561200067035e-06,
+      "loss": 0.1161,
+      "step": 20411
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.416630672494443e-06,
+      "loss": 0.1099,
+      "step": 20412
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.415700180519299e-06,
+      "loss": 0.1281,
+      "step": 20413
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.414769724150849e-06,
+      "loss": 0.0846,
+      "step": 20414
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.413839303398337e-06,
+      "loss": 0.1326,
+      "step": 20415
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4129089182710035e-06,
+      "loss": 0.1025,
+      "step": 20416
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.411978568778099e-06,
+      "loss": 0.1169,
+      "step": 20417
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4110482549288555e-06,
+      "loss": 0.1139,
+      "step": 20418
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.410117976732517e-06,
+      "loss": 0.0875,
+      "step": 20419
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4091877341983295e-06,
+      "loss": 0.0816,
+      "step": 20420
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.408257527335531e-06,
+      "loss": 0.1113,
+      "step": 20421
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.407327356153366e-06,
+      "loss": 0.1246,
+      "step": 20422
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.406397220661068e-06,
+      "loss": 0.1214,
+      "step": 20423
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.405467120867881e-06,
+      "loss": 0.0833,
+      "step": 20424
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.404537056783042e-06,
+      "loss": 0.113,
+      "step": 20425
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.403607028415797e-06,
+      "loss": 0.1002,
+      "step": 20426
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.402677035775383e-06,
+      "loss": 0.1026,
+      "step": 20427
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.40174707887103e-06,
+      "loss": 0.0953,
+      "step": 20428
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.400817157711986e-06,
+      "loss": 0.1313,
+      "step": 20429
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.399887272307485e-06,
+      "loss": 0.1014,
+      "step": 20430
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.398957422666765e-06,
+      "loss": 0.0998,
+      "step": 20431
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.398027608799064e-06,
+      "loss": 0.0987,
+      "step": 20432
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.397097830713617e-06,
+      "loss": 0.0958,
+      "step": 20433
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.39616808841966e-06,
+      "loss": 0.0888,
+      "step": 20434
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.395238381926432e-06,
+      "loss": 0.1465,
+      "step": 20435
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.394308711243171e-06,
+      "loss": 0.1161,
+      "step": 20436
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.3933790763791035e-06,
+      "loss": 0.0951,
+      "step": 20437
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.392449477343472e-06,
+      "loss": 0.0979,
+      "step": 20438
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.39151991414551e-06,
+      "loss": 0.114,
+      "step": 20439
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.390590386794448e-06,
+      "loss": 0.0991,
+      "step": 20440
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.389660895299528e-06,
+      "loss": 0.1566,
+      "step": 20441
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.388731439669975e-06,
+      "loss": 0.1252,
+      "step": 20442
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.387802019915026e-06,
+      "loss": 0.1241,
+      "step": 20443
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.386872636043913e-06,
+      "loss": 0.1418,
+      "step": 20444
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.385943288065871e-06,
+      "loss": 0.0904,
+      "step": 20445
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.385013975990133e-06,
+      "loss": 0.0877,
+      "step": 20446
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.384084699825925e-06,
+      "loss": 0.0946,
+      "step": 20447
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.383155459582482e-06,
+      "loss": 0.0907,
+      "step": 20448
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.382226255269036e-06,
+      "loss": 0.0966,
+      "step": 20449
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.381297086894814e-06,
+      "loss": 0.0942,
+      "step": 20450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.380367954469056e-06,
+      "loss": 0.1141,
+      "step": 20451
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.37943885800098e-06,
+      "loss": 0.114,
+      "step": 20452
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.378509797499821e-06,
+      "loss": 0.1103,
+      "step": 20453
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.377580772974809e-06,
+      "loss": 0.1184,
+      "step": 20454
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.376651784435174e-06,
+      "loss": 0.1375,
+      "step": 20455
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.375722831890143e-06,
+      "loss": 0.1333,
+      "step": 20456
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.374793915348945e-06,
+      "loss": 0.1173,
+      "step": 20457
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3738650348208055e-06,
+      "loss": 0.1185,
+      "step": 20458
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.372936190314951e-06,
+      "loss": 0.1384,
+      "step": 20459
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.372007381840615e-06,
+      "loss": 0.127,
+      "step": 20460
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.371078609407022e-06,
+      "loss": 0.1059,
+      "step": 20461
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.370149873023394e-06,
+      "loss": 0.1027,
+      "step": 20462
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.369221172698963e-06,
+      "loss": 0.1171,
+      "step": 20463
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3682925084429525e-06,
+      "loss": 0.1141,
+      "step": 20464
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.367363880264587e-06,
+      "loss": 0.1711,
+      "step": 20465
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.366435288173093e-06,
+      "loss": 0.1338,
+      "step": 20466
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.365506732177693e-06,
+      "loss": 0.1069,
+      "step": 20467
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.364578212287611e-06,
+      "loss": 0.1279,
+      "step": 20468
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.363649728512076e-06,
+      "loss": 0.1314,
+      "step": 20469
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.362721280860311e-06,
+      "loss": 0.1179,
+      "step": 20470
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3617928693415306e-06,
+      "loss": 0.1045,
+      "step": 20471
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360864493964966e-06,
+      "loss": 0.0845,
+      "step": 20472
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.359936154739839e-06,
+      "loss": 0.1164,
+      "step": 20473
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.359007851675368e-06,
+      "loss": 0.1038,
+      "step": 20474
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.358079584780781e-06,
+      "loss": 0.1152,
+      "step": 20475
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.357151354065293e-06,
+      "loss": 0.124,
+      "step": 20476
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.356223159538125e-06,
+      "loss": 0.1054,
+      "step": 20477
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.355295001208504e-06,
+      "loss": 0.1216,
+      "step": 20478
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.354366879085645e-06,
+      "loss": 0.1485,
+      "step": 20479
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.353438793178774e-06,
+      "loss": 0.1148,
+      "step": 20480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.352510743497104e-06,
+      "loss": 0.1085,
+      "step": 20481
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.351582730049856e-06,
+      "loss": 0.0873,
+      "step": 20482
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3506547528462505e-06,
+      "loss": 0.1304,
+      "step": 20483
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.349726811895502e-06,
+      "loss": 0.0996,
+      "step": 20484
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.34879890720684e-06,
+      "loss": 0.1373,
+      "step": 20485
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.347871038789469e-06,
+      "loss": 0.1457,
+      "step": 20486
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.346943206652611e-06,
+      "loss": 0.0881,
+      "step": 20487
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.346015410805487e-06,
+      "loss": 0.1223,
+      "step": 20488
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.34508765125731e-06,
+      "loss": 0.0769,
+      "step": 20489
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3441599280172995e-06,
+      "loss": 0.1167,
+      "step": 20490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.343232241094668e-06,
+      "loss": 0.1141,
+      "step": 20491
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.342304590498632e-06,
+      "loss": 0.0814,
+      "step": 20492
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.341376976238407e-06,
+      "loss": 0.0949,
+      "step": 20493
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.340449398323213e-06,
+      "loss": 0.069,
+      "step": 20494
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.339521856762254e-06,
+      "loss": 0.1038,
+      "step": 20495
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.338594351564754e-06,
+      "loss": 0.0905,
+      "step": 20496
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3376668827399225e-06,
+      "loss": 0.0959,
+      "step": 20497
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3367394502969736e-06,
+      "loss": 0.1418,
+      "step": 20498
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.335812054245123e-06,
+      "loss": 0.1516,
+      "step": 20499
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.334884694593579e-06,
+      "loss": 0.1216,
+      "step": 20500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.333957371351558e-06,
+      "loss": 0.0912,
+      "step": 20501
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.333030084528267e-06,
+      "loss": 0.1025,
+      "step": 20502
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.332102834132923e-06,
+      "loss": 0.1097,
+      "step": 20503
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3311756201747375e-06,
+      "loss": 0.073,
+      "step": 20504
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.330248442662919e-06,
+      "loss": 0.1326,
+      "step": 20505
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.329321301606677e-06,
+      "loss": 0.1094,
+      "step": 20506
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.328394197015224e-06,
+      "loss": 0.1194,
+      "step": 20507
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.327467128897769e-06,
+      "loss": 0.1235,
+      "step": 20508
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.326540097263525e-06,
+      "loss": 0.1193,
+      "step": 20509
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.325613102121695e-06,
+      "loss": 0.0885,
+      "step": 20510
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.32468614348149e-06,
+      "loss": 0.1426,
+      "step": 20511
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3237592213521215e-06,
+      "loss": 0.1093,
+      "step": 20512
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3228323357427945e-06,
+      "loss": 0.1279,
+      "step": 20513
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.321905486662721e-06,
+      "loss": 0.124,
+      "step": 20514
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.320978674121103e-06,
+      "loss": 0.1101,
+      "step": 20515
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.320051898127151e-06,
+      "loss": 0.1131,
+      "step": 20516
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.319125158690068e-06,
+      "loss": 0.0995,
+      "step": 20517
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.318198455819062e-06,
+      "loss": 0.1225,
+      "step": 20518
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.317271789523345e-06,
+      "loss": 0.1194,
+      "step": 20519
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3163451598121115e-06,
+      "loss": 0.1358,
+      "step": 20520
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.315418566694576e-06,
+      "loss": 0.1399,
+      "step": 20521
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.314492010179938e-06,
+      "loss": 0.1019,
+      "step": 20522
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.313565490277407e-06,
+      "loss": 0.0804,
+      "step": 20523
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31263900699618e-06,
+      "loss": 0.1405,
+      "step": 20524
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3117125603454665e-06,
+      "loss": 0.1449,
+      "step": 20525
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.310786150334467e-06,
+      "loss": 0.131,
+      "step": 20526
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.309859776972382e-06,
+      "loss": 0.1111,
+      "step": 20527
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3089334402684265e-06,
+      "loss": 0.1167,
+      "step": 20528
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.308007140231787e-06,
+      "loss": 0.1144,
+      "step": 20529
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.307080876871674e-06,
+      "loss": 0.1403,
+      "step": 20530
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.306154650197288e-06,
+      "loss": 0.1246,
+      "step": 20531
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3052284602178295e-06,
+      "loss": 0.1248,
+      "step": 20532
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3043023069425025e-06,
+      "loss": 0.1284,
+      "step": 20533
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3033761903805015e-06,
+      "loss": 0.0699,
+      "step": 20534
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3024501105410295e-06,
+      "loss": 0.0996,
+      "step": 20535
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.301524067433284e-06,
+      "loss": 0.1415,
+      "step": 20536
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.30059806106647e-06,
+      "loss": 0.1046,
+      "step": 20537
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.299672091449784e-06,
+      "loss": 0.0831,
+      "step": 20538
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.298746158592423e-06,
+      "loss": 0.1189,
+      "step": 20539
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.297820262503584e-06,
+      "loss": 0.1082,
+      "step": 20540
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2968944031924695e-06,
+      "loss": 0.1084,
+      "step": 20541
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.295968580668272e-06,
+      "loss": 0.0855,
+      "step": 20542
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.295042794940197e-06,
+      "loss": 0.1377,
+      "step": 20543
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.294117046017432e-06,
+      "loss": 0.112,
+      "step": 20544
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2931913339091755e-06,
+      "loss": 0.1215,
+      "step": 20545
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.292265658624626e-06,
+      "loss": 0.121,
+      "step": 20546
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29134002017298e-06,
+      "loss": 0.1303,
+      "step": 20547
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.290414418563433e-06,
+      "loss": 0.1048,
+      "step": 20548
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.289488853805175e-06,
+      "loss": 0.1116,
+      "step": 20549
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2885633259074044e-06,
+      "loss": 0.1589,
+      "step": 20550
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.287637834879315e-06,
+      "loss": 0.1012,
+      "step": 20551
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.286712380730104e-06,
+      "loss": 0.147,
+      "step": 20552
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.285786963468958e-06,
+      "loss": 0.0894,
+      "step": 20553
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2848615831050726e-06,
+      "loss": 0.0996,
+      "step": 20554
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.283936239647643e-06,
+      "loss": 0.1053,
+      "step": 20555
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.283010933105862e-06,
+      "loss": 0.1136,
+      "step": 20556
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.282085663488921e-06,
+      "loss": 0.1148,
+      "step": 20557
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2811604308060085e-06,
+      "loss": 0.0926,
+      "step": 20558
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.280235235066319e-06,
+      "loss": 0.1391,
+      "step": 20559
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.27931007627904e-06,
+      "loss": 0.1092,
+      "step": 20560
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.278384954453365e-06,
+      "loss": 0.1032,
+      "step": 20561
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.27745986959849e-06,
+      "loss": 0.1046,
+      "step": 20562
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.276534821723592e-06,
+      "loss": 0.1213,
+      "step": 20563
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.275609810837869e-06,
+      "loss": 0.0803,
+      "step": 20564
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.274684836950509e-06,
+      "loss": 0.1105,
+      "step": 20565
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.273759900070699e-06,
+      "loss": 0.1037,
+      "step": 20566
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.272835000207632e-06,
+      "loss": 0.1194,
+      "step": 20567
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2719101373704895e-06,
+      "loss": 0.1207,
+      "step": 20568
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.270985311568463e-06,
+      "loss": 0.1082,
+      "step": 20569
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.270060522810736e-06,
+      "loss": 0.1692,
+      "step": 20570
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2691357711065e-06,
+      "loss": 0.1014,
+      "step": 20571
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.268211056464944e-06,
+      "loss": 0.1329,
+      "step": 20572
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2672863788952465e-06,
+      "loss": 0.103,
+      "step": 20573
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.266361738406597e-06,
+      "loss": 0.1237,
+      "step": 20574
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.265437135008181e-06,
+      "loss": 0.0908,
+      "step": 20575
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.264512568709182e-06,
+      "loss": 0.0731,
+      "step": 20576
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.263588039518791e-06,
+      "loss": 0.0903,
+      "step": 20577
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.262663547446184e-06,
+      "loss": 0.1029,
+      "step": 20578
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.261739092500547e-06,
+      "loss": 0.1356,
+      "step": 20579
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.260814674691067e-06,
+      "loss": 0.0872,
+      "step": 20580
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.259890294026927e-06,
+      "loss": 0.118,
+      "step": 20581
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.258965950517306e-06,
+      "loss": 0.0827,
+      "step": 20582
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2580416441713886e-06,
+      "loss": 0.1136,
+      "step": 20583
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.257117374998358e-06,
+      "loss": 0.1393,
+      "step": 20584
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.256193143007392e-06,
+      "loss": 0.1096,
+      "step": 20585
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.25526894820768e-06,
+      "loss": 0.1169,
+      "step": 20586
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2543447906083954e-06,
+      "loss": 0.117,
+      "step": 20587
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.253420670218719e-06,
+      "loss": 0.1536,
+      "step": 20588
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.252496587047836e-06,
+      "loss": 0.1232,
+      "step": 20589
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2515725411049235e-06,
+      "loss": 0.1701,
+      "step": 20590
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.250648532399163e-06,
+      "loss": 0.1077,
+      "step": 20591
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2497245609397295e-06,
+      "loss": 0.1174,
+      "step": 20592
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2488006267358045e-06,
+      "loss": 0.1118,
+      "step": 20593
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.247876729796565e-06,
+      "loss": 0.1242,
+      "step": 20594
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2469528701311935e-06,
+      "loss": 0.1271,
+      "step": 20595
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.246029047748866e-06,
+      "loss": 0.0892,
+      "step": 20596
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.245105262658752e-06,
+      "loss": 0.0907,
+      "step": 20597
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.244181514870039e-06,
+      "loss": 0.1227,
+      "step": 20598
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.243257804391898e-06,
+      "loss": 0.1436,
+      "step": 20599
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.242334131233507e-06,
+      "loss": 0.1052,
+      "step": 20600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.241410495404043e-06,
+      "loss": 0.1277,
+      "step": 20601
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2404868969126785e-06,
+      "loss": 0.1278,
+      "step": 20602
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.239563335768587e-06,
+      "loss": 0.1212,
+      "step": 20603
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.238639811980948e-06,
+      "loss": 0.1386,
+      "step": 20604
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.237716325558939e-06,
+      "loss": 0.1041,
+      "step": 20605
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.236792876511722e-06,
+      "loss": 0.1075,
+      "step": 20606
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.235869464848481e-06,
+      "loss": 0.1296,
+      "step": 20607
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.234946090578384e-06,
+      "loss": 0.0595,
+      "step": 20608
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.234022753710607e-06,
+      "loss": 0.0925,
+      "step": 20609
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.233099454254323e-06,
+      "loss": 0.1217,
+      "step": 20610
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2321761922187e-06,
+      "loss": 0.1566,
+      "step": 20611
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.231252967612911e-06,
+      "loss": 0.1208,
+      "step": 20612
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.230329780446129e-06,
+      "loss": 0.105,
+      "step": 20613
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.229406630727525e-06,
+      "loss": 0.0811,
+      "step": 20614
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.228483518466271e-06,
+      "loss": 0.1028,
+      "step": 20615
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2275604436715346e-06,
+      "loss": 0.0989,
+      "step": 20616
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2266374063524845e-06,
+      "loss": 0.123,
+      "step": 20617
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.225714406518293e-06,
+      "loss": 0.1038,
+      "step": 20618
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.224791444178127e-06,
+      "loss": 0.1369,
+      "step": 20619
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.223868519341161e-06,
+      "loss": 0.1496,
+      "step": 20620
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.222945632016555e-06,
+      "loss": 0.0694,
+      "step": 20621
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.222022782213482e-06,
+      "loss": 0.0862,
+      "step": 20622
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.221099969941109e-06,
+      "loss": 0.1222,
+      "step": 20623
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.220177195208604e-06,
+      "loss": 0.1125,
+      "step": 20624
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.219254458025132e-06,
+      "loss": 0.0992,
+      "step": 20625
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.218331758399862e-06,
+      "loss": 0.1221,
+      "step": 20626
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.217409096341957e-06,
+      "loss": 0.132,
+      "step": 20627
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.216486471860582e-06,
+      "loss": 0.1185,
+      "step": 20628
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.215563884964908e-06,
+      "loss": 0.1125,
+      "step": 20629
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2146413356641e-06,
+      "loss": 0.1512,
+      "step": 20630
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.213718823967314e-06,
+      "loss": 0.1172,
+      "step": 20631
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.212796349883723e-06,
+      "loss": 0.0841,
+      "step": 20632
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.211873913422486e-06,
+      "loss": 0.0898,
+      "step": 20633
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.210951514592771e-06,
+      "loss": 0.1139,
+      "step": 20634
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.210029153403738e-06,
+      "loss": 0.1093,
+      "step": 20635
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.209106829864548e-06,
+      "loss": 0.1288,
+      "step": 20636
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2081845439843654e-06,
+      "loss": 0.1023,
+      "step": 20637
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.207262295772354e-06,
+      "loss": 0.0911,
+      "step": 20638
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.206340085237679e-06,
+      "loss": 0.0965,
+      "step": 20639
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.20541791238949e-06,
+      "loss": 0.1516,
+      "step": 20640
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2044957772369565e-06,
+      "loss": 0.1404,
+      "step": 20641
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.203573679789239e-06,
+      "loss": 0.0932,
+      "step": 20642
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2026516200554935e-06,
+      "loss": 0.1056,
+      "step": 20643
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.201729598044885e-06,
+      "loss": 0.1222,
+      "step": 20644
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2008076137665694e-06,
+      "loss": 0.1157,
+      "step": 20645
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.199885667229704e-06,
+      "loss": 0.1562,
+      "step": 20646
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1989637584434526e-06,
+      "loss": 0.1316,
+      "step": 20647
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.19804188741697e-06,
+      "loss": 0.1371,
+      "step": 20648
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.197120054159418e-06,
+      "loss": 0.0894,
+      "step": 20649
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.196198258679948e-06,
+      "loss": 0.1073,
+      "step": 20650
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.195276500987721e-06,
+      "loss": 0.1127,
+      "step": 20651
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1943547810918935e-06,
+      "loss": 0.1192,
+      "step": 20652
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.19343309900162e-06,
+      "loss": 0.1251,
+      "step": 20653
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.192511454726065e-06,
+      "loss": 0.1229,
+      "step": 20654
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.191589848274369e-06,
+      "loss": 0.1578,
+      "step": 20655
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1906682796557e-06,
+      "loss": 0.109,
+      "step": 20656
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1897467488792075e-06,
+      "loss": 0.0953,
+      "step": 20657
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.188825255954048e-06,
+      "loss": 0.1161,
+      "step": 20658
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.187903800889377e-06,
+      "loss": 0.1393,
+      "step": 20659
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.186982383694342e-06,
+      "loss": 0.1091,
+      "step": 20660
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.186061004378103e-06,
+      "loss": 0.1332,
+      "step": 20661
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.185139662949808e-06,
+      "loss": 0.1179,
+      "step": 20662
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.184218359418619e-06,
+      "loss": 0.0888,
+      "step": 20663
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.183297093793674e-06,
+      "loss": 0.1219,
+      "step": 20664
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1823758660841356e-06,
+      "loss": 0.1023,
+      "step": 20665
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.181454676299152e-06,
+      "loss": 0.105,
+      "step": 20666
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.180533524447876e-06,
+      "loss": 0.1113,
+      "step": 20667
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1796124105394575e-06,
+      "loss": 0.0962,
+      "step": 20668
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.178691334583046e-06,
+      "loss": 0.1035,
+      "step": 20669
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.17777029658779e-06,
+      "loss": 0.1275,
+      "step": 20670
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1768492965628415e-06,
+      "loss": 0.1379,
+      "step": 20671
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.175928334517349e-06,
+      "loss": 0.1143,
+      "step": 20672
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1750074104604655e-06,
+      "loss": 0.0979,
+      "step": 20673
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.174086524401334e-06,
+      "loss": 0.1099,
+      "step": 20674
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.1067,
+      "step": 20675
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1722448663129235e-06,
+      "loss": 0.1053,
+      "step": 20676
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.17132409430194e-06,
+      "loss": 0.1152,
+      "step": 20677
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.170403360325304e-06,
+      "loss": 0.1017,
+      "step": 20678
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.169482664392156e-06,
+      "loss": 0.0869,
+      "step": 20679
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.168562006511643e-06,
+      "loss": 0.1037,
+      "step": 20680
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.167641386692915e-06,
+      "loss": 0.1286,
+      "step": 20681
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.166720804945115e-06,
+      "loss": 0.1386,
+      "step": 20682
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.165800261277393e-06,
+      "loss": 0.1021,
+      "step": 20683
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.164879755698885e-06,
+      "loss": 0.1178,
+      "step": 20684
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.163959288218741e-06,
+      "loss": 0.1205,
+      "step": 20685
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.163038858846102e-06,
+      "loss": 0.1209,
+      "step": 20686
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.162118467590113e-06,
+      "loss": 0.0855,
+      "step": 20687
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.161198114459923e-06,
+      "loss": 0.0876,
+      "step": 20688
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.160277799464663e-06,
+      "loss": 0.1278,
+      "step": 20689
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.159357522613485e-06,
+      "loss": 0.1236,
+      "step": 20690
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.158437283915528e-06,
+      "loss": 0.1294,
+      "step": 20691
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.157517083379935e-06,
+      "loss": 0.0993,
+      "step": 20692
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1565969210158435e-06,
+      "loss": 0.0805,
+      "step": 20693
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.155676796832398e-06,
+      "loss": 0.0855,
+      "step": 20694
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.154756710838739e-06,
+      "loss": 0.0903,
+      "step": 20695
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.153836663044002e-06,
+      "loss": 0.0834,
+      "step": 20696
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.152916653457337e-06,
+      "loss": 0.0749,
+      "step": 20697
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.151996682087872e-06,
+      "loss": 0.1241,
+      "step": 20698
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.151076748944752e-06,
+      "loss": 0.1193,
+      "step": 20699
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.150156854037115e-06,
+      "loss": 0.1009,
+      "step": 20700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1492369973740995e-06,
+      "loss": 0.1045,
+      "step": 20701
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.148317178964845e-06,
+      "loss": 0.1219,
+      "step": 20702
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.147397398818486e-06,
+      "loss": 0.1333,
+      "step": 20703
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1464776569441594e-06,
+      "loss": 0.0991,
+      "step": 20704
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.145557953351002e-06,
+      "loss": 0.1556,
+      "step": 20705
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1446382880481525e-06,
+      "loss": 0.1393,
+      "step": 20706
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1437186610447496e-06,
+      "loss": 0.1198,
+      "step": 20707
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.142799072349923e-06,
+      "loss": 0.1243,
+      "step": 20708
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.141879521972809e-06,
+      "loss": 0.1239,
+      "step": 20709
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.140960009922545e-06,
+      "loss": 0.1106,
+      "step": 20710
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.14004053620826e-06,
+      "loss": 0.0835,
+      "step": 20711
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1391211008391014e-06,
+      "loss": 0.1281,
+      "step": 20712
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.138201703824189e-06,
+      "loss": 0.0924,
+      "step": 20713
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.137282345172658e-06,
+      "loss": 0.1294,
+      "step": 20714
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.136363024893647e-06,
+      "loss": 0.1101,
+      "step": 20715
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1354437429962855e-06,
+      "loss": 0.083,
+      "step": 20716
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.134524499489709e-06,
+      "loss": 0.0922,
+      "step": 20717
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.133605294383045e-06,
+      "loss": 0.1119,
+      "step": 20718
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.132686127685426e-06,
+      "loss": 0.1337,
+      "step": 20719
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.131766999405982e-06,
+      "loss": 0.0941,
+      "step": 20720
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.13084790955385e-06,
+      "loss": 0.1211,
+      "step": 20721
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.129928858138153e-06,
+      "loss": 0.1093,
+      "step": 20722
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.129009845168019e-06,
+      "loss": 0.0864,
+      "step": 20723
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.128090870652586e-06,
+      "loss": 0.1298,
+      "step": 20724
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.12717193460098e-06,
+      "loss": 0.0765,
+      "step": 20725
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.126253037022329e-06,
+      "loss": 0.1034,
+      "step": 20726
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1253341779257615e-06,
+      "loss": 0.1619,
+      "step": 20727
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.124415357320404e-06,
+      "loss": 0.0827,
+      "step": 20728
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.123496575215385e-06,
+      "loss": 0.1302,
+      "step": 20729
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.122577831619832e-06,
+      "loss": 0.1276,
+      "step": 20730
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.121659126542876e-06,
+      "loss": 0.0976,
+      "step": 20731
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.120740459993635e-06,
+      "loss": 0.0879,
+      "step": 20732
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.119821831981241e-06,
+      "loss": 0.0731,
+      "step": 20733
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1189032425148195e-06,
+      "loss": 0.0888,
+      "step": 20734
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.117984691603493e-06,
+      "loss": 0.1057,
+      "step": 20735
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.117066179256392e-06,
+      "loss": 0.09,
+      "step": 20736
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1161477054826355e-06,
+      "loss": 0.1166,
+      "step": 20737
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.115229270291347e-06,
+      "loss": 0.1328,
+      "step": 20738
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.114310873691653e-06,
+      "loss": 0.0986,
+      "step": 20739
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.113392515692678e-06,
+      "loss": 0.1317,
+      "step": 20740
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1124741963035465e-06,
+      "loss": 0.0923,
+      "step": 20741
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.111555915533376e-06,
+      "loss": 0.0627,
+      "step": 20742
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.110637673391291e-06,
+      "loss": 0.1091,
+      "step": 20743
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.109719469886413e-06,
+      "loss": 0.0866,
+      "step": 20744
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.108801305027863e-06,
+      "loss": 0.1193,
+      "step": 20745
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.107883178824767e-06,
+      "loss": 0.1006,
+      "step": 20746
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.10696509128624e-06,
+      "loss": 0.1345,
+      "step": 20747
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.106047042421401e-06,
+      "loss": 0.11,
+      "step": 20748
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.105129032239376e-06,
+      "loss": 0.1209,
+      "step": 20749
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1042110607492835e-06,
+      "loss": 0.1203,
+      "step": 20750
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1032931279602396e-06,
+      "loss": 0.1311,
+      "step": 20751
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.102375233881364e-06,
+      "loss": 0.1054,
+      "step": 20752
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.101457378521776e-06,
+      "loss": 0.1026,
+      "step": 20753
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.100539561890591e-06,
+      "loss": 0.1145,
+      "step": 20754
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.099621783996934e-06,
+      "loss": 0.0903,
+      "step": 20755
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.098704044849913e-06,
+      "loss": 0.1311,
+      "step": 20756
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.097786344458648e-06,
+      "loss": 0.1079,
+      "step": 20757
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.096868682832259e-06,
+      "loss": 0.1188,
+      "step": 20758
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.095951059979859e-06,
+      "loss": 0.1119,
+      "step": 20759
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.095033475910566e-06,
+      "loss": 0.1131,
+      "step": 20760
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.094115930633493e-06,
+      "loss": 0.0945,
+      "step": 20761
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.093198424157754e-06,
+      "loss": 0.1542,
+      "step": 20762
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.092280956492464e-06,
+      "loss": 0.0907,
+      "step": 20763
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.091363527646741e-06,
+      "loss": 0.088,
+      "step": 20764
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0904461376297e-06,
+      "loss": 0.1011,
+      "step": 20765
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.089528786450445e-06,
+      "loss": 0.1256,
+      "step": 20766
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.088611474118097e-06,
+      "loss": 0.0793,
+      "step": 20767
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.087694200641765e-06,
+      "loss": 0.1019,
+      "step": 20768
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.086776966030564e-06,
+      "loss": 0.0932,
+      "step": 20769
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.085859770293606e-06,
+      "loss": 0.1052,
+      "step": 20770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.084942613439999e-06,
+      "loss": 0.1182,
+      "step": 20771
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0840254954788535e-06,
+      "loss": 0.1181,
+      "step": 20772
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.083108416419287e-06,
+      "loss": 0.1527,
+      "step": 20773
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.082191376270408e-06,
+      "loss": 0.0806,
+      "step": 20774
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.081274375041317e-06,
+      "loss": 0.0864,
+      "step": 20775
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.080357412741135e-06,
+      "loss": 0.1131,
+      "step": 20776
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.079440489378965e-06,
+      "loss": 0.0887,
+      "step": 20777
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.078523604963919e-06,
+      "loss": 0.1387,
+      "step": 20778
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.077606759505106e-06,
+      "loss": 0.1261,
+      "step": 20779
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07668995301163e-06,
+      "loss": 0.0991,
+      "step": 20780
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.075773185492597e-06,
+      "loss": 0.1047,
+      "step": 20781
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0748564569571215e-06,
+      "loss": 0.1387,
+      "step": 20782
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.073939767414305e-06,
+      "loss": 0.1046,
+      "step": 20783
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07302311687326e-06,
+      "loss": 0.0997,
+      "step": 20784
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.072106505343085e-06,
+      "loss": 0.0871,
+      "step": 20785
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.071189932832888e-06,
+      "loss": 0.1037,
+      "step": 20786
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.070273399351777e-06,
+      "loss": 0.1172,
+      "step": 20787
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.069356904908852e-06,
+      "loss": 0.1478,
+      "step": 20788
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.068440449513227e-06,
+      "loss": 0.1126,
+      "step": 20789
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.067524033173997e-06,
+      "loss": 0.1198,
+      "step": 20790
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.066607655900265e-06,
+      "loss": 0.1253,
+      "step": 20791
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.06569131770114e-06,
+      "loss": 0.1055,
+      "step": 20792
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.064775018585724e-06,
+      "loss": 0.1168,
+      "step": 20793
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0638587585631196e-06,
+      "loss": 0.0939,
+      "step": 20794
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.062942537642427e-06,
+      "loss": 0.1413,
+      "step": 20795
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.062026355832747e-06,
+      "loss": 0.1567,
+      "step": 20796
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.061110213143182e-06,
+      "loss": 0.1277,
+      "step": 20797
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0601941095828355e-06,
+      "loss": 0.1455,
+      "step": 20798
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0592780451608105e-06,
+      "loss": 0.0869,
+      "step": 20799
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.058362019886197e-06,
+      "loss": 0.111,
+      "step": 20800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.057446033768104e-06,
+      "loss": 0.1338,
+      "step": 20801
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.056530086815629e-06,
+      "loss": 0.1236,
+      "step": 20802
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.05561417903787e-06,
+      "loss": 0.1234,
+      "step": 20803
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.054698310443925e-06,
+      "loss": 0.1051,
+      "step": 20804
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.053782481042894e-06,
+      "loss": 0.0981,
+      "step": 20805
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.05286669084387e-06,
+      "loss": 0.091,
+      "step": 20806
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.051950939855958e-06,
+      "loss": 0.1617,
+      "step": 20807
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.051035228088255e-06,
+      "loss": 0.1246,
+      "step": 20808
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.05011955554985e-06,
+      "loss": 0.159,
+      "step": 20809
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.049203922249845e-06,
+      "loss": 0.0966,
+      "step": 20810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.048288328197336e-06,
+      "loss": 0.088,
+      "step": 20811
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.047372773401417e-06,
+      "loss": 0.1169,
+      "step": 20812
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.046457257871184e-06,
+      "loss": 0.0965,
+      "step": 20813
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.045541781615732e-06,
+      "loss": 0.1133,
+      "step": 20814
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.044626344644151e-06,
+      "loss": 0.1135,
+      "step": 20815
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.04371094696554e-06,
+      "loss": 0.1091,
+      "step": 20816
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.042795588588993e-06,
+      "loss": 0.1078,
+      "step": 20817
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.041880269523604e-06,
+      "loss": 0.1217,
+      "step": 20818
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.040964989778462e-06,
+      "loss": 0.1105,
+      "step": 20819
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.040049749362659e-06,
+      "loss": 0.1084,
+      "step": 20820
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.039134548285291e-06,
+      "loss": 0.1118,
+      "step": 20821
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.038219386555444e-06,
+      "loss": 0.145,
+      "step": 20822
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.03730426418222e-06,
+      "loss": 0.1043,
+      "step": 20823
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.036389181174696e-06,
+      "loss": 0.1114,
+      "step": 20824
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.035474137541974e-06,
+      "loss": 0.0995,
+      "step": 20825
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.034559133293137e-06,
+      "loss": 0.117,
+      "step": 20826
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0336441684372785e-06,
+      "loss": 0.0925,
+      "step": 20827
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.032729242983488e-06,
+      "loss": 0.1244,
+      "step": 20828
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.031814356940851e-06,
+      "loss": 0.1693,
+      "step": 20829
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0308995103184576e-06,
+      "loss": 0.1301,
+      "step": 20830
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0299847031253956e-06,
+      "loss": 0.1128,
+      "step": 20831
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.029069935370759e-06,
+      "loss": 0.0692,
+      "step": 20832
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.028155207063624e-06,
+      "loss": 0.1262,
+      "step": 20833
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.027240518213086e-06,
+      "loss": 0.1697,
+      "step": 20834
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.026325868828228e-06,
+      "loss": 0.1423,
+      "step": 20835
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.025411258918137e-06,
+      "loss": 0.1027,
+      "step": 20836
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.024496688491901e-06,
+      "loss": 0.0786,
+      "step": 20837
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.023582157558601e-06,
+      "loss": 0.1537,
+      "step": 20838
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.022667666127325e-06,
+      "loss": 0.1237,
+      "step": 20839
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.021753214207156e-06,
+      "loss": 0.0798,
+      "step": 20840
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.020838801807178e-06,
+      "loss": 0.1099,
+      "step": 20841
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.019924428936481e-06,
+      "loss": 0.0967,
+      "step": 20842
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0190100956041405e-06,
+      "loss": 0.1002,
+      "step": 20843
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.018095801819241e-06,
+      "loss": 0.1293,
+      "step": 20844
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.017181547590869e-06,
+      "loss": 0.1093,
+      "step": 20845
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.016267332928104e-06,
+      "loss": 0.106,
+      "step": 20846
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.015353157840029e-06,
+      "loss": 0.113,
+      "step": 20847
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.014439022335726e-06,
+      "loss": 0.0829,
+      "step": 20848
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0135249264242704e-06,
+      "loss": 0.1241,
+      "step": 20849
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.012610870114749e-06,
+      "loss": 0.0911,
+      "step": 20850
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.011696853416241e-06,
+      "loss": 0.1206,
+      "step": 20851
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.010782876337828e-06,
+      "loss": 0.0931,
+      "step": 20852
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0098689388885854e-06,
+      "loss": 0.0953,
+      "step": 20853
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.008955041077594e-06,
+      "loss": 0.1228,
+      "step": 20854
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.008041182913933e-06,
+      "loss": 0.1391,
+      "step": 20855
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0071273644066785e-06,
+      "loss": 0.1361,
+      "step": 20856
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.006213585564917e-06,
+      "loss": 0.1415,
+      "step": 20857
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.005299846397712e-06,
+      "loss": 0.1055,
+      "step": 20858
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.004386146914152e-06,
+      "loss": 0.0845,
+      "step": 20859
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.003472487123308e-06,
+      "loss": 0.0793,
+      "step": 20860
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0025588670342615e-06,
+      "loss": 0.125,
+      "step": 20861
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.001645286656083e-06,
+      "loss": 0.1181,
+      "step": 20862
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.00073174599785e-06,
+      "loss": 0.1438,
+      "step": 20863
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.999818245068638e-06,
+      "loss": 0.0991,
+      "step": 20864
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.998904783877521e-06,
+      "loss": 0.1011,
+      "step": 20865
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.997991362433578e-06,
+      "loss": 0.0953,
+      "step": 20866
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.997077980745874e-06,
+      "loss": 0.1114,
+      "step": 20867
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.996164638823491e-06,
+      "loss": 0.1247,
+      "step": 20868
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.995251336675498e-06,
+      "loss": 0.1119,
+      "step": 20869
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.994338074310968e-06,
+      "loss": 0.1368,
+      "step": 20870
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.993424851738978e-06,
+      "loss": 0.1106,
+      "step": 20871
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9925116689685925e-06,
+      "loss": 0.1374,
+      "step": 20872
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.991598526008888e-06,
+      "loss": 0.0958,
+      "step": 20873
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990685422868931e-06,
+      "loss": 0.0807,
+      "step": 20874
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.989772359557799e-06,
+      "loss": 0.1272,
+      "step": 20875
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.988859336084561e-06,
+      "loss": 0.1304,
+      "step": 20876
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.987946352458285e-06,
+      "loss": 0.0828,
+      "step": 20877
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.987033408688039e-06,
+      "loss": 0.1159,
+      "step": 20878
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.986120504782895e-06,
+      "loss": 0.0999,
+      "step": 20879
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.985207640751919e-06,
+      "loss": 0.1278,
+      "step": 20880
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.984294816604188e-06,
+      "loss": 0.0879,
+      "step": 20881
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.983382032348759e-06,
+      "loss": 0.1364,
+      "step": 20882
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.982469287994701e-06,
+      "loss": 0.0998,
+      "step": 20883
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.981556583551089e-06,
+      "loss": 0.0942,
+      "step": 20884
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.980643919026985e-06,
+      "loss": 0.0968,
+      "step": 20885
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9797312944314565e-06,
+      "loss": 0.0858,
+      "step": 20886
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.978818709773568e-06,
+      "loss": 0.1157,
+      "step": 20887
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.977906165062386e-06,
+      "loss": 0.1276,
+      "step": 20888
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.976993660306976e-06,
+      "loss": 0.1251,
+      "step": 20889
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.976081195516405e-06,
+      "loss": 0.0946,
+      "step": 20890
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.975168770699733e-06,
+      "loss": 0.0989,
+      "step": 20891
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.974256385866025e-06,
+      "loss": 0.1209,
+      "step": 20892
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.973344041024347e-06,
+      "loss": 0.1274,
+      "step": 20893
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9724317361837635e-06,
+      "loss": 0.0745,
+      "step": 20894
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.971519471353335e-06,
+      "loss": 0.0872,
+      "step": 20895
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.970607246542124e-06,
+      "loss": 0.1305,
+      "step": 20896
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.969695061759192e-06,
+      "loss": 0.1138,
+      "step": 20897
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.968782917013603e-06,
+      "loss": 0.1081,
+      "step": 20898
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.967870812314413e-06,
+      "loss": 0.2035,
+      "step": 20899
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.966958747670694e-06,
+      "loss": 0.11,
+      "step": 20900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9660467230914934e-06,
+      "loss": 0.1115,
+      "step": 20901
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.965134738585879e-06,
+      "loss": 0.1215,
+      "step": 20902
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.964222794162908e-06,
+      "loss": 0.1005,
+      "step": 20903
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.963310889831642e-06,
+      "loss": 0.0908,
+      "step": 20904
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.96239902560114e-06,
+      "loss": 0.0997,
+      "step": 20905
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.961487201480458e-06,
+      "loss": 0.1209,
+      "step": 20906
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9605754174786535e-06,
+      "loss": 0.0965,
+      "step": 20907
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9596636736047855e-06,
+      "loss": 0.1065,
+      "step": 20908
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.958751969867912e-06,
+      "loss": 0.1112,
+      "step": 20909
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.957840306277092e-06,
+      "loss": 0.1152,
+      "step": 20910
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.956928682841378e-06,
+      "loss": 0.1036,
+      "step": 20911
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.956017099569828e-06,
+      "loss": 0.1075,
+      "step": 20912
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.955105556471497e-06,
+      "loss": 0.1831,
+      "step": 20913
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.95419405355544e-06,
+      "loss": 0.0993,
+      "step": 20914
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.953282590830718e-06,
+      "loss": 0.1317,
+      "step": 20915
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.952371168306376e-06,
+      "loss": 0.0887,
+      "step": 20916
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9514597859914705e-06,
+      "loss": 0.1437,
+      "step": 20917
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9505484438950586e-06,
+      "loss": 0.0955,
+      "step": 20918
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.949637142026197e-06,
+      "loss": 0.117,
+      "step": 20919
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.948725880393928e-06,
+      "loss": 0.0881,
+      "step": 20920
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.947814659007312e-06,
+      "loss": 0.0999,
+      "step": 20921
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9469034778754005e-06,
+      "loss": 0.1467,
+      "step": 20922
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.94599233700724e-06,
+      "loss": 0.1285,
+      "step": 20923
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.945081236411893e-06,
+      "loss": 0.0726,
+      "step": 20924
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.944170176098398e-06,
+      "loss": 0.1237,
+      "step": 20925
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9432591560758094e-06,
+      "loss": 0.104,
+      "step": 20926
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.942348176353181e-06,
+      "loss": 0.1159,
+      "step": 20927
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.941437236939559e-06,
+      "loss": 0.105,
+      "step": 20928
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.940526337843997e-06,
+      "loss": 0.0914,
+      "step": 20929
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.939615479075538e-06,
+      "loss": 0.0855,
+      "step": 20930
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.938704660643234e-06,
+      "loss": 0.1058,
+      "step": 20931
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.937793882556131e-06,
+      "loss": 0.1465,
+      "step": 20932
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.93688314482328e-06,
+      "loss": 0.0756,
+      "step": 20933
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.93597244745373e-06,
+      "loss": 0.0865,
+      "step": 20934
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.935061790456519e-06,
+      "loss": 0.0874,
+      "step": 20935
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.934151173840701e-06,
+      "loss": 0.166,
+      "step": 20936
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9332405976153195e-06,
+      "loss": 0.1198,
+      "step": 20937
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.932330061789422e-06,
+      "loss": 0.0945,
+      "step": 20938
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.931419566372054e-06,
+      "loss": 0.157,
+      "step": 20939
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.930509111372258e-06,
+      "loss": 0.0968,
+      "step": 20940
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.929598696799077e-06,
+      "loss": 0.095,
+      "step": 20941
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.928688322661561e-06,
+      "loss": 0.1408,
+      "step": 20942
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9277779889687546e-06,
+      "loss": 0.0957,
+      "step": 20943
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.92686769572969e-06,
+      "loss": 0.1119,
+      "step": 20944
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9259574429534195e-06,
+      "loss": 0.1358,
+      "step": 20945
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9250472306489836e-06,
+      "loss": 0.1293,
+      "step": 20946
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9241370588254236e-06,
+      "loss": 0.1467,
+      "step": 20947
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.923226927491784e-06,
+      "loss": 0.1353,
+      "step": 20948
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9223168366571024e-06,
+      "loss": 0.1075,
+      "step": 20949
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.92140678633042e-06,
+      "loss": 0.1589,
+      "step": 20950
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.920496776520777e-06,
+      "loss": 0.1329,
+      "step": 20951
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.919586807237218e-06,
+      "loss": 0.1319,
+      "step": 20952
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.91867687848878e-06,
+      "loss": 0.0964,
+      "step": 20953
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.917766990284502e-06,
+      "loss": 0.1305,
+      "step": 20954
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.916857142633421e-06,
+      "loss": 0.1261,
+      "step": 20955
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9159473355445785e-06,
+      "loss": 0.0957,
+      "step": 20956
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.915037569027009e-06,
+      "loss": 0.0965,
+      "step": 20957
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9141278430897585e-06,
+      "loss": 0.1176,
+      "step": 20958
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.913218157741856e-06,
+      "loss": 0.1161,
+      "step": 20959
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.912308512992336e-06,
+      "loss": 0.1333,
+      "step": 20960
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.911398908850243e-06,
+      "loss": 0.0656,
+      "step": 20961
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.91048934532461e-06,
+      "loss": 0.1542,
+      "step": 20962
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.909579822424476e-06,
+      "loss": 0.0873,
+      "step": 20963
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.90867034015887e-06,
+      "loss": 0.1197,
+      "step": 20964
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.907760898536828e-06,
+      "loss": 0.1329,
+      "step": 20965
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.906851497567385e-06,
+      "loss": 0.1085,
+      "step": 20966
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9059421372595795e-06,
+      "loss": 0.1191,
+      "step": 20967
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.905032817622445e-06,
+      "loss": 0.0805,
+      "step": 20968
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.904123538665005e-06,
+      "loss": 0.1021,
+      "step": 20969
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.903214300396302e-06,
+      "loss": 0.1409,
+      "step": 20970
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9023051028253645e-06,
+      "loss": 0.1173,
+      "step": 20971
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.901395945961228e-06,
+      "loss": 0.1529,
+      "step": 20972
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.90048682981292e-06,
+      "loss": 0.1138,
+      "step": 20973
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8995777543894715e-06,
+      "loss": 0.1168,
+      "step": 20974
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.898668719699914e-06,
+      "loss": 0.0911,
+      "step": 20975
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.89775972575328e-06,
+      "loss": 0.0971,
+      "step": 20976
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.896850772558605e-06,
+      "loss": 0.1263,
+      "step": 20977
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.895941860124903e-06,
+      "loss": 0.1288,
+      "step": 20978
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8950329884612155e-06,
+      "loss": 0.1031,
+      "step": 20979
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.89412415757657e-06,
+      "loss": 0.1091,
+      "step": 20980
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.893215367479991e-06,
+      "loss": 0.1305,
+      "step": 20981
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.892306618180511e-06,
+      "loss": 0.1307,
+      "step": 20982
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.891397909687153e-06,
+      "loss": 0.1164,
+      "step": 20983
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.890489242008944e-06,
+      "loss": 0.0824,
+      "step": 20984
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.889580615154915e-06,
+      "loss": 0.1129,
+      "step": 20985
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8886720291340905e-06,
+      "loss": 0.146,
+      "step": 20986
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8877634839555e-06,
+      "loss": 0.1376,
+      "step": 20987
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.88685497962816e-06,
+      "loss": 0.1275,
+      "step": 20988
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.885946516161104e-06,
+      "loss": 0.0893,
+      "step": 20989
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8850380935633536e-06,
+      "loss": 0.1025,
+      "step": 20990
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.884129711843929e-06,
+      "loss": 0.0613,
+      "step": 20991
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8832213710118665e-06,
+      "loss": 0.1295,
+      "step": 20992
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8823130710761754e-06,
+      "loss": 0.1132,
+      "step": 20993
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.881404812045886e-06,
+      "loss": 0.1102,
+      "step": 20994
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.880496593930021e-06,
+      "loss": 0.1149,
+      "step": 20995
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8795884167376004e-06,
+      "loss": 0.1096,
+      "step": 20996
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.878680280477651e-06,
+      "loss": 0.0814,
+      "step": 20997
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.877772185159187e-06,
+      "loss": 0.1248,
+      "step": 20998
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.876864130791235e-06,
+      "loss": 0.1087,
+      "step": 20999
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.875956117382811e-06,
+      "loss": 0.1101,
+      "step": 21000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.875048144942946e-06,
+      "loss": 0.1097,
+      "step": 21001
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.874140213480642e-06,
+      "loss": 0.1486,
+      "step": 21002
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8732323230049335e-06,
+      "loss": 0.1038,
+      "step": 21003
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.872324473524833e-06,
+      "loss": 0.1279,
+      "step": 21004
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.871416665049361e-06,
+      "loss": 0.13,
+      "step": 21005
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.870508897587538e-06,
+      "loss": 0.1118,
+      "step": 21006
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.869601171148377e-06,
+      "loss": 0.1286,
+      "step": 21007
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.868693485740897e-06,
+      "loss": 0.1148,
+      "step": 21008
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.867785841374114e-06,
+      "loss": 0.0907,
+      "step": 21009
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.866878238057049e-06,
+      "loss": 0.0794,
+      "step": 21010
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.865970675798716e-06,
+      "loss": 0.0784,
+      "step": 21011
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.865063154608131e-06,
+      "loss": 0.1323,
+      "step": 21012
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.864155674494305e-06,
+      "loss": 0.1092,
+      "step": 21013
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.863248235466259e-06,
+      "loss": 0.1351,
+      "step": 21014
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8623408375330035e-06,
+      "loss": 0.1135,
+      "step": 21015
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.861433480703558e-06,
+      "loss": 0.1631,
+      "step": 21016
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.860526164986929e-06,
+      "loss": 0.1168,
+      "step": 21017
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8596188903921325e-06,
+      "loss": 0.0925,
+      "step": 21018
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.858711656928184e-06,
+      "loss": 0.1343,
+      "step": 21019
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.857804464604095e-06,
+      "loss": 0.1091,
+      "step": 21020
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.856897313428878e-06,
+      "loss": 0.1121,
+      "step": 21021
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.855990203411542e-06,
+      "loss": 0.1166,
+      "step": 21022
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.855083134561099e-06,
+      "loss": 0.1085,
+      "step": 21023
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8541761068865624e-06,
+      "loss": 0.08,
+      "step": 21024
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.853269120396938e-06,
+      "loss": 0.1248,
+      "step": 21025
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8523621751012465e-06,
+      "loss": 0.1061,
+      "step": 21026
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.851455271008481e-06,
+      "loss": 0.0833,
+      "step": 21027
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.850548408127665e-06,
+      "loss": 0.1772,
+      "step": 21028
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.849641586467802e-06,
+      "loss": 0.1094,
+      "step": 21029
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.848734806037901e-06,
+      "loss": 0.1085,
+      "step": 21030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.847828066846968e-06,
+      "loss": 0.1349,
+      "step": 21031
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.846921368904013e-06,
+      "loss": 0.1012,
+      "step": 21032
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.846014712218042e-06,
+      "loss": 0.0921,
+      "step": 21033
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.84510809679806e-06,
+      "loss": 0.1036,
+      "step": 21034
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.844201522653081e-06,
+      "loss": 0.1095,
+      "step": 21035
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.843294989792101e-06,
+      "loss": 0.1395,
+      "step": 21036
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.842388498224132e-06,
+      "loss": 0.0956,
+      "step": 21037
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.841482047958176e-06,
+      "loss": 0.0968,
+      "step": 21038
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.84057563900324e-06,
+      "loss": 0.1049,
+      "step": 21039
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.839669271368329e-06,
+      "loss": 0.1391,
+      "step": 21040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.838762945062444e-06,
+      "loss": 0.0705,
+      "step": 21041
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8378566600945885e-06,
+      "loss": 0.0916,
+      "step": 21042
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.836950416473766e-06,
+      "loss": 0.0997,
+      "step": 21043
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8360442142089836e-06,
+      "loss": 0.12,
+      "step": 21044
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.835138053309239e-06,
+      "loss": 0.1251,
+      "step": 21045
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.834231933783534e-06,
+      "loss": 0.1128,
+      "step": 21046
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.833325855640869e-06,
+      "loss": 0.0929,
+      "step": 21047
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.832419818890251e-06,
+      "loss": 0.0995,
+      "step": 21048
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8315138235406745e-06,
+      "loss": 0.1369,
+      "step": 21049
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8306078696011446e-06,
+      "loss": 0.1628,
+      "step": 21050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.82970195708066e-06,
+      "loss": 0.0994,
+      "step": 21051
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.828796085988213e-06,
+      "loss": 0.1081,
+      "step": 21052
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.827890256332814e-06,
+      "loss": 0.1085,
+      "step": 21053
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.826984468123451e-06,
+      "loss": 0.084,
+      "step": 21054
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.82607872136913e-06,
+      "loss": 0.1217,
+      "step": 21055
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8251730160788445e-06,
+      "loss": 0.1124,
+      "step": 21056
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8242673522615956e-06,
+      "loss": 0.1491,
+      "step": 21057
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.823361729926373e-06,
+      "loss": 0.1097,
+      "step": 21058
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.822456149082183e-06,
+      "loss": 0.116,
+      "step": 21059
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.821550609738017e-06,
+      "loss": 0.1019,
+      "step": 21060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.820645111902865e-06,
+      "loss": 0.1078,
+      "step": 21061
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.819739655585733e-06,
+      "loss": 0.095,
+      "step": 21062
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.818834240795607e-06,
+      "loss": 0.0772,
+      "step": 21063
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.817928867541488e-06,
+      "loss": 0.1083,
+      "step": 21064
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8170235358323665e-06,
+      "loss": 0.1446,
+      "step": 21065
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8161182456772334e-06,
+      "loss": 0.0863,
+      "step": 21066
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.815212997085085e-06,
+      "loss": 0.1601,
+      "step": 21067
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.814307790064919e-06,
+      "loss": 0.0906,
+      "step": 21068
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.813402624625722e-06,
+      "loss": 0.1229,
+      "step": 21069
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.812497500776484e-06,
+      "loss": 0.103,
+      "step": 21070
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.811592418526203e-06,
+      "loss": 0.1281,
+      "step": 21071
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.810687377883863e-06,
+      "loss": 0.0809,
+      "step": 21072
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.809782378858459e-06,
+      "loss": 0.1314,
+      "step": 21073
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.808877421458987e-06,
+      "loss": 0.1154,
+      "step": 21074
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.807972505694426e-06,
+      "loss": 0.1305,
+      "step": 21075
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.807067631573767e-06,
+      "loss": 0.1419,
+      "step": 21076
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.806162799106008e-06,
+      "loss": 0.1671,
+      "step": 21077
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.805258008300128e-06,
+      "loss": 0.0951,
+      "step": 21078
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.804353259165123e-06,
+      "loss": 0.1343,
+      "step": 21079
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.803448551709977e-06,
+      "loss": 0.1081,
+      "step": 21080
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802543885943673e-06,
+      "loss": 0.1056,
+      "step": 21081
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.801639261875204e-06,
+      "loss": 0.1144,
+      "step": 21082
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.800734679513558e-06,
+      "loss": 0.0865,
+      "step": 21083
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.799830138867719e-06,
+      "loss": 0.1058,
+      "step": 21084
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.798925639946669e-06,
+      "loss": 0.1204,
+      "step": 21085
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.798021182759399e-06,
+      "loss": 0.1314,
+      "step": 21086
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.797116767314887e-06,
+      "loss": 0.1308,
+      "step": 21087
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.796212393622127e-06,
+      "loss": 0.0931,
+      "step": 21088
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.795308061690097e-06,
+      "loss": 0.1098,
+      "step": 21089
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.794403771527779e-06,
+      "loss": 0.0596,
+      "step": 21090
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.79349952314416e-06,
+      "loss": 0.1396,
+      "step": 21091
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.792595316548224e-06,
+      "loss": 0.1025,
+      "step": 21092
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7916911517489505e-06,
+      "loss": 0.1486,
+      "step": 21093
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.79078702875532e-06,
+      "loss": 0.1177,
+      "step": 21094
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.789882947576321e-06,
+      "loss": 0.0945,
+      "step": 21095
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.788978908220926e-06,
+      "loss": 0.0983,
+      "step": 21096
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.788074910698119e-06,
+      "loss": 0.0873,
+      "step": 21097
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.78717095501689e-06,
+      "loss": 0.1318,
+      "step": 21098
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.786267041186202e-06,
+      "loss": 0.1316,
+      "step": 21099
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.785363169215043e-06,
+      "loss": 0.1053,
+      "step": 21100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.784459339112397e-06,
+      "loss": 0.1243,
+      "step": 21101
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.783555550887233e-06,
+      "loss": 0.1148,
+      "step": 21102
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.782651804548538e-06,
+      "loss": 0.1073,
+      "step": 21103
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.781748100105287e-06,
+      "loss": 0.1197,
+      "step": 21104
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.780844437566451e-06,
+      "loss": 0.114,
+      "step": 21105
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.779940816941013e-06,
+      "loss": 0.0698,
+      "step": 21106
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.779037238237953e-06,
+      "loss": 0.1196,
+      "step": 21107
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.778133701466244e-06,
+      "loss": 0.0948,
+      "step": 21108
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.777230206634856e-06,
+      "loss": 0.1237,
+      "step": 21109
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.776326753752774e-06,
+      "loss": 0.1025,
+      "step": 21110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.775423342828965e-06,
+      "loss": 0.0988,
+      "step": 21111
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.77451997387241e-06,
+      "loss": 0.1351,
+      "step": 21112
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.773616646892082e-06,
+      "loss": 0.1049,
+      "step": 21113
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.772713361896946e-06,
+      "loss": 0.1182,
+      "step": 21114
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.771810118895984e-06,
+      "loss": 0.1173,
+      "step": 21115
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.770906917898171e-06,
+      "loss": 0.1274,
+      "step": 21116
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.770003758912475e-06,
+      "loss": 0.1373,
+      "step": 21117
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.769100641947864e-06,
+      "loss": 0.1038,
+      "step": 21118
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.76819756701332e-06,
+      "loss": 0.0758,
+      "step": 21119
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7672945341178025e-06,
+      "loss": 0.1011,
+      "step": 21120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.766391543270293e-06,
+      "loss": 0.1059,
+      "step": 21121
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.765488594479758e-06,
+      "loss": 0.1235,
+      "step": 21122
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.764585687755161e-06,
+      "loss": 0.0831,
+      "step": 21123
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.763682823105477e-06,
+      "loss": 0.0906,
+      "step": 21124
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.762780000539681e-06,
+      "loss": 0.1196,
+      "step": 21125
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.761877220066729e-06,
+      "loss": 0.1418,
+      "step": 21126
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7609744816956026e-06,
+      "loss": 0.1102,
+      "step": 21127
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.760071785435263e-06,
+      "loss": 0.0916,
+      "step": 21128
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.759169131294673e-06,
+      "loss": 0.0883,
+      "step": 21129
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.758266519282809e-06,
+      "loss": 0.117,
+      "step": 21130
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.757363949408629e-06,
+      "loss": 0.1523,
+      "step": 21131
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.756461421681108e-06,
+      "loss": 0.1461,
+      "step": 21132
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7555589361092026e-06,
+      "loss": 0.1626,
+      "step": 21133
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.754656492701886e-06,
+      "loss": 0.0852,
+      "step": 21134
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.753754091468116e-06,
+      "loss": 0.108,
+      "step": 21135
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.752851732416864e-06,
+      "loss": 0.0813,
+      "step": 21136
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751949415557093e-06,
+      "loss": 0.1063,
+      "step": 21137
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751047140897761e-06,
+      "loss": 0.1135,
+      "step": 21138
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7501449084478366e-06,
+      "loss": 0.1278,
+      "step": 21139
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.749242718216279e-06,
+      "loss": 0.1457,
+      "step": 21140
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7483405702120565e-06,
+      "loss": 0.1199,
+      "step": 21141
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.747438464444121e-06,
+      "loss": 0.1327,
+      "step": 21142
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.746536400921447e-06,
+      "loss": 0.1004,
+      "step": 21143
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7456343796529845e-06,
+      "loss": 0.1331,
+      "step": 21144
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7447324006477016e-06,
+      "loss": 0.0976,
+      "step": 21145
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.743830463914557e-06,
+      "loss": 0.0943,
+      "step": 21146
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.742928569462505e-06,
+      "loss": 0.1096,
+      "step": 21147
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7420267173005125e-06,
+      "loss": 0.0992,
+      "step": 21148
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.741124907437532e-06,
+      "loss": 0.1059,
+      "step": 21149
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.740223139882526e-06,
+      "loss": 0.1036,
+      "step": 21150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.739321414644457e-06,
+      "loss": 0.1411,
+      "step": 21151
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.738419731732277e-06,
+      "loss": 0.1151,
+      "step": 21152
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.737518091154941e-06,
+      "loss": 0.0979,
+      "step": 21153
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.736616492921413e-06,
+      "loss": 0.0891,
+      "step": 21154
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.735714937040642e-06,
+      "loss": 0.1043,
+      "step": 21155
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.734813423521593e-06,
+      "loss": 0.1183,
+      "step": 21156
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.733911952373217e-06,
+      "loss": 0.1468,
+      "step": 21157
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.733010523604465e-06,
+      "loss": 0.0819,
+      "step": 21158
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.732109137224295e-06,
+      "loss": 0.0859,
+      "step": 21159
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.731207793241668e-06,
+      "loss": 0.1036,
+      "step": 21160
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7303064916655315e-06,
+      "loss": 0.1321,
+      "step": 21161
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.729405232504837e-06,
+      "loss": 0.0824,
+      "step": 21162
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.728504015768544e-06,
+      "loss": 0.0645,
+      "step": 21163
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.727602841465598e-06,
+      "loss": 0.0858,
+      "step": 21164
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.726701709604956e-06,
+      "loss": 0.1107,
+      "step": 21165
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.725800620195576e-06,
+      "loss": 0.0945,
+      "step": 21166
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.724899573246395e-06,
+      "loss": 0.1308,
+      "step": 21167
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723998568766372e-06,
+      "loss": 0.0816,
+      "step": 21168
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723097606764463e-06,
+      "loss": 0.1244,
+      "step": 21169
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.72219668724961e-06,
+      "loss": 0.1123,
+      "step": 21170
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.721295810230763e-06,
+      "loss": 0.1055,
+      "step": 21171
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.720394975716878e-06,
+      "loss": 0.1206,
+      "step": 21172
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.719494183716896e-06,
+      "loss": 0.1193,
+      "step": 21173
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.718593434239767e-06,
+      "loss": 0.0953,
+      "step": 21174
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.717692727294453e-06,
+      "loss": 0.092,
+      "step": 21175
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7167920628898795e-06,
+      "loss": 0.0696,
+      "step": 21176
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.715891441035005e-06,
+      "loss": 0.1466,
+      "step": 21177
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7149908617387806e-06,
+      "loss": 0.0966,
+      "step": 21178
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.714090325010143e-06,
+      "loss": 0.1255,
+      "step": 21179
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.713189830858048e-06,
+      "loss": 0.1064,
+      "step": 21180
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7122893792914344e-06,
+      "loss": 0.1213,
+      "step": 21181
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.711388970319247e-06,
+      "loss": 0.1545,
+      "step": 21182
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.710488603950431e-06,
+      "loss": 0.1483,
+      "step": 21183
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7095882801939386e-06,
+      "loss": 0.1207,
+      "step": 21184
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7086879990587064e-06,
+      "loss": 0.0793,
+      "step": 21185
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7077877605536735e-06,
+      "loss": 0.0725,
+      "step": 21186
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7068875646877956e-06,
+      "loss": 0.1075,
+      "step": 21187
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.705987411470001e-06,
+      "loss": 0.124,
+      "step": 21188
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.70508730090924e-06,
+      "loss": 0.1147,
+      "step": 21189
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.704187233014462e-06,
+      "loss": 0.0899,
+      "step": 21190
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.703287207794591e-06,
+      "loss": 0.138,
+      "step": 21191
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.702387225258577e-06,
+      "loss": 0.1465,
+      "step": 21192
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.701487285415364e-06,
+      "loss": 0.1055,
+      "step": 21193
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7005873882738835e-06,
+      "loss": 0.1455,
+      "step": 21194
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.699687533843086e-06,
+      "loss": 0.1013,
+      "step": 21195
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.698787722131902e-06,
+      "loss": 0.1025,
+      "step": 21196
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.69788795314927e-06,
+      "loss": 0.1071,
+      "step": 21197
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.696988226904131e-06,
+      "loss": 0.072,
+      "step": 21198
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6960885434054315e-06,
+      "loss": 0.1154,
+      "step": 21199
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6951889026620925e-06,
+      "loss": 0.0781,
+      "step": 21200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.694289304683058e-06,
+      "loss": 0.0891,
+      "step": 21201
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.693389749477271e-06,
+      "loss": 0.105,
+      "step": 21202
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.692490237053658e-06,
+      "loss": 0.0829,
+      "step": 21203
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.691590767421164e-06,
+      "loss": 0.0993,
+      "step": 21204
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.690691340588721e-06,
+      "loss": 0.1162,
+      "step": 21205
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.689791956565258e-06,
+      "loss": 0.0972,
+      "step": 21206
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6888926153597136e-06,
+      "loss": 0.0966,
+      "step": 21207
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.687993316981029e-06,
+      "loss": 0.1091,
+      "step": 21208
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.68709406143813e-06,
+      "loss": 0.0904,
+      "step": 21209
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.686194848739947e-06,
+      "loss": 0.1203,
+      "step": 21210
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.685295678895422e-06,
+      "loss": 0.1072,
+      "step": 21211
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.684396551913478e-06,
+      "loss": 0.1274,
+      "step": 21212
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.683497467803056e-06,
+      "loss": 0.1061,
+      "step": 21213
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.682598426573085e-06,
+      "loss": 0.099,
+      "step": 21214
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.681699428232488e-06,
+      "loss": 0.1254,
+      "step": 21215
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6808004727902046e-06,
+      "loss": 0.1344,
+      "step": 21216
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.679901560255165e-06,
+      "loss": 0.1437,
+      "step": 21217
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.679002690636292e-06,
+      "loss": 0.1061,
+      "step": 21218
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.678103863942526e-06,
+      "loss": 0.1363,
+      "step": 21219
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.677205080182788e-06,
+      "loss": 0.1057,
+      "step": 21220
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.676306339366004e-06,
+      "loss": 0.1075,
+      "step": 21221
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.675407641501113e-06,
+      "loss": 0.1094,
+      "step": 21222
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.67450898659703e-06,
+      "loss": 0.1337,
+      "step": 21223
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.673610374662695e-06,
+      "loss": 0.1328,
+      "step": 21224
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6727118057070255e-06,
+      "loss": 0.0771,
+      "step": 21225
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.671813279738948e-06,
+      "loss": 0.0894,
+      "step": 21226
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.670914796767392e-06,
+      "loss": 0.1043,
+      "step": 21227
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.670016356801286e-06,
+      "loss": 0.1042,
+      "step": 21228
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6691179598495504e-06,
+      "loss": 0.1546,
+      "step": 21229
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6682196059211094e-06,
+      "loss": 0.1172,
+      "step": 21230
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.667321295024892e-06,
+      "loss": 0.0869,
+      "step": 21231
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.666423027169815e-06,
+      "loss": 0.1054,
+      "step": 21232
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6655248023648105e-06,
+      "loss": 0.1077,
+      "step": 21233
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.664626620618797e-06,
+      "loss": 0.1213,
+      "step": 21234
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6637284819406936e-06,
+      "loss": 0.1084,
+      "step": 21235
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.662830386339425e-06,
+      "loss": 0.1262,
+      "step": 21236
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6619323338239185e-06,
+      "loss": 0.0992,
+      "step": 21237
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.661034324403092e-06,
+      "loss": 0.1334,
+      "step": 21238
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.660136358085859e-06,
+      "loss": 0.0871,
+      "step": 21239
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6592384348811525e-06,
+      "loss": 0.1002,
+      "step": 21240
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6583405547978805e-06,
+      "loss": 0.1438,
+      "step": 21241
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.65744271784497e-06,
+      "loss": 0.1537,
+      "step": 21242
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.656544924031346e-06,
+      "loss": 0.1021,
+      "step": 21243
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.655647173365913e-06,
+      "loss": 0.0765,
+      "step": 21244
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.654749465857594e-06,
+      "loss": 0.0905,
+      "step": 21245
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.653851801515313e-06,
+      "loss": 0.1204,
+      "step": 21246
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.652954180347981e-06,
+      "loss": 0.0989,
+      "step": 21247
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.652056602364522e-06,
+      "loss": 0.0908,
+      "step": 21248
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.651159067573846e-06,
+      "loss": 0.1176,
+      "step": 21249
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.650261575984869e-06,
+      "loss": 0.1326,
+      "step": 21250
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.649364127606509e-06,
+      "loss": 0.1326,
+      "step": 21251
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.648466722447689e-06,
+      "loss": 0.111,
+      "step": 21252
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.647569360517307e-06,
+      "loss": 0.12,
+      "step": 21253
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.646672041824288e-06,
+      "loss": 0.1271,
+      "step": 21254
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.64577476637755e-06,
+      "loss": 0.1241,
+      "step": 21255
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.644877534185995e-06,
+      "loss": 0.1147,
+      "step": 21256
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.643980345258548e-06,
+      "loss": 0.1228,
+      "step": 21257
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.643083199604116e-06,
+      "loss": 0.1019,
+      "step": 21258
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.642186097231607e-06,
+      "loss": 0.0808,
+      "step": 21259
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.64128903814994e-06,
+      "loss": 0.1458,
+      "step": 21260
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.640392022368025e-06,
+      "loss": 0.1535,
+      "step": 21261
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6394950498947735e-06,
+      "loss": 0.1105,
+      "step": 21262
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.638598120739089e-06,
+      "loss": 0.1066,
+      "step": 21263
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.637701234909892e-06,
+      "loss": 0.086,
+      "step": 21264
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.636804392416084e-06,
+      "loss": 0.0917,
+      "step": 21265
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.635907593266578e-06,
+      "loss": 0.1059,
+      "step": 21266
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.635010837470289e-06,
+      "loss": 0.1259,
+      "step": 21267
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.634114125036112e-06,
+      "loss": 0.1213,
+      "step": 21268
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.633217455972961e-06,
+      "loss": 0.1432,
+      "step": 21269
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.63232083028975e-06,
+      "loss": 0.0882,
+      "step": 21270
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.631424247995376e-06,
+      "loss": 0.1279,
+      "step": 21271
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.630527709098756e-06,
+      "loss": 0.1118,
+      "step": 21272
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.629631213608788e-06,
+      "loss": 0.1093,
+      "step": 21273
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.628734761534379e-06,
+      "loss": 0.1012,
+      "step": 21274
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.627838352884435e-06,
+      "loss": 0.1203,
+      "step": 21275
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.626941987667866e-06,
+      "loss": 0.1125,
+      "step": 21276
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.626045665893574e-06,
+      "loss": 0.1013,
+      "step": 21277
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.625149387570454e-06,
+      "loss": 0.1009,
+      "step": 21278
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.624253152707424e-06,
+      "loss": 0.1095,
+      "step": 21279
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.623356961313378e-06,
+      "loss": 0.0944,
+      "step": 21280
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.622460813397224e-06,
+      "loss": 0.1142,
+      "step": 21281
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6215647089678605e-06,
+      "loss": 0.0974,
+      "step": 21282
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.620668648034189e-06,
+      "loss": 0.1115,
+      "step": 21283
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.619772630605112e-06,
+      "loss": 0.1248,
+      "step": 21284
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.618876656689534e-06,
+      "loss": 0.0967,
+      "step": 21285
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.617980726296355e-06,
+      "loss": 0.095,
+      "step": 21286
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.617084839434469e-06,
+      "loss": 0.0947,
+      "step": 21287
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.616188996112783e-06,
+      "loss": 0.121,
+      "step": 21288
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.615293196340191e-06,
+      "loss": 0.0968,
+      "step": 21289
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.614397440125599e-06,
+      "loss": 0.098,
+      "step": 21290
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.613501727477899e-06,
+      "loss": 0.1156,
+      "step": 21291
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.61260605840599e-06,
+      "loss": 0.1367,
+      "step": 21292
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.611710432918768e-06,
+      "loss": 0.1734,
+      "step": 21293
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.610814851025138e-06,
+      "loss": 0.0877,
+      "step": 21294
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.609919312733987e-06,
+      "loss": 0.0996,
+      "step": 21295
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.609023818054221e-06,
+      "loss": 0.1248,
+      "step": 21296
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.60812836699473e-06,
+      "loss": 0.1061,
+      "step": 21297
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.607232959564408e-06,
+      "loss": 0.1182,
+      "step": 21298
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606337595772155e-06,
+      "loss": 0.069,
+      "step": 21299
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.60544227562686e-06,
+      "loss": 0.0745,
+      "step": 21300
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.604546999137424e-06,
+      "loss": 0.1331,
+      "step": 21301
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.603651766312732e-06,
+      "loss": 0.1119,
+      "step": 21302
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.602756577161688e-06,
+      "loss": 0.1635,
+      "step": 21303
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.601861431693174e-06,
+      "loss": 0.103,
+      "step": 21304
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.600966329916091e-06,
+      "loss": 0.0919,
+      "step": 21305
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.600071271839329e-06,
+      "loss": 0.1011,
+      "step": 21306
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.599176257471775e-06,
+      "loss": 0.1133,
+      "step": 21307
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5982812868223245e-06,
+      "loss": 0.0843,
+      "step": 21308
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.597386359899866e-06,
+      "loss": 0.1333,
+      "step": 21309
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.596491476713293e-06,
+      "loss": 0.1057,
+      "step": 21310
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.59559663727149e-06,
+      "loss": 0.1294,
+      "step": 21311
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.594701841583352e-06,
+      "loss": 0.0996,
+      "step": 21312
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.593807089657764e-06,
+      "loss": 0.1391,
+      "step": 21313
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.592912381503618e-06,
+      "loss": 0.1241,
+      "step": 21314
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.592017717129801e-06,
+      "loss": 0.0959,
+      "step": 21315
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.591123096545196e-06,
+      "loss": 0.1075,
+      "step": 21316
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.590228519758698e-06,
+      "loss": 0.1261,
+      "step": 21317
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.589333986779184e-06,
+      "loss": 0.1254,
+      "step": 21318
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.588439497615548e-06,
+      "loss": 0.1251,
+      "step": 21319
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.587545052276678e-06,
+      "loss": 0.1381,
+      "step": 21320
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.586650650771456e-06,
+      "loss": 0.128,
+      "step": 21321
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.585756293108761e-06,
+      "loss": 0.1173,
+      "step": 21322
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.584861979297489e-06,
+      "loss": 0.1302,
+      "step": 21323
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.583967709346514e-06,
+      "loss": 0.0933,
+      "step": 21324
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.583073483264727e-06,
+      "loss": 0.0976,
+      "step": 21325
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.582179301061009e-06,
+      "loss": 0.1425,
+      "step": 21326
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.581285162744239e-06,
+      "loss": 0.1094,
+      "step": 21327
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.580391068323302e-06,
+      "loss": 0.1275,
+      "step": 21328
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.579497017807085e-06,
+      "loss": 0.1014,
+      "step": 21329
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.578603011204466e-06,
+      "loss": 0.0938,
+      "step": 21330
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.57770904852432e-06,
+      "loss": 0.0981,
+      "step": 21331
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.576815129775539e-06,
+      "loss": 0.121,
+      "step": 21332
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.575921254966992e-06,
+      "loss": 0.1388,
+      "step": 21333
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.575027424107564e-06,
+      "loss": 0.0684,
+      "step": 21334
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.574133637206144e-06,
+      "loss": 0.1216,
+      "step": 21335
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.573239894271593e-06,
+      "loss": 0.1019,
+      "step": 21336
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.572346195312797e-06,
+      "loss": 0.1166,
+      "step": 21337
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.571452540338641e-06,
+      "loss": 0.0802,
+      "step": 21338
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.570558929357995e-06,
+      "loss": 0.088,
+      "step": 21339
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.569665362379735e-06,
+      "loss": 0.1014,
+      "step": 21340
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.568771839412745e-06,
+      "loss": 0.1171,
+      "step": 21341
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.567878360465893e-06,
+      "loss": 0.1295,
+      "step": 21342
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.566984925548058e-06,
+      "loss": 0.1208,
+      "step": 21343
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.566091534668124e-06,
+      "loss": 0.1011,
+      "step": 21344
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.565198187834952e-06,
+      "loss": 0.1113,
+      "step": 21345
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.564304885057421e-06,
+      "loss": 0.1201,
+      "step": 21346
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5634116263444125e-06,
+      "loss": 0.1733,
+      "step": 21347
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.56251841170479e-06,
+      "loss": 0.1256,
+      "step": 21348
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.561625241147437e-06,
+      "loss": 0.1088,
+      "step": 21349
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.56073211468122e-06,
+      "loss": 0.1117,
+      "step": 21350
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.559839032315006e-06,
+      "loss": 0.142,
+      "step": 21351
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.558945994057676e-06,
+      "loss": 0.093,
+      "step": 21352
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.558052999918102e-06,
+      "loss": 0.093,
+      "step": 21353
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.557160049905152e-06,
+      "loss": 0.1052,
+      "step": 21354
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.556267144027692e-06,
+      "loss": 0.1241,
+      "step": 21355
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.555374282294601e-06,
+      "loss": 0.1129,
+      "step": 21356
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.55448146471474e-06,
+      "loss": 0.1151,
+      "step": 21357
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.553588691296983e-06,
+      "loss": 0.1049,
+      "step": 21358
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.552695962050205e-06,
+      "loss": 0.1452,
+      "step": 21359
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.551803276983262e-06,
+      "loss": 0.1083,
+      "step": 21360
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.550910636105027e-06,
+      "loss": 0.088,
+      "step": 21361
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.550018039424372e-06,
+      "loss": 0.0773,
+      "step": 21362
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.549125486950158e-06,
+      "loss": 0.109,
+      "step": 21363
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.548232978691257e-06,
+      "loss": 0.1463,
+      "step": 21364
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.547340514656534e-06,
+      "loss": 0.0895,
+      "step": 21365
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.546448094854849e-06,
+      "loss": 0.1163,
+      "step": 21366
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.545555719295071e-06,
+      "loss": 0.1005,
+      "step": 21367
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.544663387986073e-06,
+      "loss": 0.0874,
+      "step": 21368
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.543771100936706e-06,
+      "loss": 0.1067,
+      "step": 21369
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.542878858155839e-06,
+      "loss": 0.1059,
+      "step": 21370
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.541986659652341e-06,
+      "loss": 0.0925,
+      "step": 21371
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5410945054350675e-06,
+      "loss": 0.1101,
+      "step": 21372
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.540202395512888e-06,
+      "loss": 0.0859,
+      "step": 21373
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.53931032989466e-06,
+      "loss": 0.113,
+      "step": 21374
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.538418308589246e-06,
+      "loss": 0.1118,
+      "step": 21375
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.537526331605506e-06,
+      "loss": 0.1057,
+      "step": 21376
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.536634398952307e-06,
+      "loss": 0.1011,
+      "step": 21377
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5357425106385075e-06,
+      "loss": 0.0917,
+      "step": 21378
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5348506666729615e-06,
+      "loss": 0.0931,
+      "step": 21379
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.533958867064535e-06,
+      "loss": 0.1271,
+      "step": 21380
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.533067111822084e-06,
+      "loss": 0.1359,
+      "step": 21381
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.53217540095447e-06,
+      "loss": 0.1163,
+      "step": 21382
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.531283734470551e-06,
+      "loss": 0.0986,
+      "step": 21383
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.530392112379178e-06,
+      "loss": 0.108,
+      "step": 21384
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.529500534689216e-06,
+      "loss": 0.0971,
+      "step": 21385
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5286090014095225e-06,
+      "loss": 0.101,
+      "step": 21386
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.527717512548949e-06,
+      "loss": 0.1241,
+      "step": 21387
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.526826068116356e-06,
+      "loss": 0.1094,
+      "step": 21388
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5259346681205996e-06,
+      "loss": 0.1137,
+      "step": 21389
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.525043312570528e-06,
+      "loss": 0.1317,
+      "step": 21390
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.524152001475005e-06,
+      "loss": 0.0897,
+      "step": 21391
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.523260734842876e-06,
+      "loss": 0.1502,
+      "step": 21392
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5223695126830035e-06,
+      "loss": 0.1112,
+      "step": 21393
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.521478335004238e-06,
+      "loss": 0.0903,
+      "step": 21394
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.520587201815428e-06,
+      "loss": 0.1136,
+      "step": 21395
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.519696113125429e-06,
+      "loss": 0.1218,
+      "step": 21396
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.518805068943098e-06,
+      "loss": 0.1114,
+      "step": 21397
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5179140692772816e-06,
+      "loss": 0.1012,
+      "step": 21398
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.51702311413683e-06,
+      "loss": 0.1455,
+      "step": 21399
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.516132203530599e-06,
+      "loss": 0.1501,
+      "step": 21400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.515241337467432e-06,
+      "loss": 0.1035,
+      "step": 21401
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.514350515956186e-06,
+      "loss": 0.1025,
+      "step": 21402
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5134597390057086e-06,
+      "loss": 0.1251,
+      "step": 21403
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5125690066248435e-06,
+      "loss": 0.1121,
+      "step": 21404
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.511678318822443e-06,
+      "loss": 0.115,
+      "step": 21405
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.510787675607361e-06,
+      "loss": 0.1598,
+      "step": 21406
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.509897076988438e-06,
+      "loss": 0.1297,
+      "step": 21407
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5090065229745206e-06,
+      "loss": 0.1115,
+      "step": 21408
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5081160135744605e-06,
+      "loss": 0.0926,
+      "step": 21409
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.507225548797099e-06,
+      "loss": 0.0933,
+      "step": 21410
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.506335128651286e-06,
+      "loss": 0.0771,
+      "step": 21411
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5054447531458725e-06,
+      "loss": 0.0735,
+      "step": 21412
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.504554422289689e-06,
+      "loss": 0.0946,
+      "step": 21413
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.503664136091589e-06,
+      "loss": 0.0946,
+      "step": 21414
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.502773894560418e-06,
+      "loss": 0.1552,
+      "step": 21415
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.501883697705015e-06,
+      "loss": 0.1352,
+      "step": 21416
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.500993545534228e-06,
+      "loss": 0.1352,
+      "step": 21417
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.500103438056899e-06,
+      "loss": 0.0938,
+      "step": 21418
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.499213375281864e-06,
+      "loss": 0.1131,
+      "step": 21419
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.49832335721797e-06,
+      "loss": 0.1365,
+      "step": 21420
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.497433383874067e-06,
+      "loss": 0.1154,
+      "step": 21421
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.496543455258978e-06,
+      "loss": 0.1241,
+      "step": 21422
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.495653571381554e-06,
+      "loss": 0.067,
+      "step": 21423
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.494763732250638e-06,
+      "loss": 0.1275,
+      "step": 21424
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.493873937875063e-06,
+      "loss": 0.1382,
+      "step": 21425
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.492984188263675e-06,
+      "loss": 0.1125,
+      "step": 21426
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.492094483425308e-06,
+      "loss": 0.0996,
+      "step": 21427
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4912048233687984e-06,
+      "loss": 0.1334,
+      "step": 21428
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.490315208102986e-06,
+      "loss": 0.1146,
+      "step": 21429
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.489425637636714e-06,
+      "loss": 0.1561,
+      "step": 21430
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.488536111978816e-06,
+      "loss": 0.1184,
+      "step": 21431
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.487646631138122e-06,
+      "loss": 0.1261,
+      "step": 21432
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.48675719512348e-06,
+      "loss": 0.1319,
+      "step": 21433
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.485867803943713e-06,
+      "loss": 0.119,
+      "step": 21434
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.484978457607664e-06,
+      "loss": 0.106,
+      "step": 21435
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.484089156124175e-06,
+      "loss": 0.1157,
+      "step": 21436
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.483199899502063e-06,
+      "loss": 0.1185,
+      "step": 21437
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.48231068775017e-06,
+      "loss": 0.124,
+      "step": 21438
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.481421520877335e-06,
+      "loss": 0.128,
+      "step": 21439
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.480532398892384e-06,
+      "loss": 0.1236,
+      "step": 21440
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.479643321804155e-06,
+      "loss": 0.0903,
+      "step": 21441
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.478754289621477e-06,
+      "loss": 0.1256,
+      "step": 21442
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.477865302353178e-06,
+      "loss": 0.0978,
+      "step": 21443
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.476976360008093e-06,
+      "loss": 0.1318,
+      "step": 21444
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.476087462595058e-06,
+      "loss": 0.098,
+      "step": 21445
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.475198610122897e-06,
+      "loss": 0.1153,
+      "step": 21446
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.474309802600438e-06,
+      "loss": 0.0865,
+      "step": 21447
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.473421040036517e-06,
+      "loss": 0.1323,
+      "step": 21448
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.472532322439957e-06,
+      "loss": 0.0921,
+      "step": 21449
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.471643649819594e-06,
+      "loss": 0.124,
+      "step": 21450
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.470755022184249e-06,
+      "loss": 0.0967,
+      "step": 21451
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.469866439542751e-06,
+      "loss": 0.1193,
+      "step": 21452
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4689779019039255e-06,
+      "loss": 0.1041,
+      "step": 21453
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.468089409276608e-06,
+      "loss": 0.1136,
+      "step": 21454
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.467200961669619e-06,
+      "loss": 0.1295,
+      "step": 21455
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.466312559091779e-06,
+      "loss": 0.1085,
+      "step": 21456
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.465424201551925e-06,
+      "loss": 0.1396,
+      "step": 21457
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.464535889058872e-06,
+      "loss": 0.1102,
+      "step": 21458
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.463647621621451e-06,
+      "loss": 0.1169,
+      "step": 21459
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.462759399248485e-06,
+      "loss": 0.1194,
+      "step": 21460
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.461871221948791e-06,
+      "loss": 0.1227,
+      "step": 21461
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4609830897311985e-06,
+      "loss": 0.1003,
+      "step": 21462
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.0911,
+      "step": 21463
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4592069605776096e-06,
+      "loss": 0.087,
+      "step": 21464
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.458318963659257e-06,
+      "loss": 0.1547,
+      "step": 21465
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.457431011858294e-06,
+      "loss": 0.1376,
+      "step": 21466
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.456543105183536e-06,
+      "loss": 0.0928,
+      "step": 21467
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4556552436438125e-06,
+      "loss": 0.0913,
+      "step": 21468
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.454767427247935e-06,
+      "loss": 0.1202,
+      "step": 21469
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4538796560047335e-06,
+      "loss": 0.114,
+      "step": 21470
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.452991929923017e-06,
+      "loss": 0.1284,
+      "step": 21471
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4521042490116115e-06,
+      "loss": 0.1343,
+      "step": 21472
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.451216613279329e-06,
+      "loss": 0.1047,
+      "step": 21473
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.450329022734995e-06,
+      "loss": 0.1257,
+      "step": 21474
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4494414773874205e-06,
+      "loss": 0.1137,
+      "step": 21475
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.448553977245424e-06,
+      "loss": 0.0836,
+      "step": 21476
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.447666522317823e-06,
+      "loss": 0.1148,
+      "step": 21477
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.44677911261343e-06,
+      "loss": 0.0742,
+      "step": 21478
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.445891748141068e-06,
+      "loss": 0.1244,
+      "step": 21479
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.445004428909544e-06,
+      "loss": 0.1425,
+      "step": 21480
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.444117154927682e-06,
+      "loss": 0.1095,
+      "step": 21481
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.443229926204284e-06,
+      "loss": 0.0935,
+      "step": 21482
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4423427427481755e-06,
+      "loss": 0.0654,
+      "step": 21483
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.441455604568164e-06,
+      "loss": 0.0997,
+      "step": 21484
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.440568511673061e-06,
+      "loss": 0.1018,
+      "step": 21485
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.439681464071685e-06,
+      "loss": 0.1163,
+      "step": 21486
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.438794461772839e-06,
+      "loss": 0.0824,
+      "step": 21487
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.43790750478534e-06,
+      "loss": 0.1132,
+      "step": 21488
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.437020593118003e-06,
+      "loss": 0.1154,
+      "step": 21489
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4361337267796355e-06,
+      "loss": 0.0854,
+      "step": 21490
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.435246905779041e-06,
+      "loss": 0.117,
+      "step": 21491
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.434360130125041e-06,
+      "loss": 0.1406,
+      "step": 21492
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.433473399826433e-06,
+      "loss": 0.0867,
+      "step": 21493
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.432586714892036e-06,
+      "loss": 0.1112,
+      "step": 21494
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.431700075330654e-06,
+      "loss": 0.132,
+      "step": 21495
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.43081348115109e-06,
+      "loss": 0.0859,
+      "step": 21496
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.429926932362158e-06,
+      "loss": 0.0958,
+      "step": 21497
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.429040428972666e-06,
+      "loss": 0.1275,
+      "step": 21498
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.428153970991419e-06,
+      "loss": 0.0961,
+      "step": 21499
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.427267558427217e-06,
+      "loss": 0.1191,
+      "step": 21500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.426381191288877e-06,
+      "loss": 0.121,
+      "step": 21501
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.42549486958519e-06,
+      "loss": 0.1304,
+      "step": 21502
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4246085933249725e-06,
+      "loss": 0.1652,
+      "step": 21503
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.423722362517032e-06,
+      "loss": 0.1134,
+      "step": 21504
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.422836177170159e-06,
+      "loss": 0.1026,
+      "step": 21505
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.421950037293161e-06,
+      "loss": 0.1348,
+      "step": 21506
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.421063942894848e-06,
+      "loss": 0.1125,
+      "step": 21507
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.42017789398402e-06,
+      "loss": 0.0928,
+      "step": 21508
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.419291890569471e-06,
+      "loss": 0.1264,
+      "step": 21509
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4184059326600136e-06,
+      "loss": 0.1311,
+      "step": 21510
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.41752002026444e-06,
+      "loss": 0.1113,
+      "step": 21511
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.416634153391554e-06,
+      "loss": 0.1149,
+      "step": 21512
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.415748332050166e-06,
+      "loss": 0.1006,
+      "step": 21513
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.414862556249059e-06,
+      "loss": 0.1108,
+      "step": 21514
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4139768259970385e-06,
+      "loss": 0.1103,
+      "step": 21515
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.413091141302911e-06,
+      "loss": 0.0977,
+      "step": 21516
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.412205502175462e-06,
+      "loss": 0.1272,
+      "step": 21517
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.411319908623503e-06,
+      "loss": 0.1206,
+      "step": 21518
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.410434360655824e-06,
+      "loss": 0.1008,
+      "step": 21519
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.40954885828122e-06,
+      "loss": 0.0998,
+      "step": 21520
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.408663401508489e-06,
+      "loss": 0.1319,
+      "step": 21521
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.407777990346433e-06,
+      "loss": 0.0994,
+      "step": 21522
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.406892624803845e-06,
+      "loss": 0.1206,
+      "step": 21523
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.406007304889513e-06,
+      "loss": 0.101,
+      "step": 21524
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.405122030612243e-06,
+      "loss": 0.1232,
+      "step": 21525
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.40423680198082e-06,
+      "loss": 0.1206,
+      "step": 21526
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.403351619004043e-06,
+      "loss": 0.1077,
+      "step": 21527
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4024664816907115e-06,
+      "loss": 0.1356,
+      "step": 21528
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.401581390049602e-06,
+      "loss": 0.0998,
+      "step": 21529
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.400696344089519e-06,
+      "loss": 0.0962,
+      "step": 21530
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.399811343819254e-06,
+      "loss": 0.0855,
+      "step": 21531
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.398926389247595e-06,
+      "loss": 0.1229,
+      "step": 21532
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.398041480383338e-06,
+      "loss": 0.0855,
+      "step": 21533
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3971566172352704e-06,
+      "loss": 0.1209,
+      "step": 21534
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.39627179981218e-06,
+      "loss": 0.1183,
+      "step": 21535
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.39538702812286e-06,
+      "loss": 0.1193,
+      "step": 21536
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.394502302176106e-06,
+      "loss": 0.125,
+      "step": 21537
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.393617621980691e-06,
+      "loss": 0.1228,
+      "step": 21538
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.392732987545415e-06,
+      "loss": 0.1699,
+      "step": 21539
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.391848398879067e-06,
+      "loss": 0.1164,
+      "step": 21540
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3909638559904276e-06,
+      "loss": 0.0922,
+      "step": 21541
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.390079358888291e-06,
+      "loss": 0.1033,
+      "step": 21542
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.389194907581441e-06,
+      "loss": 0.0996,
+      "step": 21543
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.38831050207866e-06,
+      "loss": 0.1279,
+      "step": 21544
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.387426142388736e-06,
+      "loss": 0.1,
+      "step": 21545
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.38654182852046e-06,
+      "loss": 0.1137,
+      "step": 21546
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.385657560482611e-06,
+      "loss": 0.0962,
+      "step": 21547
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.384773338283973e-06,
+      "loss": 0.1254,
+      "step": 21548
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.383889161933333e-06,
+      "loss": 0.0943,
+      "step": 21549
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.38300503143947e-06,
+      "loss": 0.1098,
+      "step": 21550
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.382120946811174e-06,
+      "loss": 0.1121,
+      "step": 21551
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.381236908057224e-06,
+      "loss": 0.1115,
+      "step": 21552
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.380352915186396e-06,
+      "loss": 0.0986,
+      "step": 21553
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3794689682074816e-06,
+      "loss": 0.0967,
+      "step": 21554
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.378585067129256e-06,
+      "loss": 0.1195,
+      "step": 21555
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3777012119605e-06,
+      "loss": 0.1122,
+      "step": 21556
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.376817402709999e-06,
+      "loss": 0.1453,
+      "step": 21557
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.37593363938653e-06,
+      "loss": 0.1149,
+      "step": 21558
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.375049921998867e-06,
+      "loss": 0.0961,
+      "step": 21559
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.374166250555799e-06,
+      "loss": 0.1124,
+      "step": 21560
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.373282625066094e-06,
+      "loss": 0.1119,
+      "step": 21561
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.372399045538539e-06,
+      "loss": 0.1107,
+      "step": 21562
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.371515511981909e-06,
+      "loss": 0.11,
+      "step": 21563
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.370632024404974e-06,
+      "loss": 0.101,
+      "step": 21564
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.369748582816517e-06,
+      "loss": 0.1442,
+      "step": 21565
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.368865187225318e-06,
+      "loss": 0.1204,
+      "step": 21566
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.367981837640147e-06,
+      "loss": 0.1005,
+      "step": 21567
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.367098534069777e-06,
+      "loss": 0.0952,
+      "step": 21568
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.366215276522991e-06,
+      "loss": 0.1697,
+      "step": 21569
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.365332065008554e-06,
+      "loss": 0.0788,
+      "step": 21570
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.364448899535248e-06,
+      "loss": 0.1172,
+      "step": 21571
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.363565780111844e-06,
+      "loss": 0.128,
+      "step": 21572
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36268270674711e-06,
+      "loss": 0.0864,
+      "step": 21573
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.361799679449821e-06,
+      "loss": 0.0954,
+      "step": 21574
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.360916698228756e-06,
+      "loss": 0.1146,
+      "step": 21575
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36003376309268e-06,
+      "loss": 0.1054,
+      "step": 21576
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.359150874050363e-06,
+      "loss": 0.1646,
+      "step": 21577
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.358268031110581e-06,
+      "loss": 0.115,
+      "step": 21578
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.357385234282095e-06,
+      "loss": 0.0896,
+      "step": 21579
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.356502483573684e-06,
+      "loss": 0.1297,
+      "step": 21580
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.35561977899412e-06,
+      "loss": 0.1287,
+      "step": 21581
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.354737120552159e-06,
+      "loss": 0.0811,
+      "step": 21582
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.353854508256576e-06,
+      "loss": 0.0917,
+      "step": 21583
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.352971942116144e-06,
+      "loss": 0.1341,
+      "step": 21584
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3520894221396214e-06,
+      "loss": 0.1255,
+      "step": 21585
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.351206948335784e-06,
+      "loss": 0.0862,
+      "step": 21586
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.350324520713394e-06,
+      "loss": 0.0945,
+      "step": 21587
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.349442139281214e-06,
+      "loss": 0.1319,
+      "step": 21588
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.348559804048013e-06,
+      "loss": 0.1687,
+      "step": 21589
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3476775150225645e-06,
+      "loss": 0.1171,
+      "step": 21590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.346795272213617e-06,
+      "loss": 0.0884,
+      "step": 21591
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.345913075629942e-06,
+      "loss": 0.124,
+      "step": 21592
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.345030925280308e-06,
+      "loss": 0.1012,
+      "step": 21593
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.344148821173472e-06,
+      "loss": 0.1092,
+      "step": 21594
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.343266763318202e-06,
+      "loss": 0.1477,
+      "step": 21595
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.342384751723259e-06,
+      "loss": 0.1228,
+      "step": 21596
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.341502786397399e-06,
+      "loss": 0.0828,
+      "step": 21597
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.340620867349389e-06,
+      "loss": 0.1421,
+      "step": 21598
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.339738994587993e-06,
+      "loss": 0.1066,
+      "step": 21599
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.338857168121967e-06,
+      "loss": 0.1212,
+      "step": 21600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.337975387960068e-06,
+      "loss": 0.1171,
+      "step": 21601
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.337093654111065e-06,
+      "loss": 0.096,
+      "step": 21602
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3362119665837084e-06,
+      "loss": 0.0947,
+      "step": 21603
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.335330325386758e-06,
+      "loss": 0.083,
+      "step": 21604
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.334448730528985e-06,
+      "loss": 0.1228,
+      "step": 21605
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.333567182019128e-06,
+      "loss": 0.1177,
+      "step": 21606
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.332685679865953e-06,
+      "loss": 0.0847,
+      "step": 21607
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3318042240782205e-06,
+      "loss": 0.1222,
+      "step": 21608
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.33092281466468e-06,
+      "loss": 0.12,
+      "step": 21609
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.330041451634096e-06,
+      "loss": 0.0944,
+      "step": 21610
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.329160134995219e-06,
+      "loss": 0.1143,
+      "step": 21611
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.328278864756799e-06,
+      "loss": 0.1225,
+      "step": 21612
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.327397640927596e-06,
+      "loss": 0.1082,
+      "step": 21613
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.326516463516368e-06,
+      "loss": 0.0859,
+      "step": 21614
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.325635332531864e-06,
+      "loss": 0.0862,
+      "step": 21615
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.324754247982834e-06,
+      "loss": 0.1129,
+      "step": 21616
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.323873209878039e-06,
+      "loss": 0.1008,
+      "step": 21617
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.322992218226221e-06,
+      "loss": 0.0626,
+      "step": 21618
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.322111273036144e-06,
+      "loss": 0.1002,
+      "step": 21619
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3212303743165526e-06,
+      "loss": 0.0865,
+      "step": 21620
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3203495220761936e-06,
+      "loss": 0.1093,
+      "step": 21621
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.319468716323822e-06,
+      "loss": 0.1173,
+      "step": 21622
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.318587957068191e-06,
+      "loss": 0.0896,
+      "step": 21623
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.317707244318047e-06,
+      "loss": 0.1222,
+      "step": 21624
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.316826578082136e-06,
+      "loss": 0.1002,
+      "step": 21625
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.315945958369213e-06,
+      "loss": 0.0912,
+      "step": 21626
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.315065385188019e-06,
+      "loss": 0.143,
+      "step": 21627
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3141848585473084e-06,
+      "loss": 0.1468,
+      "step": 21628
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.313304378455826e-06,
+      "loss": 0.1513,
+      "step": 21629
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.312423944922314e-06,
+      "loss": 0.0815,
+      "step": 21630
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.311543557955522e-06,
+      "loss": 0.0962,
+      "step": 21631
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.310663217564201e-06,
+      "loss": 0.1187,
+      "step": 21632
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.309782923757088e-06,
+      "loss": 0.0794,
+      "step": 21633
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3089026765429355e-06,
+      "loss": 0.1314,
+      "step": 21634
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.308022475930484e-06,
+      "loss": 0.1129,
+      "step": 21635
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3071423219284735e-06,
+      "loss": 0.1413,
+      "step": 21636
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.306262214545657e-06,
+      "loss": 0.1296,
+      "step": 21637
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.305382153790765e-06,
+      "loss": 0.1091,
+      "step": 21638
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3045021396725535e-06,
+      "loss": 0.1622,
+      "step": 21639
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.303622172199755e-06,
+      "loss": 0.0895,
+      "step": 21640
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3027422513811175e-06,
+      "loss": 0.1232,
+      "step": 21641
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.301862377225375e-06,
+      "loss": 0.1036,
+      "step": 21642
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.300982549741278e-06,
+      "loss": 0.0978,
+      "step": 21643
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.300102768937561e-06,
+      "loss": 0.1176,
+      "step": 21644
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.299223034822959e-06,
+      "loss": 0.0759,
+      "step": 21645
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2983433474062204e-06,
+      "loss": 0.1213,
+      "step": 21646
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.297463706696076e-06,
+      "loss": 0.1436,
+      "step": 21647
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.296584112701273e-06,
+      "loss": 0.1229,
+      "step": 21648
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.295704565430542e-06,
+      "loss": 0.0863,
+      "step": 21649
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2948250648926245e-06,
+      "loss": 0.142,
+      "step": 21650
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.293945611096254e-06,
+      "loss": 0.1112,
+      "step": 21651
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.293066204050172e-06,
+      "loss": 0.1054,
+      "step": 21652
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.292186843763112e-06,
+      "loss": 0.1098,
+      "step": 21653
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.291307530243805e-06,
+      "loss": 0.1135,
+      "step": 21654
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.290428263500996e-06,
+      "loss": 0.1121,
+      "step": 21655
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.289549043543409e-06,
+      "loss": 0.12,
+      "step": 21656
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.288669870379786e-06,
+      "loss": 0.1012,
+      "step": 21657
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.287790744018859e-06,
+      "loss": 0.1395,
+      "step": 21658
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.286911664469363e-06,
+      "loss": 0.1149,
+      "step": 21659
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.286032631740023e-06,
+      "loss": 0.1333,
+      "step": 21660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2851536458395826e-06,
+      "loss": 0.1099,
+      "step": 21661
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.284274706776762e-06,
+      "loss": 0.0896,
+      "step": 21662
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.283395814560303e-06,
+      "loss": 0.077,
+      "step": 21663
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.282516969198933e-06,
+      "loss": 0.1055,
+      "step": 21664
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.281638170701378e-06,
+      "loss": 0.1041,
+      "step": 21665
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.280759419076371e-06,
+      "loss": 0.1186,
+      "step": 21666
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.279880714332646e-06,
+      "loss": 0.1015,
+      "step": 21667
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.279002056478928e-06,
+      "loss": 0.109,
+      "step": 21668
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.278123445523944e-06,
+      "loss": 0.0946,
+      "step": 21669
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.277244881476427e-06,
+      "loss": 0.1143,
+      "step": 21670
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.276366364345098e-06,
+      "loss": 0.1063,
+      "step": 21671
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.275487894138688e-06,
+      "loss": 0.1241,
+      "step": 21672
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.274609470865932e-06,
+      "loss": 0.0585,
+      "step": 21673
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.273731094535541e-06,
+      "loss": 0.0926,
+      "step": 21674
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2728527651562475e-06,
+      "loss": 0.1055,
+      "step": 21675
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2719744827367815e-06,
+      "loss": 0.0791,
+      "step": 21676
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.271096247285865e-06,
+      "loss": 0.0798,
+      "step": 21677
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.270218058812217e-06,
+      "loss": 0.1143,
+      "step": 21678
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.26933991732457e-06,
+      "loss": 0.1065,
+      "step": 21679
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.268461822831638e-06,
+      "loss": 0.1248,
+      "step": 21680
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2675837753421525e-06,
+      "loss": 0.15,
+      "step": 21681
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.266705774864838e-06,
+      "loss": 0.1108,
+      "step": 21682
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.265827821408407e-06,
+      "loss": 0.1157,
+      "step": 21683
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.264949914981583e-06,
+      "loss": 0.1206,
+      "step": 21684
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.264072055593094e-06,
+      "loss": 0.1343,
+      "step": 21685
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.263194243251655e-06,
+      "loss": 0.1355,
+      "step": 21686
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.262316477965992e-06,
+      "loss": 0.1155,
+      "step": 21687
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.261438759744819e-06,
+      "loss": 0.1265,
+      "step": 21688
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.260561088596854e-06,
+      "loss": 0.1236,
+      "step": 21689
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.259683464530819e-06,
+      "loss": 0.1282,
+      "step": 21690
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.258805887555436e-06,
+      "loss": 0.1148,
+      "step": 21691
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.257928357679421e-06,
+      "loss": 0.0943,
+      "step": 21692
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.257050874911483e-06,
+      "loss": 0.0944,
+      "step": 21693
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.256173439260351e-06,
+      "loss": 0.1032,
+      "step": 21694
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.255296050734733e-06,
+      "loss": 0.1037,
+      "step": 21695
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.254418709343347e-06,
+      "loss": 0.0754,
+      "step": 21696
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.253541415094918e-06,
+      "loss": 0.1638,
+      "step": 21697
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.252664167998146e-06,
+      "loss": 0.1355,
+      "step": 21698
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.251786968061752e-06,
+      "loss": 0.1114,
+      "step": 21699
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.250909815294455e-06,
+      "loss": 0.1385,
+      "step": 21700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.25003270970496e-06,
+      "loss": 0.1295,
+      "step": 21701
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.249155651301989e-06,
+      "loss": 0.1035,
+      "step": 21702
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.24827864009425e-06,
+      "loss": 0.1137,
+      "step": 21703
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.247401676090451e-06,
+      "loss": 0.0753,
+      "step": 21704
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2465247592993095e-06,
+      "loss": 0.0907,
+      "step": 21705
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.245647889729544e-06,
+      "loss": 0.0928,
+      "step": 21706
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.244771067389849e-06,
+      "loss": 0.1128,
+      "step": 21707
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.243894292288945e-06,
+      "loss": 0.1083,
+      "step": 21708
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.243017564435542e-06,
+      "loss": 0.1198,
+      "step": 21709
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.242140883838344e-06,
+      "loss": 0.1057,
+      "step": 21710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.241264250506067e-06,
+      "loss": 0.1179,
+      "step": 21711
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.240387664447417e-06,
+      "loss": 0.0994,
+      "step": 21712
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.239511125671097e-06,
+      "loss": 0.1265,
+      "step": 21713
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.238634634185824e-06,
+      "loss": 0.117,
+      "step": 21714
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.237758190000295e-06,
+      "loss": 0.0936,
+      "step": 21715
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.236881793123225e-06,
+      "loss": 0.1095,
+      "step": 21716
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.236005443563315e-06,
+      "loss": 0.1135,
+      "step": 21717
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.235129141329274e-06,
+      "loss": 0.0793,
+      "step": 21718
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.234252886429803e-06,
+      "loss": 0.1605,
+      "step": 21719
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.233376678873614e-06,
+      "loss": 0.0754,
+      "step": 21720
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.232500518669406e-06,
+      "loss": 0.1115,
+      "step": 21721
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.23162440582588e-06,
+      "loss": 0.0991,
+      "step": 21722
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2307483403517476e-06,
+      "loss": 0.1027,
+      "step": 21723
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.229872322255702e-06,
+      "loss": 0.1195,
+      "step": 21724
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2289963515464525e-06,
+      "loss": 0.1326,
+      "step": 21725
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.228120428232703e-06,
+      "loss": 0.1155,
+      "step": 21726
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.227244552323149e-06,
+      "loss": 0.0869,
+      "step": 21727
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.226368723826494e-06,
+      "loss": 0.1004,
+      "step": 21728
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.22549294275144e-06,
+      "loss": 0.1109,
+      "step": 21729
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.224617209106681e-06,
+      "loss": 0.136,
+      "step": 21730
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2237415229009255e-06,
+      "loss": 0.1275,
+      "step": 21731
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.222865884142869e-06,
+      "loss": 0.0881,
+      "step": 21732
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.221990292841205e-06,
+      "loss": 0.1212,
+      "step": 21733
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.221114749004636e-06,
+      "loss": 0.1273,
+      "step": 21734
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.220239252641863e-06,
+      "loss": 0.1303,
+      "step": 21735
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.219363803761581e-06,
+      "loss": 0.0807,
+      "step": 21736
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.218488402372481e-06,
+      "loss": 0.067,
+      "step": 21737
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.21761304848327e-06,
+      "loss": 0.1131,
+      "step": 21738
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.216737742102632e-06,
+      "loss": 0.1449,
+      "step": 21739
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.215862483239273e-06,
+      "loss": 0.1093,
+      "step": 21740
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2149872719018836e-06,
+      "loss": 0.1019,
+      "step": 21741
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.214112108099153e-06,
+      "loss": 0.1278,
+      "step": 21742
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.213236991839781e-06,
+      "loss": 0.0963,
+      "step": 21743
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.212361923132463e-06,
+      "loss": 0.0901,
+      "step": 21744
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.211486901985889e-06,
+      "loss": 0.0948,
+      "step": 21745
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.210611928408747e-06,
+      "loss": 0.0928,
+      "step": 21746
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.209737002409738e-06,
+      "loss": 0.0665,
+      "step": 21747
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.208862123997547e-06,
+      "loss": 0.12,
+      "step": 21748
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.207987293180865e-06,
+      "loss": 0.1037,
+      "step": 21749
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.207112509968394e-06,
+      "loss": 0.1384,
+      "step": 21750
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.206237774368808e-06,
+      "loss": 0.0843,
+      "step": 21751
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.205363086390801e-06,
+      "loss": 0.1227,
+      "step": 21752
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.20448844604307e-06,
+      "loss": 0.1106,
+      "step": 21753
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.203613853334295e-06,
+      "loss": 0.1328,
+      "step": 21754
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.202739308273172e-06,
+      "loss": 0.0713,
+      "step": 21755
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.201864810868386e-06,
+      "loss": 0.131,
+      "step": 21756
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.200990361128618e-06,
+      "loss": 0.1116,
+      "step": 21757
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.20011595906256e-06,
+      "loss": 0.1151,
+      "step": 21758
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.199241604678905e-06,
+      "loss": 0.0818,
+      "step": 21759
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.198367297986326e-06,
+      "loss": 0.1367,
+      "step": 21760
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.197493038993513e-06,
+      "loss": 0.1225,
+      "step": 21761
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.196618827709159e-06,
+      "loss": 0.1138,
+      "step": 21762
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.195744664141935e-06,
+      "loss": 0.0978,
+      "step": 21763
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.194870548300539e-06,
+      "loss": 0.1071,
+      "step": 21764
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.193996480193646e-06,
+      "loss": 0.1393,
+      "step": 21765
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.193122459829935e-06,
+      "loss": 0.0932,
+      "step": 21766
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.192248487218097e-06,
+      "loss": 0.0991,
+      "step": 21767
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.191374562366816e-06,
+      "loss": 0.1307,
+      "step": 21768
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.190500685284766e-06,
+      "loss": 0.1106,
+      "step": 21769
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.189626855980629e-06,
+      "loss": 0.1141,
+      "step": 21770
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.188753074463091e-06,
+      "loss": 0.1012,
+      "step": 21771
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.187879340740824e-06,
+      "loss": 0.1075,
+      "step": 21772
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.187005654822515e-06,
+      "loss": 0.0909,
+      "step": 21773
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1861320167168485e-06,
+      "loss": 0.0976,
+      "step": 21774
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1852584264324864e-06,
+      "loss": 0.1861,
+      "step": 21775
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.184384883978117e-06,
+      "loss": 0.1294,
+      "step": 21776
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1835113893624235e-06,
+      "loss": 0.0915,
+      "step": 21777
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.182637942594071e-06,
+      "loss": 0.0759,
+      "step": 21778
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.181764543681748e-06,
+      "loss": 0.1164,
+      "step": 21779
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1808911926341245e-06,
+      "loss": 0.096,
+      "step": 21780
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.180017889459874e-06,
+      "loss": 0.1497,
+      "step": 21781
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.179144634167675e-06,
+      "loss": 0.0921,
+      "step": 21782
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.178271426766209e-06,
+      "loss": 0.1027,
+      "step": 21783
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.177398267264142e-06,
+      "loss": 0.0952,
+      "step": 21784
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.176525155670146e-06,
+      "loss": 0.1043,
+      "step": 21785
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.175652091992906e-06,
+      "loss": 0.1094,
+      "step": 21786
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.174779076241082e-06,
+      "loss": 0.0864,
+      "step": 21787
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.173906108423357e-06,
+      "loss": 0.1099,
+      "step": 21788
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1730331885484e-06,
+      "loss": 0.1226,
+      "step": 21789
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1721603166248755e-06,
+      "loss": 0.1095,
+      "step": 21790
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.171287492661462e-06,
+      "loss": 0.0973,
+      "step": 21791
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.170414716666833e-06,
+      "loss": 0.1552,
+      "step": 21792
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.169541988649654e-06,
+      "loss": 0.1238,
+      "step": 21793
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.168669308618593e-06,
+      "loss": 0.1432,
+      "step": 21794
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1677966765823226e-06,
+      "loss": 0.1076,
+      "step": 21795
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.166924092549508e-06,
+      "loss": 0.1335,
+      "step": 21796
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.166051556528823e-06,
+      "loss": 0.0895,
+      "step": 21797
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.165179068528936e-06,
+      "loss": 0.0972,
+      "step": 21798
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1643066285585045e-06,
+      "loss": 0.0884,
+      "step": 21799
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.163434236626201e-06,
+      "loss": 0.1196,
+      "step": 21800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.162561892740699e-06,
+      "loss": 0.0961,
+      "step": 21801
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.161689596910653e-06,
+      "loss": 0.1208,
+      "step": 21802
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.160817349144737e-06,
+      "loss": 0.0988,
+      "step": 21803
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.159945149451613e-06,
+      "loss": 0.17,
+      "step": 21804
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.15907299783994e-06,
+      "loss": 0.1635,
+      "step": 21805
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.158200894318392e-06,
+      "loss": 0.1176,
+      "step": 21806
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.157328838895623e-06,
+      "loss": 0.1025,
+      "step": 21807
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.156456831580305e-06,
+      "loss": 0.0883,
+      "step": 21808
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.155584872381094e-06,
+      "loss": 0.1007,
+      "step": 21809
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.154712961306656e-06,
+      "loss": 0.1069,
+      "step": 21810
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.153841098365648e-06,
+      "loss": 0.1085,
+      "step": 21811
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.152969283566738e-06,
+      "loss": 0.085,
+      "step": 21812
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.152097516918583e-06,
+      "loss": 0.0916,
+      "step": 21813
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.151225798429838e-06,
+      "loss": 0.1181,
+      "step": 21814
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.150354128109173e-06,
+      "loss": 0.0906,
+      "step": 21815
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.149482505965237e-06,
+      "loss": 0.0745,
+      "step": 21816
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.148610932006698e-06,
+      "loss": 0.1069,
+      "step": 21817
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1477394062422056e-06,
+      "loss": 0.0875,
+      "step": 21818
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.146867928680427e-06,
+      "loss": 0.113,
+      "step": 21819
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1459964993300104e-06,
+      "loss": 0.1068,
+      "step": 21820
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1451251181996185e-06,
+      "loss": 0.1197,
+      "step": 21821
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.144253785297908e-06,
+      "loss": 0.1013,
+      "step": 21822
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.143382500633529e-06,
+      "loss": 0.1217,
+      "step": 21823
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1425112642151435e-06,
+      "loss": 0.1205,
+      "step": 21824
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1416400760514e-06,
+      "loss": 0.1211,
+      "step": 21825
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1407689361509565e-06,
+      "loss": 0.1772,
+      "step": 21826
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.13989784452247e-06,
+      "loss": 0.1177,
+      "step": 21827
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.139026801174592e-06,
+      "loss": 0.1108,
+      "step": 21828
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.13815580611597e-06,
+      "loss": 0.1246,
+      "step": 21829
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.137284859355265e-06,
+      "loss": 0.1137,
+      "step": 21830
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.136413960901121e-06,
+      "loss": 0.0966,
+      "step": 21831
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.135543110762198e-06,
+      "loss": 0.0768,
+      "step": 21832
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.134672308947142e-06,
+      "loss": 0.1093,
+      "step": 21833
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.133801555464601e-06,
+      "loss": 0.1009,
+      "step": 21834
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.132930850323228e-06,
+      "loss": 0.09,
+      "step": 21835
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.132060193531676e-06,
+      "loss": 0.0991,
+      "step": 21836
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.131189585098593e-06,
+      "loss": 0.1017,
+      "step": 21837
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.13031902503262e-06,
+      "loss": 0.1018,
+      "step": 21838
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.129448513342417e-06,
+      "loss": 0.1294,
+      "step": 21839
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.12857805003662e-06,
+      "loss": 0.072,
+      "step": 21840
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.127707635123884e-06,
+      "loss": 0.1241,
+      "step": 21841
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1268372686128595e-06,
+      "loss": 0.0754,
+      "step": 21842
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.125966950512181e-06,
+      "loss": 0.1074,
+      "step": 21843
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.125096680830501e-06,
+      "loss": 0.1568,
+      "step": 21844
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1242264595764664e-06,
+      "loss": 0.1188,
+      "step": 21845
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.12335628675872e-06,
+      "loss": 0.0799,
+      "step": 21846
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.122486162385902e-06,
+      "loss": 0.1097,
+      "step": 21847
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1216160864666655e-06,
+      "loss": 0.1195,
+      "step": 21848
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.120746059009645e-06,
+      "loss": 0.1443,
+      "step": 21849
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.119876080023487e-06,
+      "loss": 0.1104,
+      "step": 21850
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.119006149516841e-06,
+      "loss": 0.1111,
+      "step": 21851
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.118136267498335e-06,
+      "loss": 0.0863,
+      "step": 21852
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.117266433976616e-06,
+      "loss": 0.1036,
+      "step": 21853
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.116396648960331e-06,
+      "loss": 0.1131,
+      "step": 21854
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.115526912458113e-06,
+      "loss": 0.1564,
+      "step": 21855
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1146572244786085e-06,
+      "loss": 0.0852,
+      "step": 21856
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1137875850304545e-06,
+      "loss": 0.1225,
+      "step": 21857
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.112917994122284e-06,
+      "loss": 0.1241,
+      "step": 21858
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.112048451762741e-06,
+      "loss": 0.1001,
+      "step": 21859
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.111178957960467e-06,
+      "loss": 0.1282,
+      "step": 21860
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.110309512724095e-06,
+      "loss": 0.1249,
+      "step": 21861
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.109440116062261e-06,
+      "loss": 0.1151,
+      "step": 21862
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.108570767983607e-06,
+      "loss": 0.1363,
+      "step": 21863
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.107701468496762e-06,
+      "loss": 0.0941,
+      "step": 21864
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1068322176103646e-06,
+      "loss": 0.1047,
+      "step": 21865
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.105963015333059e-06,
+      "loss": 0.0916,
+      "step": 21866
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.105093861673465e-06,
+      "loss": 0.0946,
+      "step": 21867
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.104224756640224e-06,
+      "loss": 0.1625,
+      "step": 21868
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.103355700241972e-06,
+      "loss": 0.1432,
+      "step": 21869
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.1024866924873365e-06,
+      "loss": 0.1673,
+      "step": 21870
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.101617733384957e-06,
+      "loss": 0.1309,
+      "step": 21871
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.100748822943462e-06,
+      "loss": 0.088,
+      "step": 21872
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.099879961171481e-06,
+      "loss": 0.1192,
+      "step": 21873
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.099011148077651e-06,
+      "loss": 0.088,
+      "step": 21874
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0981423836706e-06,
+      "loss": 0.1083,
+      "step": 21875
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.097273667958955e-06,
+      "loss": 0.1178,
+      "step": 21876
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.096405000951348e-06,
+      "loss": 0.1097,
+      "step": 21877
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.095536382656414e-06,
+      "loss": 0.1192,
+      "step": 21878
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0946678130827735e-06,
+      "loss": 0.0844,
+      "step": 21879
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.093799292239062e-06,
+      "loss": 0.1181,
+      "step": 21880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.092930820133905e-06,
+      "loss": 0.1035,
+      "step": 21881
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.092062396775926e-06,
+      "loss": 0.0971,
+      "step": 21882
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.091194022173758e-06,
+      "loss": 0.1069,
+      "step": 21883
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.090325696336021e-06,
+      "loss": 0.1047,
+      "step": 21884
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.089457419271349e-06,
+      "loss": 0.1093,
+      "step": 21885
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.088589190988358e-06,
+      "loss": 0.1033,
+      "step": 21886
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.087721011495683e-06,
+      "loss": 0.0884,
+      "step": 21887
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.086852880801942e-06,
+      "loss": 0.0906,
+      "step": 21888
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.085984798915761e-06,
+      "loss": 0.0936,
+      "step": 21889
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.085116765845767e-06,
+      "loss": 0.1203,
+      "step": 21890
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.084248781600574e-06,
+      "loss": 0.1403,
+      "step": 21891
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.083380846188814e-06,
+      "loss": 0.0876,
+      "step": 21892
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0825129596191025e-06,
+      "loss": 0.1181,
+      "step": 21893
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0816451219000626e-06,
+      "loss": 0.1005,
+      "step": 21894
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.080777333040323e-06,
+      "loss": 0.0829,
+      "step": 21895
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.079909593048497e-06,
+      "loss": 0.1205,
+      "step": 21896
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.079041901933202e-06,
+      "loss": 0.0853,
+      "step": 21897
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.078174259703066e-06,
+      "loss": 0.129,
+      "step": 21898
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.077306666366699e-06,
+      "loss": 0.1206,
+      "step": 21899
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.076439121932729e-06,
+      "loss": 0.1315,
+      "step": 21900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.075571626409771e-06,
+      "loss": 0.0709,
+      "step": 21901
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.074704179806438e-06,
+      "loss": 0.1377,
+      "step": 21902
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.07383678213135e-06,
+      "loss": 0.0838,
+      "step": 21903
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.072969433393129e-06,
+      "loss": 0.0846,
+      "step": 21904
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.072102133600386e-06,
+      "loss": 0.0903,
+      "step": 21905
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.071234882761735e-06,
+      "loss": 0.0938,
+      "step": 21906
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.070367680885799e-06,
+      "loss": 0.1122,
+      "step": 21907
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.069500527981184e-06,
+      "loss": 0.1225,
+      "step": 21908
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.068633424056511e-06,
+      "loss": 0.1247,
+      "step": 21909
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0677663691203925e-06,
+      "loss": 0.1274,
+      "step": 21910
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.066899363181435e-06,
+      "loss": 0.0944,
+      "step": 21911
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.066032406248261e-06,
+      "loss": 0.1063,
+      "step": 21912
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.065165498329478e-06,
+      "loss": 0.1702,
+      "step": 21913
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.064298639433703e-06,
+      "loss": 0.1349,
+      "step": 21914
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.063431829569537e-06,
+      "loss": 0.1073,
+      "step": 21915
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.062565068745602e-06,
+      "loss": 0.1647,
+      "step": 21916
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.061698356970501e-06,
+      "loss": 0.1023,
+      "step": 21917
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.060831694252846e-06,
+      "loss": 0.1339,
+      "step": 21918
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.059965080601255e-06,
+      "loss": 0.1241,
+      "step": 21919
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.059098516024321e-06,
+      "loss": 0.1321,
+      "step": 21920
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.058232000530662e-06,
+      "loss": 0.1421,
+      "step": 21921
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.057365534128888e-06,
+      "loss": 0.115,
+      "step": 21922
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.056499116827599e-06,
+      "loss": 0.0984,
+      "step": 21923
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.05563274863541e-06,
+      "loss": 0.1059,
+      "step": 21924
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.054766429560927e-06,
+      "loss": 0.1019,
+      "step": 21925
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.053900159612746e-06,
+      "loss": 0.139,
+      "step": 21926
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.053033938799481e-06,
+      "loss": 0.1147,
+      "step": 21927
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.052167767129743e-06,
+      "loss": 0.1184,
+      "step": 21928
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.051301644612121e-06,
+      "loss": 0.097,
+      "step": 21929
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.050435571255229e-06,
+      "loss": 0.1299,
+      "step": 21930
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.049569547067672e-06,
+      "loss": 0.1512,
+      "step": 21931
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.048703572058047e-06,
+      "loss": 0.1309,
+      "step": 21932
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.047837646234963e-06,
+      "loss": 0.1017,
+      "step": 21933
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.04697176960702e-06,
+      "loss": 0.1371,
+      "step": 21934
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.046105942182815e-06,
+      "loss": 0.111,
+      "step": 21935
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.045240163970953e-06,
+      "loss": 0.1258,
+      "step": 21936
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.044374434980037e-06,
+      "loss": 0.1146,
+      "step": 21937
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0435087552186655e-06,
+      "loss": 0.1025,
+      "step": 21938
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.042643124695434e-06,
+      "loss": 0.1211,
+      "step": 21939
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.041777543418951e-06,
+      "loss": 0.0901,
+      "step": 21940
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.040912011397803e-06,
+      "loss": 0.1572,
+      "step": 21941
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.040046528640596e-06,
+      "loss": 0.0896,
+      "step": 21942
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.039181095155934e-06,
+      "loss": 0.143,
+      "step": 21943
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0383157109524e-06,
+      "loss": 0.0875,
+      "step": 21944
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.037450376038598e-06,
+      "loss": 0.0993,
+      "step": 21945
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.036585090423127e-06,
+      "loss": 0.0954,
+      "step": 21946
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.035719854114576e-06,
+      "loss": 0.1141,
+      "step": 21947
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.034854667121547e-06,
+      "loss": 0.1164,
+      "step": 21948
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.033989529452634e-06,
+      "loss": 0.1618,
+      "step": 21949
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.033124441116424e-06,
+      "loss": 0.086,
+      "step": 21950
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.032259402121516e-06,
+      "loss": 0.1005,
+      "step": 21951
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.031394412476509e-06,
+      "loss": 0.1451,
+      "step": 21952
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.030529472189989e-06,
+      "loss": 0.1016,
+      "step": 21953
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.029664581270546e-06,
+      "loss": 0.1179,
+      "step": 21954
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.02879973972678e-06,
+      "loss": 0.1186,
+      "step": 21955
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0279349475672745e-06,
+      "loss": 0.0961,
+      "step": 21956
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.027070204800626e-06,
+      "loss": 0.1202,
+      "step": 21957
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0262055114354245e-06,
+      "loss": 0.0794,
+      "step": 21958
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.025340867480255e-06,
+      "loss": 0.1142,
+      "step": 21959
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.024476272943708e-06,
+      "loss": 0.1266,
+      "step": 21960
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.023611727834381e-06,
+      "loss": 0.1078,
+      "step": 21961
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.022747232160854e-06,
+      "loss": 0.096,
+      "step": 21962
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.021882785931714e-06,
+      "loss": 0.1376,
+      "step": 21963
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.021018389155556e-06,
+      "loss": 0.0978,
+      "step": 21964
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.020154041840959e-06,
+      "loss": 0.1248,
+      "step": 21965
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.019289743996516e-06,
+      "loss": 0.1261,
+      "step": 21966
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.01842549563081e-06,
+      "loss": 0.0995,
+      "step": 21967
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.017561296752422e-06,
+      "loss": 0.1308,
+      "step": 21968
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.016697147369943e-06,
+      "loss": 0.077,
+      "step": 21969
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.015833047491961e-06,
+      "loss": 0.1252,
+      "step": 21970
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0149689971270485e-06,
+      "loss": 0.1045,
+      "step": 21971
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.014104996283801e-06,
+      "loss": 0.136,
+      "step": 21972
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.013241044970795e-06,
+      "loss": 0.1296,
+      "step": 21973
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.012377143196612e-06,
+      "loss": 0.1116,
+      "step": 21974
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.011513290969839e-06,
+      "loss": 0.0899,
+      "step": 21975
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.010649488299052e-06,
+      "loss": 0.0942,
+      "step": 21976
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.009785735192838e-06,
+      "loss": 0.095,
+      "step": 21977
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0089220316597705e-06,
+      "loss": 0.1508,
+      "step": 21978
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.008058377708438e-06,
+      "loss": 0.1013,
+      "step": 21979
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.007194773347413e-06,
+      "loss": 0.1146,
+      "step": 21980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.006331218585279e-06,
+      "loss": 0.0697,
+      "step": 21981
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.005467713430614e-06,
+      "loss": 0.1087,
+      "step": 21982
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.004604257891992e-06,
+      "loss": 0.1053,
+      "step": 21983
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.003740851977996e-06,
+      "loss": 0.0959,
+      "step": 21984
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.002877495697198e-06,
+      "loss": 0.0852,
+      "step": 21985
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.002014189058182e-06,
+      "loss": 0.0957,
+      "step": 21986
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.001150932069514e-06,
+      "loss": 0.11,
+      "step": 21987
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.000287724739781e-06,
+      "loss": 0.0868,
+      "step": 21988
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.999424567077547e-06,
+      "loss": 0.0773,
+      "step": 21989
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9985614590913965e-06,
+      "loss": 0.1068,
+      "step": 21990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9976984007898995e-06,
+      "loss": 0.1371,
+      "step": 21991
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.996835392181626e-06,
+      "loss": 0.0985,
+      "step": 21992
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9959724332751565e-06,
+      "loss": 0.0992,
+      "step": 21993
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.995109524079056e-06,
+      "loss": 0.1245,
+      "step": 21994
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9942466646019e-06,
+      "loss": 0.1081,
+      "step": 21995
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.993383854852265e-06,
+      "loss": 0.0917,
+      "step": 21996
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.992521094838718e-06,
+      "loss": 0.1178,
+      "step": 21997
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.991658384569826e-06,
+      "loss": 0.124,
+      "step": 21998
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.990795724054167e-06,
+      "loss": 0.0942,
+      "step": 21999
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.989933113300302e-06,
+      "loss": 0.1178,
+      "step": 22000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.989070552316811e-06,
+      "loss": 0.108,
+      "step": 22001
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9882080411122545e-06,
+      "loss": 0.089,
+      "step": 22002
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9873455796952e-06,
+      "loss": 0.1217,
+      "step": 22003
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.986483168074217e-06,
+      "loss": 0.1452,
+      "step": 22004
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.98562080625788e-06,
+      "loss": 0.122,
+      "step": 22005
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.984758494254747e-06,
+      "loss": 0.1122,
+      "step": 22006
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.983896232073385e-06,
+      "loss": 0.1296,
+      "step": 22007
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.983034019722365e-06,
+      "loss": 0.0809,
+      "step": 22008
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.982171857210245e-06,
+      "loss": 0.1134,
+      "step": 22009
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.981309744545594e-06,
+      "loss": 0.122,
+      "step": 22010
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.980447681736984e-06,
+      "loss": 0.0945,
+      "step": 22011
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.979585668792963e-06,
+      "loss": 0.0816,
+      "step": 22012
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9787237057221026e-06,
+      "loss": 0.1143,
+      "step": 22013
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.977861792532969e-06,
+      "loss": 0.1237,
+      "step": 22014
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.97699992923412e-06,
+      "loss": 0.1271,
+      "step": 22015
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.976138115834115e-06,
+      "loss": 0.1045,
+      "step": 22016
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.975276352341521e-06,
+      "loss": 0.1732,
+      "step": 22017
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.974414638764894e-06,
+      "loss": 0.1109,
+      "step": 22018
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.973552975112796e-06,
+      "loss": 0.1031,
+      "step": 22019
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.972691361393795e-06,
+      "loss": 0.1006,
+      "step": 22020
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.971829797616435e-06,
+      "loss": 0.1572,
+      "step": 22021
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.970968283789283e-06,
+      "loss": 0.1146,
+      "step": 22022
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9701068199209005e-06,
+      "loss": 0.0748,
+      "step": 22023
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.969245406019838e-06,
+      "loss": 0.1373,
+      "step": 22024
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.96838404209466e-06,
+      "loss": 0.1416,
+      "step": 22025
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.967522728153921e-06,
+      "loss": 0.1242,
+      "step": 22026
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.966661464206171e-06,
+      "loss": 0.1017,
+      "step": 22027
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.965800250259972e-06,
+      "loss": 0.1517,
+      "step": 22028
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.964939086323884e-06,
+      "loss": 0.1139,
+      "step": 22029
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.964077972406454e-06,
+      "loss": 0.1074,
+      "step": 22030
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.963216908516235e-06,
+      "loss": 0.1168,
+      "step": 22031
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.96235589466179e-06,
+      "loss": 0.0834,
+      "step": 22032
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.961494930851663e-06,
+      "loss": 0.1282,
+      "step": 22033
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.960634017094413e-06,
+      "loss": 0.0949,
+      "step": 22034
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9597731533985915e-06,
+      "loss": 0.1341,
+      "step": 22035
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.958912339772746e-06,
+      "loss": 0.1153,
+      "step": 22036
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.95805157622543e-06,
+      "loss": 0.1168,
+      "step": 22037
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9571908627652e-06,
+      "loss": 0.1273,
+      "step": 22038
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.956330199400598e-06,
+      "loss": 0.0853,
+      "step": 22039
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.95546958614018e-06,
+      "loss": 0.1083,
+      "step": 22040
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.954609022992495e-06,
+      "loss": 0.1077,
+      "step": 22041
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.953748509966085e-06,
+      "loss": 0.1376,
+      "step": 22042
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9528880470695075e-06,
+      "loss": 0.1263,
+      "step": 22043
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.952027634311306e-06,
+      "loss": 0.083,
+      "step": 22044
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.951167271700026e-06,
+      "loss": 0.1198,
+      "step": 22045
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.950306959244214e-06,
+      "loss": 0.1247,
+      "step": 22046
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.949446696952425e-06,
+      "loss": 0.0856,
+      "step": 22047
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.948586484833194e-06,
+      "loss": 0.0883,
+      "step": 22048
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.947726322895076e-06,
+      "loss": 0.0688,
+      "step": 22049
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.946866211146612e-06,
+      "loss": 0.0987,
+      "step": 22050
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.94600614959634e-06,
+      "loss": 0.0898,
+      "step": 22051
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9451461382528145e-06,
+      "loss": 0.0912,
+      "step": 22052
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9442861771245685e-06,
+      "loss": 0.1101,
+      "step": 22053
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.943426266220156e-06,
+      "loss": 0.1281,
+      "step": 22054
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.942566405548109e-06,
+      "loss": 0.1139,
+      "step": 22055
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9417065951169794e-06,
+      "loss": 0.1363,
+      "step": 22056
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.940846834935299e-06,
+      "loss": 0.1033,
+      "step": 22057
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.939987125011616e-06,
+      "loss": 0.1267,
+      "step": 22058
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.93912746535447e-06,
+      "loss": 0.0984,
+      "step": 22059
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.938267855972393e-06,
+      "loss": 0.1626,
+      "step": 22060
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9374082968739355e-06,
+      "loss": 0.1011,
+      "step": 22061
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.936548788067627e-06,
+      "loss": 0.0775,
+      "step": 22062
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.93568932956201e-06,
+      "loss": 0.1304,
+      "step": 22063
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.934829921365627e-06,
+      "loss": 0.1299,
+      "step": 22064
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.933970563487011e-06,
+      "loss": 0.0991,
+      "step": 22065
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.933111255934694e-06,
+      "loss": 0.0917,
+      "step": 22066
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9322519987172225e-06,
+      "loss": 0.1042,
+      "step": 22067
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.931392791843123e-06,
+      "loss": 0.1003,
+      "step": 22068
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.930533635320939e-06,
+      "loss": 0.1116,
+      "step": 22069
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.929674529159202e-06,
+      "loss": 0.1406,
+      "step": 22070
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9288154733664415e-06,
+      "loss": 0.1195,
+      "step": 22071
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9279564679511975e-06,
+      "loss": 0.082,
+      "step": 22072
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.927097512922005e-06,
+      "loss": 0.1403,
+      "step": 22073
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.926238608287392e-06,
+      "loss": 0.1067,
+      "step": 22074
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.925379754055891e-06,
+      "loss": 0.1347,
+      "step": 22075
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.924520950236039e-06,
+      "loss": 0.1104,
+      "step": 22076
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.923662196836361e-06,
+      "loss": 0.0953,
+      "step": 22077
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.922803493865394e-06,
+      "loss": 0.1212,
+      "step": 22078
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.921944841331665e-06,
+      "loss": 0.1231,
+      "step": 22079
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.921086239243702e-06,
+      "loss": 0.1444,
+      "step": 22080
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.920227687610035e-06,
+      "loss": 0.1286,
+      "step": 22081
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9193691864392e-06,
+      "loss": 0.112,
+      "step": 22082
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.91851073573972e-06,
+      "loss": 0.1444,
+      "step": 22083
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.917652335520119e-06,
+      "loss": 0.1194,
+      "step": 22084
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.916793985788931e-06,
+      "loss": 0.1287,
+      "step": 22085
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.915935686554678e-06,
+      "loss": 0.1107,
+      "step": 22086
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.915077437825888e-06,
+      "loss": 0.096,
+      "step": 22087
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.914219239611096e-06,
+      "loss": 0.0815,
+      "step": 22088
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.913361091918811e-06,
+      "loss": 0.1396,
+      "step": 22089
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9125029947575656e-06,
+      "loss": 0.094,
+      "step": 22090
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.911644948135888e-06,
+      "loss": 0.1539,
+      "step": 22091
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9107869520622965e-06,
+      "loss": 0.1022,
+      "step": 22092
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9099290065453196e-06,
+      "loss": 0.104,
+      "step": 22093
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.909071111593476e-06,
+      "loss": 0.0959,
+      "step": 22094
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.908213267215287e-06,
+      "loss": 0.1133,
+      "step": 22095
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.907355473419277e-06,
+      "loss": 0.1209,
+      "step": 22096
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.906497730213974e-06,
+      "loss": 0.1155,
+      "step": 22097
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.905640037607885e-06,
+      "loss": 0.1034,
+      "step": 22098
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9047823956095366e-06,
+      "loss": 0.077,
+      "step": 22099
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.903924804227454e-06,
+      "loss": 0.1298,
+      "step": 22100
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.903067263470148e-06,
+      "loss": 0.0876,
+      "step": 22101
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.902209773346146e-06,
+      "loss": 0.1052,
+      "step": 22102
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.901352333863961e-06,
+      "loss": 0.0969,
+      "step": 22103
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.900494945032109e-06,
+      "loss": 0.0877,
+      "step": 22104
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.89963760685911e-06,
+      "loss": 0.1099,
+      "step": 22105
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8987803193534845e-06,
+      "loss": 0.1159,
+      "step": 22106
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.897923082523746e-06,
+      "loss": 0.1055,
+      "step": 22107
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.897065896378406e-06,
+      "loss": 0.092,
+      "step": 22108
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.896208760925986e-06,
+      "loss": 0.1246,
+      "step": 22109
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8953516761749955e-06,
+      "loss": 0.0823,
+      "step": 22110
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8944946421339525e-06,
+      "loss": 0.0857,
+      "step": 22111
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.893637658811376e-06,
+      "loss": 0.1623,
+      "step": 22112
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.892780726215768e-06,
+      "loss": 0.1137,
+      "step": 22113
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.891923844355645e-06,
+      "loss": 0.0948,
+      "step": 22114
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.891067013239526e-06,
+      "loss": 0.1219,
+      "step": 22115
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8902102328759134e-06,
+      "loss": 0.0861,
+      "step": 22116
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.889353503273327e-06,
+      "loss": 0.1129,
+      "step": 22117
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.888496824440274e-06,
+      "loss": 0.1001,
+      "step": 22118
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.88764019638526e-06,
+      "loss": 0.106,
+      "step": 22119
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8867836191168e-06,
+      "loss": 0.1036,
+      "step": 22120
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.885927092643407e-06,
+      "loss": 0.0949,
+      "step": 22121
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.885070616973584e-06,
+      "loss": 0.1159,
+      "step": 22122
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.884214192115837e-06,
+      "loss": 0.0988,
+      "step": 22123
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8833578180786815e-06,
+      "loss": 0.1367,
+      "step": 22124
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.882501494870616e-06,
+      "loss": 0.1053,
+      "step": 22125
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.881645222500157e-06,
+      "loss": 0.1021,
+      "step": 22126
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.880789000975804e-06,
+      "loss": 0.0992,
+      "step": 22127
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.879932830306062e-06,
+      "loss": 0.1027,
+      "step": 22128
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.879076710499439e-06,
+      "loss": 0.1218,
+      "step": 22129
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8782206415644415e-06,
+      "loss": 0.1171,
+      "step": 22130
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.877364623509574e-06,
+      "loss": 0.1207,
+      "step": 22131
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.876508656343333e-06,
+      "loss": 0.098,
+      "step": 22132
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.87565274007423e-06,
+      "loss": 0.1057,
+      "step": 22133
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.874796874710761e-06,
+      "loss": 0.1179,
+      "step": 22134
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.873941060261437e-06,
+      "loss": 0.14,
+      "step": 22135
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.873085296734752e-06,
+      "loss": 0.0772,
+      "step": 22136
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.872229584139208e-06,
+      "loss": 0.1071,
+      "step": 22137
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.871373922483306e-06,
+      "loss": 0.1424,
+      "step": 22138
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.870518311775552e-06,
+      "loss": 0.0934,
+      "step": 22139
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.869662752024437e-06,
+      "loss": 0.1403,
+      "step": 22140
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.868807243238468e-06,
+      "loss": 0.0885,
+      "step": 22141
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.86795178542614e-06,
+      "loss": 0.1414,
+      "step": 22142
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8670963785959475e-06,
+      "loss": 0.1161,
+      "step": 22143
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.866241022756396e-06,
+      "loss": 0.1115,
+      "step": 22144
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.865385717915976e-06,
+      "loss": 0.1407,
+      "step": 22145
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.864530464083189e-06,
+      "loss": 0.0914,
+      "step": 22146
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.863675261266526e-06,
+      "loss": 0.1438,
+      "step": 22147
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.862820109474489e-06,
+      "loss": 0.0955,
+      "step": 22148
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.861965008715565e-06,
+      "loss": 0.1585,
+      "step": 22149
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8611099589982595e-06,
+      "loss": 0.1097,
+      "step": 22150
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8602549603310586e-06,
+      "loss": 0.093,
+      "step": 22151
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.859400012722455e-06,
+      "loss": 0.1168,
+      "step": 22152
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.858545116180948e-06,
+      "loss": 0.0756,
+      "step": 22153
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8576902707150234e-06,
+      "loss": 0.1102,
+      "step": 22154
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.856835476333181e-06,
+      "loss": 0.0927,
+      "step": 22155
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.855980733043903e-06,
+      "loss": 0.114,
+      "step": 22156
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8551260408556895e-06,
+      "loss": 0.1316,
+      "step": 22157
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.854271399777024e-06,
+      "loss": 0.1061,
+      "step": 22158
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.853416809816404e-06,
+      "loss": 0.1405,
+      "step": 22159
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.852562270982314e-06,
+      "loss": 0.1181,
+      "step": 22160
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.851707783283239e-06,
+      "loss": 0.1105,
+      "step": 22161
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.850853346727679e-06,
+      "loss": 0.1138,
+      "step": 22162
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.84999896132411e-06,
+      "loss": 0.0828,
+      "step": 22163
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.849144627081024e-06,
+      "loss": 0.1357,
+      "step": 22164
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.848290344006917e-06,
+      "loss": 0.1154,
+      "step": 22165
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.847436112110261e-06,
+      "loss": 0.1112,
+      "step": 22166
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.846581931399548e-06,
+      "loss": 0.1032,
+      "step": 22167
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.845727801883268e-06,
+      "loss": 0.0936,
+      "step": 22168
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.844873723569897e-06,
+      "loss": 0.1387,
+      "step": 22169
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.84401969646793e-06,
+      "loss": 0.1021,
+      "step": 22170
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.843165720585844e-06,
+      "loss": 0.145,
+      "step": 22171
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.842311795932121e-06,
+      "loss": 0.1114,
+      "step": 22172
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.841457922515246e-06,
+      "loss": 0.107,
+      "step": 22173
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.840604100343706e-06,
+      "loss": 0.0892,
+      "step": 22174
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.839750329425979e-06,
+      "loss": 0.1176,
+      "step": 22175
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.838896609770544e-06,
+      "loss": 0.1246,
+      "step": 22176
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.838042941385888e-06,
+      "loss": 0.1092,
+      "step": 22177
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.837189324280484e-06,
+      "loss": 0.1163,
+      "step": 22178
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8363357584628155e-06,
+      "loss": 0.1081,
+      "step": 22179
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.835482243941369e-06,
+      "loss": 0.0902,
+      "step": 22180
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8346287807246105e-06,
+      "loss": 0.1471,
+      "step": 22181
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.833775368821025e-06,
+      "loss": 0.1223,
+      "step": 22182
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.832922008239091e-06,
+      "loss": 0.1207,
+      "step": 22183
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.832068698987287e-06,
+      "loss": 0.0834,
+      "step": 22184
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8312154410740815e-06,
+      "loss": 0.113,
+      "step": 22185
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.830362234507963e-06,
+      "loss": 0.0999,
+      "step": 22186
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8295090792973955e-06,
+      "loss": 0.1397,
+      "step": 22187
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.828655975450861e-06,
+      "loss": 0.0996,
+      "step": 22188
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.827802922976841e-06,
+      "loss": 0.0973,
+      "step": 22189
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.826949921883794e-06,
+      "loss": 0.118,
+      "step": 22190
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.826096972180201e-06,
+      "loss": 0.0676,
+      "step": 22191
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.82524407387454e-06,
+      "loss": 0.093,
+      "step": 22192
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.824391226975277e-06,
+      "loss": 0.1548,
+      "step": 22193
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.823538431490891e-06,
+      "loss": 0.113,
+      "step": 22194
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.822685687429848e-06,
+      "loss": 0.1083,
+      "step": 22195
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.821832994800618e-06,
+      "loss": 0.1143,
+      "step": 22196
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.820980353611675e-06,
+      "loss": 0.0937,
+      "step": 22197
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8201277638714925e-06,
+      "loss": 0.099,
+      "step": 22198
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.819275225588538e-06,
+      "loss": 0.0942,
+      "step": 22199
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.818422738771273e-06,
+      "loss": 0.0961,
+      "step": 22200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.817570303428178e-06,
+      "loss": 0.1019,
+      "step": 22201
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.816717919567711e-06,
+      "loss": 0.1196,
+      "step": 22202
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8158655871983486e-06,
+      "loss": 0.119,
+      "step": 22203
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.815013306328553e-06,
+      "loss": 0.1349,
+      "step": 22204
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8141610769667876e-06,
+      "loss": 0.1093,
+      "step": 22205
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.813308899121524e-06,
+      "loss": 0.1311,
+      "step": 22206
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.812456772801228e-06,
+      "loss": 0.1246,
+      "step": 22207
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8116046980143594e-06,
+      "loss": 0.0755,
+      "step": 22208
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.81075267476939e-06,
+      "loss": 0.1354,
+      "step": 22209
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.80990070307478e-06,
+      "loss": 0.1725,
+      "step": 22210
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.809048782938989e-06,
+      "loss": 0.1182,
+      "step": 22211
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8081969143704876e-06,
+      "loss": 0.1021,
+      "step": 22212
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.807345097377736e-06,
+      "loss": 0.1118,
+      "step": 22213
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.80649333196919e-06,
+      "loss": 0.1062,
+      "step": 22214
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.805641618153316e-06,
+      "loss": 0.1056,
+      "step": 22215
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.804789955938579e-06,
+      "loss": 0.0781,
+      "step": 22216
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.803938345333432e-06,
+      "loss": 0.0996,
+      "step": 22217
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8030867863463415e-06,
+      "loss": 0.0968,
+      "step": 22218
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.802235278985764e-06,
+      "loss": 0.105,
+      "step": 22219
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.801383823260155e-06,
+      "loss": 0.1194,
+      "step": 22220
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.80053241917798e-06,
+      "loss": 0.117,
+      "step": 22221
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.79968106674769e-06,
+      "loss": 0.1454,
+      "step": 22222
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.798829765977748e-06,
+      "loss": 0.1119,
+      "step": 22223
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7979785168766055e-06,
+      "loss": 0.113,
+      "step": 22224
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7971273194527254e-06,
+      "loss": 0.0893,
+      "step": 22225
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.796276173714557e-06,
+      "loss": 0.0885,
+      "step": 22226
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.795425079670561e-06,
+      "loss": 0.1149,
+      "step": 22227
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.79457403732919e-06,
+      "loss": 0.1202,
+      "step": 22228
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.793723046698896e-06,
+      "loss": 0.1201,
+      "step": 22229
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.792872107788138e-06,
+      "loss": 0.1251,
+      "step": 22230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.792021220605362e-06,
+      "loss": 0.1246,
+      "step": 22231
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.791170385159025e-06,
+      "loss": 0.0777,
+      "step": 22232
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.790319601457584e-06,
+      "loss": 0.097,
+      "step": 22233
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7894688695094845e-06,
+      "loss": 0.1085,
+      "step": 22234
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.788618189323176e-06,
+      "loss": 0.0644,
+      "step": 22235
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7877675609071174e-06,
+      "loss": 0.1233,
+      "step": 22236
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7869169842697525e-06,
+      "loss": 0.1583,
+      "step": 22237
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.78606645941953e-06,
+      "loss": 0.113,
+      "step": 22238
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.785215986364904e-06,
+      "loss": 0.1064,
+      "step": 22239
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.784365565114319e-06,
+      "loss": 0.1029,
+      "step": 22240
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.783515195676223e-06,
+      "loss": 0.0823,
+      "step": 22241
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.782664878059069e-06,
+      "loss": 0.092,
+      "step": 22242
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.781814612271302e-06,
+      "loss": 0.1261,
+      "step": 22243
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.780964398321363e-06,
+      "loss": 0.0846,
+      "step": 22244
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.780114236217706e-06,
+      "loss": 0.0936,
+      "step": 22245
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.779264125968769e-06,
+      "loss": 0.092,
+      "step": 22246
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7784140675830045e-06,
+      "loss": 0.1135,
+      "step": 22247
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.777564061068853e-06,
+      "loss": 0.1137,
+      "step": 22248
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.776714106434756e-06,
+      "loss": 0.1419,
+      "step": 22249
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.775864203689161e-06,
+      "loss": 0.1424,
+      "step": 22250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.775014352840512e-06,
+      "loss": 0.0965,
+      "step": 22251
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.774164553897251e-06,
+      "loss": 0.1174,
+      "step": 22252
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7733148068678134e-06,
+      "loss": 0.1559,
+      "step": 22253
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.77246511176065e-06,
+      "loss": 0.1018,
+      "step": 22254
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.771615468584194e-06,
+      "loss": 0.1085,
+      "step": 22255
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.770765877346889e-06,
+      "loss": 0.1726,
+      "step": 22256
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.769916338057182e-06,
+      "loss": 0.1028,
+      "step": 22257
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7690668507234996e-06,
+      "loss": 0.1026,
+      "step": 22258
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.768217415354286e-06,
+      "loss": 0.102,
+      "step": 22259
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.767368031957983e-06,
+      "loss": 0.1017,
+      "step": 22260
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.766518700543022e-06,
+      "loss": 0.1388,
+      "step": 22261
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.765669421117849e-06,
+      "loss": 0.1206,
+      "step": 22262
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.764820193690895e-06,
+      "loss": 0.1167,
+      "step": 22263
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7639710182705925e-06,
+      "loss": 0.1044,
+      "step": 22264
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.763121894865382e-06,
+      "loss": 0.1314,
+      "step": 22265
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.762272823483707e-06,
+      "loss": 0.1326,
+      "step": 22266
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.761423804133986e-06,
+      "loss": 0.0964,
+      "step": 22267
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.760574836824661e-06,
+      "loss": 0.1098,
+      "step": 22268
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.759725921564169e-06,
+      "loss": 0.1017,
+      "step": 22269
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.758877058360936e-06,
+      "loss": 0.1448,
+      "step": 22270
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.758028247223402e-06,
+      "loss": 0.1055,
+      "step": 22271
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.757179488159996e-06,
+      "loss": 0.1288,
+      "step": 22272
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.756330781179146e-06,
+      "loss": 0.1095,
+      "step": 22273
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.755482126289287e-06,
+      "loss": 0.1301,
+      "step": 22274
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7546335234988515e-06,
+      "loss": 0.1211,
+      "step": 22275
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.753784972816268e-06,
+      "loss": 0.0968,
+      "step": 22276
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.752936474249961e-06,
+      "loss": 0.1005,
+      "step": 22277
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.752088027808368e-06,
+      "loss": 0.0914,
+      "step": 22278
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.751239633499909e-06,
+      "loss": 0.0968,
+      "step": 22279
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.750391291333016e-06,
+      "loss": 0.1112,
+      "step": 22280
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.749543001316125e-06,
+      "loss": 0.124,
+      "step": 22281
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7486947634576465e-06,
+      "loss": 0.1304,
+      "step": 22282
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7478465777660155e-06,
+      "loss": 0.0919,
+      "step": 22283
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.746998444249662e-06,
+      "loss": 0.1234,
+      "step": 22284
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.746150362917003e-06,
+      "loss": 0.1048,
+      "step": 22285
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.745302333776472e-06,
+      "loss": 0.1077,
+      "step": 22286
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.744454356836487e-06,
+      "loss": 0.0916,
+      "step": 22287
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.743606432105471e-06,
+      "loss": 0.1382,
+      "step": 22288
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7427585595918506e-06,
+      "loss": 0.0945,
+      "step": 22289
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.74191073930405e-06,
+      "loss": 0.136,
+      "step": 22290
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.74106297125049e-06,
+      "loss": 0.1003,
+      "step": 22291
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.74021525543959e-06,
+      "loss": 0.1253,
+      "step": 22292
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.739367591879776e-06,
+      "loss": 0.0815,
+      "step": 22293
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7385199805794625e-06,
+      "loss": 0.0804,
+      "step": 22294
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.737672421547076e-06,
+      "loss": 0.1091,
+      "step": 22295
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.736824914791035e-06,
+      "loss": 0.0777,
+      "step": 22296
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.735977460319752e-06,
+      "loss": 0.1048,
+      "step": 22297
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.735130058141651e-06,
+      "loss": 0.093,
+      "step": 22298
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7342827082651544e-06,
+      "loss": 0.1131,
+      "step": 22299
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.733435410698676e-06,
+      "loss": 0.1023,
+      "step": 22300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.732588165450627e-06,
+      "loss": 0.1117,
+      "step": 22301
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.731740972529434e-06,
+      "loss": 0.0981,
+      "step": 22302
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.730893831943504e-06,
+      "loss": 0.114,
+      "step": 22303
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7300467437012606e-06,
+      "loss": 0.0995,
+      "step": 22304
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7291997078111165e-06,
+      "loss": 0.0864,
+      "step": 22305
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7283527242814796e-06,
+      "loss": 0.0969,
+      "step": 22306
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.72750579312077e-06,
+      "loss": 0.1198,
+      "step": 22307
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.726658914337404e-06,
+      "loss": 0.1192,
+      "step": 22308
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.725812087939787e-06,
+      "loss": 0.1006,
+      "step": 22309
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.724965313936339e-06,
+      "loss": 0.0607,
+      "step": 22310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.724118592335469e-06,
+      "loss": 0.1116,
+      "step": 22311
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.723271923145584e-06,
+      "loss": 0.0993,
+      "step": 22312
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.722425306375101e-06,
+      "loss": 0.1163,
+      "step": 22313
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.721578742032425e-06,
+      "loss": 0.1215,
+      "step": 22314
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.720732230125973e-06,
+      "loss": 0.1103,
+      "step": 22315
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.719885770664145e-06,
+      "loss": 0.1487,
+      "step": 22316
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.71903936365536e-06,
+      "loss": 0.109,
+      "step": 22317
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7181930091080165e-06,
+      "loss": 0.1108,
+      "step": 22318
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.717346707030531e-06,
+      "loss": 0.0856,
+      "step": 22319
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.716500457431306e-06,
+      "loss": 0.0936,
+      "step": 22320
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7156542603187445e-06,
+      "loss": 0.0948,
+      "step": 22321
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.714808115701263e-06,
+      "loss": 0.1113,
+      "step": 22322
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.713962023587255e-06,
+      "loss": 0.1148,
+      "step": 22323
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.713115983985137e-06,
+      "loss": 0.1219,
+      "step": 22324
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.712269996903307e-06,
+      "loss": 0.0948,
+      "step": 22325
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.711424062350168e-06,
+      "loss": 0.1113,
+      "step": 22326
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.710578180334125e-06,
+      "loss": 0.104,
+      "step": 22327
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.709732350863587e-06,
+      "loss": 0.1111,
+      "step": 22328
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.708886573946952e-06,
+      "loss": 0.1068,
+      "step": 22329
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7080408495926155e-06,
+      "loss": 0.1256,
+      "step": 22330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.707195177808991e-06,
+      "loss": 0.1141,
+      "step": 22331
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7063495586044695e-06,
+      "loss": 0.1348,
+      "step": 22332
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.705503991987455e-06,
+      "loss": 0.1126,
+      "step": 22333
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.704658477966355e-06,
+      "loss": 0.0942,
+      "step": 22334
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.703813016549555e-06,
+      "loss": 0.1294,
+      "step": 22335
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7029676077454614e-06,
+      "loss": 0.0982,
+      "step": 22336
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7021222515624745e-06,
+      "loss": 0.11,
+      "step": 22337
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.701276948008987e-06,
+      "loss": 0.1073,
+      "step": 22338
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7004316970934015e-06,
+      "loss": 0.1163,
+      "step": 22339
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.699586498824112e-06,
+      "loss": 0.0691,
+      "step": 22340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.698741353209512e-06,
+      "loss": 0.0923,
+      "step": 22341
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.697896260257998e-06,
+      "loss": 0.1266,
+      "step": 22342
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.697051219977973e-06,
+      "loss": 0.1401,
+      "step": 22343
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.696206232377825e-06,
+      "loss": 0.0774,
+      "step": 22344
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.695361297465946e-06,
+      "loss": 0.0963,
+      "step": 22345
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.694516415250735e-06,
+      "loss": 0.1238,
+      "step": 22346
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.69367158574058e-06,
+      "loss": 0.0815,
+      "step": 22347
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.692826808943877e-06,
+      "loss": 0.0932,
+      "step": 22348
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.691982084869024e-06,
+      "loss": 0.1057,
+      "step": 22349
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.691137413524399e-06,
+      "loss": 0.0954,
+      "step": 22350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.690292794918401e-06,
+      "loss": 0.1286,
+      "step": 22351
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.689448229059423e-06,
+      "loss": 0.0794,
+      "step": 22352
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.688603715955851e-06,
+      "loss": 0.0993,
+      "step": 22353
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.68775925561607e-06,
+      "loss": 0.0898,
+      "step": 22354
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6869148480484795e-06,
+      "loss": 0.1064,
+      "step": 22355
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.686070493261459e-06,
+      "loss": 0.1034,
+      "step": 22356
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.685226191263398e-06,
+      "loss": 0.1056,
+      "step": 22357
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.684381942062694e-06,
+      "loss": 0.0861,
+      "step": 22358
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.683537745667718e-06,
+      "loss": 0.0976,
+      "step": 22359
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6826936020868644e-06,
+      "loss": 0.0958,
+      "step": 22360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.681849511328521e-06,
+      "loss": 0.1176,
+      "step": 22361
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.681005473401067e-06,
+      "loss": 0.0954,
+      "step": 22362
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6801614883128946e-06,
+      "loss": 0.1002,
+      "step": 22363
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.679317556072383e-06,
+      "loss": 0.1359,
+      "step": 22364
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6784736766879136e-06,
+      "loss": 0.1027,
+      "step": 22365
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.677629850167873e-06,
+      "loss": 0.1166,
+      "step": 22366
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.676786076520647e-06,
+      "loss": 0.1102,
+      "step": 22367
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.675942355754615e-06,
+      "loss": 0.1461,
+      "step": 22368
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.675098687878153e-06,
+      "loss": 0.1065,
+      "step": 22369
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6742550728996514e-06,
+      "loss": 0.1021,
+      "step": 22370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.673411510827484e-06,
+      "loss": 0.1072,
+      "step": 22371
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.672568001670035e-06,
+      "loss": 0.1257,
+      "step": 22372
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.671724545435682e-06,
+      "loss": 0.0747,
+      "step": 22373
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6708811421328e-06,
+      "loss": 0.0957,
+      "step": 22374
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.670037791769772e-06,
+      "loss": 0.1111,
+      "step": 22375
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6691944943549795e-06,
+      "loss": 0.1633,
+      "step": 22376
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.668351249896791e-06,
+      "loss": 0.0977,
+      "step": 22377
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6675080584035926e-06,
+      "loss": 0.1001,
+      "step": 22378
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.666664919883755e-06,
+      "loss": 0.0892,
+      "step": 22379
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.665821834345652e-06,
+      "loss": 0.1283,
+      "step": 22380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.664978801797665e-06,
+      "loss": 0.0854,
+      "step": 22381
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.664135822248166e-06,
+      "loss": 0.1243,
+      "step": 22382
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.663292895705526e-06,
+      "loss": 0.1051,
+      "step": 22383
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.66245002217812e-06,
+      "loss": 0.1571,
+      "step": 22384
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.661607201674328e-06,
+      "loss": 0.1121,
+      "step": 22385
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.660764434202512e-06,
+      "loss": 0.1135,
+      "step": 22386
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.659921719771055e-06,
+      "loss": 0.1527,
+      "step": 22387
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.659079058388321e-06,
+      "loss": 0.1103,
+      "step": 22388
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.658236450062681e-06,
+      "loss": 0.0699,
+      "step": 22389
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.65739389480251e-06,
+      "loss": 0.1147,
+      "step": 22390
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6565513926161736e-06,
+      "loss": 0.0953,
+      "step": 22391
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.655708943512045e-06,
+      "loss": 0.1358,
+      "step": 22392
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.65486654749849e-06,
+      "loss": 0.0948,
+      "step": 22393
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.65402420458388e-06,
+      "loss": 0.1235,
+      "step": 22394
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.65318191477658e-06,
+      "loss": 0.124,
+      "step": 22395
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6523396780849604e-06,
+      "loss": 0.1173,
+      "step": 22396
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.651497494517387e-06,
+      "loss": 0.0955,
+      "step": 22397
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.650655364082223e-06,
+      "loss": 0.0949,
+      "step": 22398
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.64981328678784e-06,
+      "loss": 0.0899,
+      "step": 22399
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.648971262642596e-06,
+      "loss": 0.1045,
+      "step": 22400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.648129291654859e-06,
+      "loss": 0.115,
+      "step": 22401
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.647287373832998e-06,
+      "loss": 0.1249,
+      "step": 22402
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.646445509185373e-06,
+      "loss": 0.1038,
+      "step": 22403
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6456036977203424e-06,
+      "loss": 0.1222,
+      "step": 22404
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.644761939446277e-06,
+      "loss": 0.1071,
+      "step": 22405
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.643920234371536e-06,
+      "loss": 0.1089,
+      "step": 22406
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.643078582504475e-06,
+      "loss": 0.0881,
+      "step": 22407
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.642236983853464e-06,
+      "loss": 0.1414,
+      "step": 22408
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.641395438426857e-06,
+      "loss": 0.1074,
+      "step": 22409
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.640553946233015e-06,
+      "loss": 0.1126,
+      "step": 22410
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.639712507280303e-06,
+      "loss": 0.1101,
+      "step": 22411
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.638871121577075e-06,
+      "loss": 0.0885,
+      "step": 22412
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.638029789131688e-06,
+      "loss": 0.1301,
+      "step": 22413
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.637188509952505e-06,
+      "loss": 0.0895,
+      "step": 22414
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.636347284047878e-06,
+      "loss": 0.1092,
+      "step": 22415
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.635506111426168e-06,
+      "loss": 0.0684,
+      "step": 22416
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.634664992095731e-06,
+      "loss": 0.0937,
+      "step": 22417
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.633823926064917e-06,
+      "loss": 0.1599,
+      "step": 22418
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.632982913342084e-06,
+      "loss": 0.1015,
+      "step": 22419
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.632141953935594e-06,
+      "loss": 0.0964,
+      "step": 22420
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.631301047853795e-06,
+      "loss": 0.0802,
+      "step": 22421
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.630460195105035e-06,
+      "loss": 0.087,
+      "step": 22422
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.629619395697679e-06,
+      "loss": 0.107,
+      "step": 22423
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6287786496400685e-06,
+      "loss": 0.1146,
+      "step": 22424
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6279379569405626e-06,
+      "loss": 0.1546,
+      "step": 22425
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.627097317607516e-06,
+      "loss": 0.105,
+      "step": 22426
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.626256731649268e-06,
+      "loss": 0.1127,
+      "step": 22427
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.625416199074175e-06,
+      "loss": 0.1005,
+      "step": 22428
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.624575719890592e-06,
+      "loss": 0.0932,
+      "step": 22429
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.623735294106859e-06,
+      "loss": 0.092,
+      "step": 22430
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.622894921731335e-06,
+      "loss": 0.1328,
+      "step": 22431
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6220546027723615e-06,
+      "loss": 0.1198,
+      "step": 22432
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.621214337238284e-06,
+      "loss": 0.1031,
+      "step": 22433
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.620374125137453e-06,
+      "loss": 0.0684,
+      "step": 22434
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.619533966478225e-06,
+      "loss": 0.0959,
+      "step": 22435
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.618693861268928e-06,
+      "loss": 0.0898,
+      "step": 22436
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617853809517917e-06,
+      "loss": 0.0999,
+      "step": 22437
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617013811233541e-06,
+      "loss": 0.1115,
+      "step": 22438
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.616173866424136e-06,
+      "loss": 0.1214,
+      "step": 22439
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.615333975098054e-06,
+      "loss": 0.1077,
+      "step": 22440
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6144941372636355e-06,
+      "loss": 0.113,
+      "step": 22441
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.613654352929219e-06,
+      "loss": 0.0895,
+      "step": 22442
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.612814622103151e-06,
+      "loss": 0.0942,
+      "step": 22443
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.611974944793776e-06,
+      "loss": 0.0895,
+      "step": 22444
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6111353210094346e-06,
+      "loss": 0.0888,
+      "step": 22445
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.610295750758461e-06,
+      "loss": 0.0881,
+      "step": 22446
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.609456234049206e-06,
+      "loss": 0.0752,
+      "step": 22447
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.608616770889998e-06,
+      "loss": 0.1702,
+      "step": 22448
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.607777361289183e-06,
+      "loss": 0.097,
+      "step": 22449
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.606938005255106e-06,
+      "loss": 0.1022,
+      "step": 22450
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.60609870279609e-06,
+      "loss": 0.0844,
+      "step": 22451
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.605259453920482e-06,
+      "loss": 0.1025,
+      "step": 22452
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.604420258636621e-06,
+      "loss": 0.0873,
+      "step": 22453
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6035811169528375e-06,
+      "loss": 0.1379,
+      "step": 22454
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.602742028877475e-06,
+      "loss": 0.122,
+      "step": 22455
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.601902994418863e-06,
+      "loss": 0.1414,
+      "step": 22456
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.601064013585336e-06,
+      "loss": 0.0831,
+      "step": 22457
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.600225086385231e-06,
+      "loss": 0.0965,
+      "step": 22458
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.599386212826884e-06,
+      "loss": 0.0835,
+      "step": 22459
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.598547392918628e-06,
+      "loss": 0.1221,
+      "step": 22460
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.597708626668789e-06,
+      "loss": 0.1136,
+      "step": 22461
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.596869914085708e-06,
+      "loss": 0.0996,
+      "step": 22462
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.59603125517771e-06,
+      "loss": 0.1213,
+      "step": 22463
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.595192649953133e-06,
+      "loss": 0.106,
+      "step": 22464
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.594354098420305e-06,
+      "loss": 0.108,
+      "step": 22465
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.593515600587551e-06,
+      "loss": 0.1128,
+      "step": 22466
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.592677156463206e-06,
+      "loss": 0.073,
+      "step": 22467
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591838766055602e-06,
+      "loss": 0.0945,
+      "step": 22468
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591000429373063e-06,
+      "loss": 0.0725,
+      "step": 22469
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.590162146423916e-06,
+      "loss": 0.1164,
+      "step": 22470
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.589323917216494e-06,
+      "loss": 0.1655,
+      "step": 22471
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.588485741759117e-06,
+      "loss": 0.1438,
+      "step": 22472
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5876476200601185e-06,
+      "loss": 0.0967,
+      "step": 22473
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.586809552127821e-06,
+      "loss": 0.0965,
+      "step": 22474
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5859715379705475e-06,
+      "loss": 0.0694,
+      "step": 22475
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.585133577596624e-06,
+      "loss": 0.1026,
+      "step": 22476
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.584295671014381e-06,
+      "loss": 0.0901,
+      "step": 22477
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.583457818232134e-06,
+      "loss": 0.0798,
+      "step": 22478
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5826200192582145e-06,
+      "loss": 0.0994,
+      "step": 22479
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.58178227410094e-06,
+      "loss": 0.1326,
+      "step": 22480
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.58094458276863e-06,
+      "loss": 0.0803,
+      "step": 22481
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.580106945269614e-06,
+      "loss": 0.1074,
+      "step": 22482
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.579269361612203e-06,
+      "loss": 0.0972,
+      "step": 22483
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.57843183180473e-06,
+      "loss": 0.0728,
+      "step": 22484
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.577594355855506e-06,
+      "loss": 0.1375,
+      "step": 22485
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.57675693377285e-06,
+      "loss": 0.1006,
+      "step": 22486
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.575919565565085e-06,
+      "loss": 0.1411,
+      "step": 22487
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.575082251240532e-06,
+      "loss": 0.1301,
+      "step": 22488
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.574244990807505e-06,
+      "loss": 0.0992,
+      "step": 22489
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.573407784274317e-06,
+      "loss": 0.1013,
+      "step": 22490
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.572570631649296e-06,
+      "loss": 0.0762,
+      "step": 22491
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.571733532940746e-06,
+      "loss": 0.151,
+      "step": 22492
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.570896488156995e-06,
+      "loss": 0.1765,
+      "step": 22493
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.57005949730635e-06,
+      "loss": 0.1267,
+      "step": 22494
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.569222560397126e-06,
+      "loss": 0.0902,
+      "step": 22495
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.568385677437638e-06,
+      "loss": 0.0636,
+      "step": 22496
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.567548848436205e-06,
+      "loss": 0.071,
+      "step": 22497
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.566712073401135e-06,
+      "loss": 0.0912,
+      "step": 22498
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.565875352340739e-06,
+      "loss": 0.0964,
+      "step": 22499
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.565038685263334e-06,
+      "loss": 0.1345,
+      "step": 22500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5642020721772266e-06,
+      "loss": 0.1479,
+      "step": 22501
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.56336551309073e-06,
+      "loss": 0.0814,
+      "step": 22502
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.562529008012161e-06,
+      "loss": 0.107,
+      "step": 22503
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.561692556949816e-06,
+      "loss": 0.1215,
+      "step": 22504
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.560856159912013e-06,
+      "loss": 0.1038,
+      "step": 22505
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.560019816907062e-06,
+      "loss": 0.0988,
+      "step": 22506
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5591835279432664e-06,
+      "loss": 0.1032,
+      "step": 22507
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5583472930289384e-06,
+      "loss": 0.102,
+      "step": 22508
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5575111121723835e-06,
+      "loss": 0.0922,
+      "step": 22509
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.556674985381906e-06,
+      "loss": 0.1254,
+      "step": 22510
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.555838912665812e-06,
+      "loss": 0.0886,
+      "step": 22511
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.555002894032414e-06,
+      "loss": 0.1146,
+      "step": 22512
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.554166929490013e-06,
+      "loss": 0.0997,
+      "step": 22513
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.553331019046908e-06,
+      "loss": 0.1015,
+      "step": 22514
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.552495162711412e-06,
+      "loss": 0.1306,
+      "step": 22515
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.55165936049182e-06,
+      "loss": 0.1234,
+      "step": 22516
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.55082361239644e-06,
+      "loss": 0.0959,
+      "step": 22517
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.54998791843358e-06,
+      "loss": 0.094,
+      "step": 22518
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.549152278611529e-06,
+      "loss": 0.1282,
+      "step": 22519
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.548316692938595e-06,
+      "loss": 0.1267,
+      "step": 22520
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.547481161423081e-06,
+      "loss": 0.1466,
+      "step": 22521
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.546645684073285e-06,
+      "loss": 0.1246,
+      "step": 22522
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.545810260897503e-06,
+      "loss": 0.1123,
+      "step": 22523
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.544974891904042e-06,
+      "loss": 0.0999,
+      "step": 22524
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.544139577101191e-06,
+      "loss": 0.1158,
+      "step": 22525
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.543304316497254e-06,
+      "loss": 0.105,
+      "step": 22526
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.542469110100536e-06,
+      "loss": 0.1053,
+      "step": 22527
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.541633957919317e-06,
+      "loss": 0.1183,
+      "step": 22528
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5407988599619025e-06,
+      "loss": 0.1042,
+      "step": 22529
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.539963816236592e-06,
+      "loss": 0.0896,
+      "step": 22530
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5391288267516744e-06,
+      "loss": 0.0833,
+      "step": 22531
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.538293891515449e-06,
+      "loss": 0.1086,
+      "step": 22532
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.53745901053621e-06,
+      "loss": 0.0913,
+      "step": 22533
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5366241838222454e-06,
+      "loss": 0.0889,
+      "step": 22534
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.535789411381852e-06,
+      "loss": 0.1016,
+      "step": 22535
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.534954693223328e-06,
+      "loss": 0.102,
+      "step": 22536
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.534120029354959e-06,
+      "loss": 0.0971,
+      "step": 22537
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.533285419785035e-06,
+      "loss": 0.087,
+      "step": 22538
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.532450864521854e-06,
+      "loss": 0.1072,
+      "step": 22539
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5316163635737e-06,
+      "loss": 0.0981,
+      "step": 22540
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.530781916948869e-06,
+      "loss": 0.1061,
+      "step": 22541
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.529947524655647e-06,
+      "loss": 0.1427,
+      "step": 22542
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.529113186702319e-06,
+      "loss": 0.1288,
+      "step": 22543
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.528278903097177e-06,
+      "loss": 0.1085,
+      "step": 22544
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.527444673848514e-06,
+      "loss": 0.1012,
+      "step": 22545
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.52661049896461e-06,
+      "loss": 0.1029,
+      "step": 22546
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.525776378453757e-06,
+      "loss": 0.0938,
+      "step": 22547
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.524942312324239e-06,
+      "loss": 0.1162,
+      "step": 22548
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.524108300584336e-06,
+      "loss": 0.0982,
+      "step": 22549
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.523274343242343e-06,
+      "loss": 0.1168,
+      "step": 22550
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.522440440306541e-06,
+      "loss": 0.1687,
+      "step": 22551
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.521606591785208e-06,
+      "loss": 0.1169,
+      "step": 22552
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.520772797686633e-06,
+      "loss": 0.1523,
+      "step": 22553
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.519939058019103e-06,
+      "loss": 0.1082,
+      "step": 22554
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.519105372790892e-06,
+      "loss": 0.096,
+      "step": 22555
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5182717420102895e-06,
+      "loss": 0.1324,
+      "step": 22556
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.517438165685573e-06,
+      "loss": 0.1196,
+      "step": 22557
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.51660464382502e-06,
+      "loss": 0.131,
+      "step": 22558
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.515771176436917e-06,
+      "loss": 0.0657,
+      "step": 22559
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.514937763529539e-06,
+      "loss": 0.156,
+      "step": 22560
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.514104405111169e-06,
+      "loss": 0.0973,
+      "step": 22561
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.513271101190082e-06,
+      "loss": 0.1301,
+      "step": 22562
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.51243785177456e-06,
+      "loss": 0.0976,
+      "step": 22563
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.511604656872877e-06,
+      "loss": 0.1211,
+      "step": 22564
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.510771516493313e-06,
+      "loss": 0.1039,
+      "step": 22565
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.509938430644144e-06,
+      "loss": 0.0885,
+      "step": 22566
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.509105399333641e-06,
+      "loss": 0.0987,
+      "step": 22567
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.508272422570087e-06,
+      "loss": 0.1194,
+      "step": 22568
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.50743950036175e-06,
+      "loss": 0.0932,
+      "step": 22569
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.506606632716906e-06,
+      "loss": 0.0992,
+      "step": 22570
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.505773819643836e-06,
+      "loss": 0.1085,
+      "step": 22571
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.504941061150807e-06,
+      "loss": 0.1232,
+      "step": 22572
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.504108357246087e-06,
+      "loss": 0.1391,
+      "step": 22573
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.503275707937958e-06,
+      "loss": 0.101,
+      "step": 22574
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.502443113234688e-06,
+      "loss": 0.0845,
+      "step": 22575
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5016105731445415e-06,
+      "loss": 0.1349,
+      "step": 22576
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.500778087675799e-06,
+      "loss": 0.0952,
+      "step": 22577
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4999456568367215e-06,
+      "loss": 0.1274,
+      "step": 22578
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4991132806355845e-06,
+      "loss": 0.1141,
+      "step": 22579
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.498280959080658e-06,
+      "loss": 0.1344,
+      "step": 22580
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4974486921802075e-06,
+      "loss": 0.0866,
+      "step": 22581
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.496616479942497e-06,
+      "loss": 0.0843,
+      "step": 22582
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.495784322375801e-06,
+      "loss": 0.1128,
+      "step": 22583
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.49495221948838e-06,
+      "loss": 0.1084,
+      "step": 22584
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4941201712885065e-06,
+      "loss": 0.093,
+      "step": 22585
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.493288177784443e-06,
+      "loss": 0.126,
+      "step": 22586
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.492456238984451e-06,
+      "loss": 0.1064,
+      "step": 22587
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4916243548967965e-06,
+      "loss": 0.1274,
+      "step": 22588
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.49079252552975e-06,
+      "loss": 0.1021,
+      "step": 22589
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.489960750891571e-06,
+      "loss": 0.1244,
+      "step": 22590
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4891290309905165e-06,
+      "loss": 0.0746,
+      "step": 22591
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.488297365834858e-06,
+      "loss": 0.1151,
+      "step": 22592
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.48746575543285e-06,
+      "loss": 0.0945,
+      "step": 22593
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.486634199792757e-06,
+      "loss": 0.1278,
+      "step": 22594
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.485802698922849e-06,
+      "loss": 0.0853,
+      "step": 22595
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4849712528313685e-06,
+      "loss": 0.0917,
+      "step": 22596
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.484139861526583e-06,
+      "loss": 0.1435,
+      "step": 22597
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.483308525016756e-06,
+      "loss": 0.0999,
+      "step": 22598
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.482477243310138e-06,
+      "loss": 0.1169,
+      "step": 22599
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.481646016414997e-06,
+      "loss": 0.1132,
+      "step": 22600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4808148443395836e-06,
+      "loss": 0.1421,
+      "step": 22601
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.479983727092152e-06,
+      "loss": 0.0951,
+      "step": 22602
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.479152664680963e-06,
+      "loss": 0.1026,
+      "step": 22603
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.478321657114279e-06,
+      "loss": 0.109,
+      "step": 22604
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4774907044003426e-06,
+      "loss": 0.0749,
+      "step": 22605
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.476659806547412e-06,
+      "loss": 0.1302,
+      "step": 22606
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.475828963563748e-06,
+      "loss": 0.0924,
+      "step": 22607
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474998175457597e-06,
+      "loss": 0.0934,
+      "step": 22608
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474167442237218e-06,
+      "loss": 0.0862,
+      "step": 22609
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.473336763910861e-06,
+      "loss": 0.1133,
+      "step": 22610
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4725061404867735e-06,
+      "loss": 0.0972,
+      "step": 22611
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.47167557197321e-06,
+      "loss": 0.0953,
+      "step": 22612
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.470845058378427e-06,
+      "loss": 0.0977,
+      "step": 22613
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4700145997106705e-06,
+      "loss": 0.1396,
+      "step": 22614
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4691841959781865e-06,
+      "loss": 0.0866,
+      "step": 22615
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.468353847189231e-06,
+      "loss": 0.1148,
+      "step": 22616
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.467523553352047e-06,
+      "loss": 0.1262,
+      "step": 22617
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.466693314474885e-06,
+      "loss": 0.1069,
+      "step": 22618
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.465863130565999e-06,
+      "loss": 0.098,
+      "step": 22619
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.465033001633625e-06,
+      "loss": 0.1437,
+      "step": 22620
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.464202927686014e-06,
+      "loss": 0.1047,
+      "step": 22621
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.463372908731416e-06,
+      "loss": 0.0915,
+      "step": 22622
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.46254294477807e-06,
+      "loss": 0.1025,
+      "step": 22623
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4617130358342274e-06,
+      "loss": 0.1446,
+      "step": 22624
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.46088318190813e-06,
+      "loss": 0.1131,
+      "step": 22625
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.460053383008016e-06,
+      "loss": 0.1011,
+      "step": 22626
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.459223639142135e-06,
+      "loss": 0.0947,
+      "step": 22627
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4583939503187315e-06,
+      "loss": 0.0993,
+      "step": 22628
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.457564316546043e-06,
+      "loss": 0.1201,
+      "step": 22629
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.456734737832311e-06,
+      "loss": 0.1078,
+      "step": 22630
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.455905214185782e-06,
+      "loss": 0.1419,
+      "step": 22631
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.455075745614689e-06,
+      "loss": 0.1,
+      "step": 22632
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.454246332127279e-06,
+      "loss": 0.1265,
+      "step": 22633
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.453416973731788e-06,
+      "loss": 0.1175,
+      "step": 22634
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.452587670436451e-06,
+      "loss": 0.124,
+      "step": 22635
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.451758422249513e-06,
+      "loss": 0.1237,
+      "step": 22636
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.45092922917921e-06,
+      "loss": 0.1077,
+      "step": 22637
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.450100091233781e-06,
+      "loss": 0.1073,
+      "step": 22638
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.449271008421455e-06,
+      "loss": 0.1358,
+      "step": 22639
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.448441980750479e-06,
+      "loss": 0.0918,
+      "step": 22640
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.447613008229078e-06,
+      "loss": 0.0978,
+      "step": 22641
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.446784090865496e-06,
+      "loss": 0.1087,
+      "step": 22642
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.445955228667964e-06,
+      "loss": 0.0884,
+      "step": 22643
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.445126421644712e-06,
+      "loss": 0.1132,
+      "step": 22644
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.1116,
+      "step": 22645
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4434689731539956e-06,
+      "loss": 0.1021,
+      "step": 22646
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.442640331702993e-06,
+      "loss": 0.1077,
+      "step": 22647
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.441811745459209e-06,
+      "loss": 0.1396,
+      "step": 22648
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.440983214430869e-06,
+      "loss": 0.0927,
+      "step": 22649
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.440154738626202e-06,
+      "loss": 0.1059,
+      "step": 22650
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.439326318053445e-06,
+      "loss": 0.1221,
+      "step": 22651
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.43849795272082e-06,
+      "loss": 0.1177,
+      "step": 22652
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4376696426365636e-06,
+      "loss": 0.1021,
+      "step": 22653
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.436841387808901e-06,
+      "loss": 0.1309,
+      "step": 22654
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.436013188246056e-06,
+      "loss": 0.1393,
+      "step": 22655
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.435185043956258e-06,
+      "loss": 0.1421,
+      "step": 22656
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.434356954947741e-06,
+      "loss": 0.1048,
+      "step": 22657
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4335289212287245e-06,
+      "loss": 0.1353,
+      "step": 22658
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.432700942807432e-06,
+      "loss": 0.1044,
+      "step": 22659
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.431873019692097e-06,
+      "loss": 0.0993,
+      "step": 22660
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.431045151890935e-06,
+      "loss": 0.081,
+      "step": 22661
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.430217339412178e-06,
+      "loss": 0.1118,
+      "step": 22662
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4293895822640464e-06,
+      "loss": 0.119,
+      "step": 22663
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.428561880454759e-06,
+      "loss": 0.0874,
+      "step": 22664
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.427734233992542e-06,
+      "loss": 0.1215,
+      "step": 22665
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.426906642885621e-06,
+      "loss": 0.1055,
+      "step": 22666
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.426079107142214e-06,
+      "loss": 0.0812,
+      "step": 22667
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.425251626770538e-06,
+      "loss": 0.1334,
+      "step": 22668
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.424424201778821e-06,
+      "loss": 0.0818,
+      "step": 22669
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.423596832175273e-06,
+      "loss": 0.1331,
+      "step": 22670
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.422769517968119e-06,
+      "loss": 0.0983,
+      "step": 22671
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421942259165586e-06,
+      "loss": 0.0876,
+      "step": 22672
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421115055775875e-06,
+      "loss": 0.0961,
+      "step": 22673
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.420287907807212e-06,
+      "loss": 0.105,
+      "step": 22674
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.419460815267818e-06,
+      "loss": 0.1225,
+      "step": 22675
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4186337781659006e-06,
+      "loss": 0.1222,
+      "step": 22676
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.417806796509685e-06,
+      "loss": 0.1287,
+      "step": 22677
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.416979870307381e-06,
+      "loss": 0.1361,
+      "step": 22678
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.416152999567199e-06,
+      "loss": 0.0982,
+      "step": 22679
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.415326184297359e-06,
+      "loss": 0.1159,
+      "step": 22680
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.414499424506079e-06,
+      "loss": 0.1188,
+      "step": 22681
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.413672720201565e-06,
+      "loss": 0.1019,
+      "step": 22682
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.41284607139203e-06,
+      "loss": 0.1312,
+      "step": 22683
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.412019478085688e-06,
+      "loss": 0.1184,
+      "step": 22684
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.41119294029075e-06,
+      "loss": 0.1132,
+      "step": 22685
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.410366458015424e-06,
+      "loss": 0.081,
+      "step": 22686
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.409540031267932e-06,
+      "loss": 0.0965,
+      "step": 22687
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4087136600564664e-06,
+      "loss": 0.1233,
+      "step": 22688
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.407887344389244e-06,
+      "loss": 0.1096,
+      "step": 22689
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.407061084274481e-06,
+      "loss": 0.0916,
+      "step": 22690
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.406234879720377e-06,
+      "loss": 0.0997,
+      "step": 22691
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.405408730735138e-06,
+      "loss": 0.0872,
+      "step": 22692
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.404582637326978e-06,
+      "loss": 0.0828,
+      "step": 22693
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.403756599504098e-06,
+      "loss": 0.1336,
+      "step": 22694
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.402930617274704e-06,
+      "loss": 0.1539,
+      "step": 22695
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4021046906470125e-06,
+      "loss": 0.083,
+      "step": 22696
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.401278819629211e-06,
+      "loss": 0.1238,
+      "step": 22697
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.400453004229511e-06,
+      "loss": 0.1023,
+      "step": 22698
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.399627244456122e-06,
+      "loss": 0.1135,
+      "step": 22699
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.398801540317238e-06,
+      "loss": 0.1304,
+      "step": 22700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.397975891821069e-06,
+      "loss": 0.1255,
+      "step": 22701
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.397150298975815e-06,
+      "loss": 0.0868,
+      "step": 22702
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.396324761789672e-06,
+      "loss": 0.0948,
+      "step": 22703
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.395499280270846e-06,
+      "loss": 0.0922,
+      "step": 22704
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.394673854427541e-06,
+      "loss": 0.101,
+      "step": 22705
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.393848484267951e-06,
+      "loss": 0.1101,
+      "step": 22706
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.393023169800273e-06,
+      "loss": 0.1216,
+      "step": 22707
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.392197911032715e-06,
+      "loss": 0.092,
+      "step": 22708
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.391372707973465e-06,
+      "loss": 0.141,
+      "step": 22709
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.390547560630729e-06,
+      "loss": 0.1136,
+      "step": 22710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.389722469012702e-06,
+      "loss": 0.1295,
+      "step": 22711
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.388897433127573e-06,
+      "loss": 0.1006,
+      "step": 22712
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.388072452983546e-06,
+      "loss": 0.1026,
+      "step": 22713
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.387247528588817e-06,
+      "loss": 0.0997,
+      "step": 22714
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.386422659951575e-06,
+      "loss": 0.1341,
+      "step": 22715
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.385597847080021e-06,
+      "loss": 0.0718,
+      "step": 22716
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.384773089982346e-06,
+      "loss": 0.1017,
+      "step": 22717
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.383948388666738e-06,
+      "loss": 0.0933,
+      "step": 22718
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3831237431413984e-06,
+      "loss": 0.1036,
+      "step": 22719
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.382299153414514e-06,
+      "loss": 0.1042,
+      "step": 22720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.381474619494275e-06,
+      "loss": 0.1071,
+      "step": 22721
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.380650141388875e-06,
+      "loss": 0.0948,
+      "step": 22722
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3798257191065066e-06,
+      "loss": 0.1124,
+      "step": 22723
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.379001352655353e-06,
+      "loss": 0.0829,
+      "step": 22724
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.378177042043613e-06,
+      "loss": 0.103,
+      "step": 22725
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.37735278727947e-06,
+      "loss": 0.1104,
+      "step": 22726
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.376528588371108e-06,
+      "loss": 0.0719,
+      "step": 22727
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.375704445326723e-06,
+      "loss": 0.104,
+      "step": 22728
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.374880358154494e-06,
+      "loss": 0.1197,
+      "step": 22729
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.374056326862615e-06,
+      "loss": 0.1227,
+      "step": 22730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.373232351459266e-06,
+      "loss": 0.1209,
+      "step": 22731
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.372408431952638e-06,
+      "loss": 0.1209,
+      "step": 22732
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.37158456835091e-06,
+      "loss": 0.1004,
+      "step": 22733
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.370760760662273e-06,
+      "loss": 0.085,
+      "step": 22734
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.369937008894906e-06,
+      "loss": 0.0931,
+      "step": 22735
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.369113313056991e-06,
+      "loss": 0.1742,
+      "step": 22736
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.368289673156715e-06,
+      "loss": 0.1273,
+      "step": 22737
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.367466089202255e-06,
+      "loss": 0.0997,
+      "step": 22738
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.366642561201795e-06,
+      "loss": 0.1154,
+      "step": 22739
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.365819089163521e-06,
+      "loss": 0.0885,
+      "step": 22740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.364995673095608e-06,
+      "loss": 0.0944,
+      "step": 22741
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.364172313006233e-06,
+      "loss": 0.1228,
+      "step": 22742
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3633490089035835e-06,
+      "loss": 0.1141,
+      "step": 22743
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.362525760795834e-06,
+      "loss": 0.1111,
+      "step": 22744
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.361702568691159e-06,
+      "loss": 0.0969,
+      "step": 22745
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3608794325977435e-06,
+      "loss": 0.0979,
+      "step": 22746
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.360056352523756e-06,
+      "loss": 0.0783,
+      "step": 22747
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.359233328477379e-06,
+      "loss": 0.1302,
+      "step": 22748
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3584103604667905e-06,
+      "loss": 0.1366,
+      "step": 22749
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.357587448500163e-06,
+      "loss": 0.1138,
+      "step": 22750
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.356764592585667e-06,
+      "loss": 0.0999,
+      "step": 22751
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.355941792731485e-06,
+      "loss": 0.0824,
+      "step": 22752
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.355119048945783e-06,
+      "loss": 0.158,
+      "step": 22753
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.354296361236742e-06,
+      "loss": 0.1229,
+      "step": 22754
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.353473729612531e-06,
+      "loss": 0.0981,
+      "step": 22755
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.352651154081316e-06,
+      "loss": 0.1143,
+      "step": 22756
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.351828634651276e-06,
+      "loss": 0.1046,
+      "step": 22757
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3510061713305825e-06,
+      "loss": 0.1132,
+      "step": 22758
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.350183764127405e-06,
+      "loss": 0.072,
+      "step": 22759
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.349361413049906e-06,
+      "loss": 0.0962,
+      "step": 22760
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.348539118106266e-06,
+      "loss": 0.099,
+      "step": 22761
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3477168793046445e-06,
+      "loss": 0.1466,
+      "step": 22762
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.346894696653213e-06,
+      "loss": 0.1761,
+      "step": 22763
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.346072570160148e-06,
+      "loss": 0.1088,
+      "step": 22764
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3452504998336005e-06,
+      "loss": 0.1366,
+      "step": 22765
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.344428485681746e-06,
+      "loss": 0.1215,
+      "step": 22766
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.34360652771275e-06,
+      "loss": 0.1395,
+      "step": 22767
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.342784625934776e-06,
+      "loss": 0.0797,
+      "step": 22768
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.341962780355993e-06,
+      "loss": 0.1316,
+      "step": 22769
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.341140990984561e-06,
+      "loss": 0.1297,
+      "step": 22770
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3403192578286435e-06,
+      "loss": 0.1435,
+      "step": 22771
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.339497580896404e-06,
+      "loss": 0.1295,
+      "step": 22772
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.338675960196012e-06,
+      "loss": 0.1332,
+      "step": 22773
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.337854395735619e-06,
+      "loss": 0.0818,
+      "step": 22774
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.33703288752339e-06,
+      "loss": 0.0973,
+      "step": 22775
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.336211435567491e-06,
+      "loss": 0.1547,
+      "step": 22776
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.335390039876074e-06,
+      "loss": 0.0935,
+      "step": 22777
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3345687004573075e-06,
+      "loss": 0.1213,
+      "step": 22778
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.333747417319348e-06,
+      "loss": 0.1282,
+      "step": 22779
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.332926190470347e-06,
+      "loss": 0.1287,
+      "step": 22780
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.33210501991847e-06,
+      "loss": 0.1084,
+      "step": 22781
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.331283905671875e-06,
+      "loss": 0.1091,
+      "step": 22782
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.330462847738718e-06,
+      "loss": 0.1225,
+      "step": 22783
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.32964184612715e-06,
+      "loss": 0.1174,
+      "step": 22784
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.328820900845336e-06,
+      "loss": 0.1049,
+      "step": 22785
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.328000011901422e-06,
+      "loss": 0.1537,
+      "step": 22786
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.327179179303568e-06,
+      "loss": 0.121,
+      "step": 22787
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.326358403059935e-06,
+      "loss": 0.1456,
+      "step": 22788
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.325537683178661e-06,
+      "loss": 0.079,
+      "step": 22789
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.324717019667908e-06,
+      "loss": 0.0967,
+      "step": 22790
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.323896412535832e-06,
+      "loss": 0.1081,
+      "step": 22791
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.323075861790576e-06,
+      "loss": 0.1049,
+      "step": 22792
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.322255367440301e-06,
+      "loss": 0.1078,
+      "step": 22793
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3214349294931525e-06,
+      "loss": 0.1121,
+      "step": 22794
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.320614547957278e-06,
+      "loss": 0.1332,
+      "step": 22795
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.319794222840831e-06,
+      "loss": 0.1,
+      "step": 22796
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.318973954151963e-06,
+      "loss": 0.0801,
+      "step": 22797
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.318153741898821e-06,
+      "loss": 0.1123,
+      "step": 22798
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.317333586089547e-06,
+      "loss": 0.0999,
+      "step": 22799
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.316513486732297e-06,
+      "loss": 0.0911,
+      "step": 22800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.315693443835213e-06,
+      "loss": 0.1132,
+      "step": 22801
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.314873457406444e-06,
+      "loss": 0.1601,
+      "step": 22802
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.314053527454136e-06,
+      "loss": 0.0951,
+      "step": 22803
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.313233653986428e-06,
+      "loss": 0.1112,
+      "step": 22804
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.31241383701147e-06,
+      "loss": 0.0828,
+      "step": 22805
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.31159407653741e-06,
+      "loss": 0.1257,
+      "step": 22806
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.310774372572386e-06,
+      "loss": 0.1227,
+      "step": 22807
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.309954725124539e-06,
+      "loss": 0.1047,
+      "step": 22808
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.309135134202017e-06,
+      "loss": 0.1164,
+      "step": 22809
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.308315599812958e-06,
+      "loss": 0.1581,
+      "step": 22810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.307496121965506e-06,
+      "loss": 0.1076,
+      "step": 22811
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.306676700667802e-06,
+      "loss": 0.1296,
+      "step": 22812
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.305857335927981e-06,
+      "loss": 0.1454,
+      "step": 22813
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.30503802775419e-06,
+      "loss": 0.0875,
+      "step": 22814
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.30421877615456e-06,
+      "loss": 0.1291,
+      "step": 22815
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.303399581137236e-06,
+      "loss": 0.1246,
+      "step": 22816
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.302580442710355e-06,
+      "loss": 0.1057,
+      "step": 22817
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.301761360882054e-06,
+      "loss": 0.1186,
+      "step": 22818
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.300942335660466e-06,
+      "loss": 0.1049,
+      "step": 22819
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3001233670537335e-06,
+      "loss": 0.1078,
+      "step": 22820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.299304455069985e-06,
+      "loss": 0.0955,
+      "step": 22821
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.298485599717364e-06,
+      "loss": 0.1161,
+      "step": 22822
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.297666801004e-06,
+      "loss": 0.1163,
+      "step": 22823
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.296848058938024e-06,
+      "loss": 0.1021,
+      "step": 22824
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.296029373527573e-06,
+      "loss": 0.1162,
+      "step": 22825
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.295210744780786e-06,
+      "loss": 0.0795,
+      "step": 22826
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.294392172705787e-06,
+      "loss": 0.0916,
+      "step": 22827
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.293573657310706e-06,
+      "loss": 0.0873,
+      "step": 22828
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.292755198603684e-06,
+      "loss": 0.1196,
+      "step": 22829
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.291936796592841e-06,
+      "loss": 0.1073,
+      "step": 22830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.291118451286315e-06,
+      "loss": 0.1293,
+      "step": 22831
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.290300162692235e-06,
+      "loss": 0.0945,
+      "step": 22832
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.289481930818722e-06,
+      "loss": 0.1267,
+      "step": 22833
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.28866375567391e-06,
+      "loss": 0.1407,
+      "step": 22834
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2878456372659295e-06,
+      "loss": 0.1228,
+      "step": 22835
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.287027575602906e-06,
+      "loss": 0.1202,
+      "step": 22836
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2862095706929605e-06,
+      "loss": 0.1213,
+      "step": 22837
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.285391622544228e-06,
+      "loss": 0.1034,
+      "step": 22838
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.284573731164826e-06,
+      "loss": 0.1127,
+      "step": 22839
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2837558965628835e-06,
+      "loss": 0.0901,
+      "step": 22840
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.282938118746531e-06,
+      "loss": 0.118,
+      "step": 22841
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.282120397723879e-06,
+      "loss": 0.1197,
+      "step": 22842
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.281302733503058e-06,
+      "loss": 0.1023,
+      "step": 22843
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.280485126092194e-06,
+      "loss": 0.1272,
+      "step": 22844
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.279667575499402e-06,
+      "loss": 0.1273,
+      "step": 22845
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.278850081732812e-06,
+      "loss": 0.1041,
+      "step": 22846
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.278032644800539e-06,
+      "loss": 0.1093,
+      "step": 22847
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.277215264710703e-06,
+      "loss": 0.0843,
+      "step": 22848
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.276397941471424e-06,
+      "loss": 0.1379,
+      "step": 22849
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.275580675090827e-06,
+      "loss": 0.0995,
+      "step": 22850
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.274763465577027e-06,
+      "loss": 0.124,
+      "step": 22851
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.273946312938138e-06,
+      "loss": 0.0986,
+      "step": 22852
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.273129217182287e-06,
+      "loss": 0.1068,
+      "step": 22853
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.27231217831758e-06,
+      "loss": 0.1261,
+      "step": 22854
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.271495196352141e-06,
+      "loss": 0.1099,
+      "step": 22855
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.270678271294091e-06,
+      "loss": 0.1302,
+      "step": 22856
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.269861403151532e-06,
+      "loss": 0.097,
+      "step": 22857
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.269044591932585e-06,
+      "loss": 0.1506,
+      "step": 22858
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.268227837645369e-06,
+      "loss": 0.1028,
+      "step": 22859
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.267411140297992e-06,
+      "loss": 0.1177,
+      "step": 22860
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.266594499898567e-06,
+      "loss": 0.1526,
+      "step": 22861
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2657779164552106e-06,
+      "loss": 0.1351,
+      "step": 22862
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.264961389976029e-06,
+      "loss": 0.0938,
+      "step": 22863
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2641449204691375e-06,
+      "loss": 0.1406,
+      "step": 22864
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.263328507942653e-06,
+      "loss": 0.106,
+      "step": 22865
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.262512152404672e-06,
+      "loss": 0.1503,
+      "step": 22866
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.261695853863312e-06,
+      "loss": 0.1007,
+      "step": 22867
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.260879612326686e-06,
+      "loss": 0.1025,
+      "step": 22868
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.260063427802893e-06,
+      "loss": 0.1216,
+      "step": 22869
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.259247300300051e-06,
+      "loss": 0.1579,
+      "step": 22870
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.258431229826263e-06,
+      "loss": 0.0974,
+      "step": 22871
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2576152163896324e-06,
+      "loss": 0.1093,
+      "step": 22872
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.256799259998268e-06,
+      "loss": 0.0957,
+      "step": 22873
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.255983360660279e-06,
+      "loss": 0.1038,
+      "step": 22874
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.255167518383771e-06,
+      "loss": 0.1126,
+      "step": 22875
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.254351733176839e-06,
+      "loss": 0.1525,
+      "step": 22876
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253536005047599e-06,
+      "loss": 0.102,
+      "step": 22877
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.252720334004146e-06,
+      "loss": 0.0951,
+      "step": 22878
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.25190472005459e-06,
+      "loss": 0.1231,
+      "step": 22879
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.251089163207027e-06,
+      "loss": 0.1011,
+      "step": 22880
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.25027366346956e-06,
+      "loss": 0.0934,
+      "step": 22881
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.249458220850291e-06,
+      "loss": 0.096,
+      "step": 22882
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.248642835357325e-06,
+      "loss": 0.0968,
+      "step": 22883
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.247827506998754e-06,
+      "loss": 0.1048,
+      "step": 22884
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.247012235782686e-06,
+      "loss": 0.1484,
+      "step": 22885
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.246197021717215e-06,
+      "loss": 0.1121,
+      "step": 22886
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.245381864810438e-06,
+      "loss": 0.109,
+      "step": 22887
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.244566765070458e-06,
+      "loss": 0.111,
+      "step": 22888
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.243751722505369e-06,
+      "loss": 0.1149,
+      "step": 22889
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2429367371232656e-06,
+      "loss": 0.1288,
+      "step": 22890
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.242121808932246e-06,
+      "loss": 0.1291,
+      "step": 22891
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.241306937940408e-06,
+      "loss": 0.1163,
+      "step": 22892
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.240492124155843e-06,
+      "loss": 0.1208,
+      "step": 22893
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.23967736758665e-06,
+      "loss": 0.1211,
+      "step": 22894
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.23886266824092e-06,
+      "loss": 0.129,
+      "step": 22895
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.238048026126742e-06,
+      "loss": 0.1138,
+      "step": 22896
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.237233441252218e-06,
+      "loss": 0.13,
+      "step": 22897
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.23641891362543e-06,
+      "loss": 0.1332,
+      "step": 22898
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.235604443254478e-06,
+      "loss": 0.1054,
+      "step": 22899
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.234790030147447e-06,
+      "loss": 0.0964,
+      "step": 22900
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.233975674312434e-06,
+      "loss": 0.1114,
+      "step": 22901
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.233161375757522e-06,
+      "loss": 0.0905,
+      "step": 22902
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.232347134490805e-06,
+      "loss": 0.1127,
+      "step": 22903
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.231532950520371e-06,
+      "loss": 0.0913,
+      "step": 22904
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.230718823854304e-06,
+      "loss": 0.1311,
+      "step": 22905
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.229904754500697e-06,
+      "loss": 0.1105,
+      "step": 22906
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.229090742467632e-06,
+      "loss": 0.1129,
+      "step": 22907
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.228276787763198e-06,
+      "loss": 0.1551,
+      "step": 22908
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.227462890395485e-06,
+      "loss": 0.1011,
+      "step": 22909
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.226649050372575e-06,
+      "loss": 0.1077,
+      "step": 22910
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.225835267702548e-06,
+      "loss": 0.0852,
+      "step": 22911
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.225021542393496e-06,
+      "loss": 0.1228,
+      "step": 22912
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.224207874453499e-06,
+      "loss": 0.089,
+      "step": 22913
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.223394263890637e-06,
+      "loss": 0.1085,
+      "step": 22914
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.222580710712998e-06,
+      "loss": 0.1081,
+      "step": 22915
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.221767214928658e-06,
+      "loss": 0.1207,
+      "step": 22916
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.220953776545702e-06,
+      "loss": 0.1146,
+      "step": 22917
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2201403955722145e-06,
+      "loss": 0.0917,
+      "step": 22918
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.21932707201627e-06,
+      "loss": 0.1091,
+      "step": 22919
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.218513805885948e-06,
+      "loss": 0.0691,
+      "step": 22920
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.217700597189331e-06,
+      "loss": 0.1183,
+      "step": 22921
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2168874459344945e-06,
+      "loss": 0.0901,
+      "step": 22922
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.21607435212952e-06,
+      "loss": 0.1333,
+      "step": 22923
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.215261315782483e-06,
+      "loss": 0.0853,
+      "step": 22924
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2144483369014555e-06,
+      "loss": 0.1674,
+      "step": 22925
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.213635415494518e-06,
+      "loss": 0.0874,
+      "step": 22926
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.21282255156975e-06,
+      "loss": 0.1183,
+      "step": 22927
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.212009745135223e-06,
+      "loss": 0.0889,
+      "step": 22928
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.211196996199006e-06,
+      "loss": 0.0975,
+      "step": 22929
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.210384304769183e-06,
+      "loss": 0.0957,
+      "step": 22930
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.20957167085382e-06,
+      "loss": 0.1043,
+      "step": 22931
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.208759094460989e-06,
+      "loss": 0.1079,
+      "step": 22932
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.207946575598774e-06,
+      "loss": 0.0721,
+      "step": 22933
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.207134114275231e-06,
+      "loss": 0.0827,
+      "step": 22934
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2063217104984374e-06,
+      "loss": 0.1376,
+      "step": 22935
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2055093642764685e-06,
+      "loss": 0.101,
+      "step": 22936
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2046970756173865e-06,
+      "loss": 0.1483,
+      "step": 22937
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.203884844529268e-06,
+      "loss": 0.1617,
+      "step": 22938
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2030726710201764e-06,
+      "loss": 0.092,
+      "step": 22939
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.20226055509818e-06,
+      "loss": 0.115,
+      "step": 22940
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.201448496771346e-06,
+      "loss": 0.1321,
+      "step": 22941
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.200636496047751e-06,
+      "loss": 0.1324,
+      "step": 22942
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.199824552935447e-06,
+      "loss": 0.1084,
+      "step": 22943
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.199012667442507e-06,
+      "loss": 0.1373,
+      "step": 22944
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.198200839576999e-06,
+      "loss": 0.1128,
+      "step": 22945
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.197389069346981e-06,
+      "loss": 0.0845,
+      "step": 22946
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.196577356760525e-06,
+      "loss": 0.131,
+      "step": 22947
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.19576570182569e-06,
+      "loss": 0.1404,
+      "step": 22948
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.194954104550537e-06,
+      "loss": 0.1173,
+      "step": 22949
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.194142564943131e-06,
+      "loss": 0.1189,
+      "step": 22950
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.193331083011537e-06,
+      "loss": 0.1265,
+      "step": 22951
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.192519658763813e-06,
+      "loss": 0.0908,
+      "step": 22952
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.191708292208017e-06,
+      "loss": 0.1529,
+      "step": 22953
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.190896983352216e-06,
+      "loss": 0.1243,
+      "step": 22954
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.190085732204462e-06,
+      "loss": 0.0931,
+      "step": 22955
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.189274538772817e-06,
+      "loss": 0.1268,
+      "step": 22956
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1884634030653496e-06,
+      "loss": 0.1021,
+      "step": 22957
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.187652325090099e-06,
+      "loss": 0.1206,
+      "step": 22958
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1868413048551325e-06,
+      "loss": 0.1266,
+      "step": 22959
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.18603034236851e-06,
+      "loss": 0.1015,
+      "step": 22960
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.18521943763828e-06,
+      "loss": 0.0992,
+      "step": 22961
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.184408590672506e-06,
+      "loss": 0.1186,
+      "step": 22962
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1835978014792386e-06,
+      "loss": 0.151,
+      "step": 22963
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.182787070066529e-06,
+      "loss": 0.1144,
+      "step": 22964
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.181976396442436e-06,
+      "loss": 0.073,
+      "step": 22965
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.181165780615014e-06,
+      "loss": 0.1013,
+      "step": 22966
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.180355222592314e-06,
+      "loss": 0.1017,
+      "step": 22967
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.179544722382383e-06,
+      "loss": 0.0927,
+      "step": 22968
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.178734279993281e-06,
+      "loss": 0.1047,
+      "step": 22969
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.177923895433052e-06,
+      "loss": 0.1266,
+      "step": 22970
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.177113568709752e-06,
+      "loss": 0.1326,
+      "step": 22971
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1763032998314276e-06,
+      "loss": 0.1106,
+      "step": 22972
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.175493088806126e-06,
+      "loss": 0.0972,
+      "step": 22973
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.174682935641902e-06,
+      "loss": 0.0883,
+      "step": 22974
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.173872840346798e-06,
+      "loss": 0.113,
+      "step": 22975
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.173062802928867e-06,
+      "loss": 0.0913,
+      "step": 22976
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.172252823396149e-06,
+      "loss": 0.1299,
+      "step": 22977
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.171442901756697e-06,
+      "loss": 0.1337,
+      "step": 22978
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.170633038018551e-06,
+      "loss": 0.0831,
+      "step": 22979
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.169823232189763e-06,
+      "loss": 0.1266,
+      "step": 22980
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1690134842783744e-06,
+      "loss": 0.1179,
+      "step": 22981
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.168203794292425e-06,
+      "loss": 0.1011,
+      "step": 22982
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.167394162239966e-06,
+      "loss": 0.1047,
+      "step": 22983
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.166584588129032e-06,
+      "loss": 0.1253,
+      "step": 22984
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1657750719676715e-06,
+      "loss": 0.1227,
+      "step": 22985
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.164965613763927e-06,
+      "loss": 0.1179,
+      "step": 22986
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.164156213525838e-06,
+      "loss": 0.1304,
+      "step": 22987
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.163346871261441e-06,
+      "loss": 0.1265,
+      "step": 22988
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.162537586978783e-06,
+      "loss": 0.088,
+      "step": 22989
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1617283606858975e-06,
+      "loss": 0.1141,
+      "step": 22990
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.160919192390827e-06,
+      "loss": 0.0888,
+      "step": 22991
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.160110082101612e-06,
+      "loss": 0.1046,
+      "step": 22992
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.159301029826282e-06,
+      "loss": 0.1088,
+      "step": 22993
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.158492035572879e-06,
+      "loss": 0.1365,
+      "step": 22994
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.157683099349444e-06,
+      "loss": 0.1046,
+      "step": 22995
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.156874221164009e-06,
+      "loss": 0.1005,
+      "step": 22996
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.156065401024606e-06,
+      "loss": 0.0874,
+      "step": 22997
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.155256638939276e-06,
+      "loss": 0.1072,
+      "step": 22998
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.154447934916048e-06,
+      "loss": 0.1299,
+      "step": 22999
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.153639288962962e-06,
+      "loss": 0.1344,
+      "step": 23000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.152830701088046e-06,
+      "loss": 0.0991,
+      "step": 23001
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1520221712993316e-06,
+      "loss": 0.1432,
+      "step": 23002
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.151213699604853e-06,
+      "loss": 0.1215,
+      "step": 23003
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.150405286012645e-06,
+      "loss": 0.098,
+      "step": 23004
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.149596930530736e-06,
+      "loss": 0.0898,
+      "step": 23005
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1487886331671515e-06,
+      "loss": 0.1616,
+      "step": 23006
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.147980393929929e-06,
+      "loss": 0.1026,
+      "step": 23007
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.147172212827091e-06,
+      "loss": 0.1066,
+      "step": 23008
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.146364089866668e-06,
+      "loss": 0.1252,
+      "step": 23009
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.145556025056695e-06,
+      "loss": 0.0911,
+      "step": 23010
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.144748018405188e-06,
+      "loss": 0.1355,
+      "step": 23011
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.143940069920178e-06,
+      "loss": 0.0997,
+      "step": 23012
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.143132179609695e-06,
+      "loss": 0.098,
+      "step": 23013
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.14232434748176e-06,
+      "loss": 0.1378,
+      "step": 23014
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1415165735444016e-06,
+      "loss": 0.0996,
+      "step": 23015
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.140708857805645e-06,
+      "loss": 0.1388,
+      "step": 23016
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.139901200273506e-06,
+      "loss": 0.1194,
+      "step": 23017
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.139093600956014e-06,
+      "loss": 0.1242,
+      "step": 23018
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1382860598611955e-06,
+      "loss": 0.1168,
+      "step": 23019
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.137478576997068e-06,
+      "loss": 0.1155,
+      "step": 23020
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.136671152371651e-06,
+      "loss": 0.0997,
+      "step": 23021
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.135863785992971e-06,
+      "loss": 0.1084,
+      "step": 23022
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.135056477869043e-06,
+      "loss": 0.075,
+      "step": 23023
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.134249228007888e-06,
+      "loss": 0.1721,
+      "step": 23024
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.133442036417534e-06,
+      "loss": 0.1259,
+      "step": 23025
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.132634903105987e-06,
+      "loss": 0.1205,
+      "step": 23026
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131827828081268e-06,
+      "loss": 0.1133,
+      "step": 23027
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131020811351403e-06,
+      "loss": 0.0954,
+      "step": 23028
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1302138529244015e-06,
+      "loss": 0.0961,
+      "step": 23029
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.129406952808278e-06,
+      "loss": 0.1052,
+      "step": 23030
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.128600111011056e-06,
+      "loss": 0.1258,
+      "step": 23031
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.127793327540742e-06,
+      "loss": 0.0674,
+      "step": 23032
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.126986602405356e-06,
+      "loss": 0.157,
+      "step": 23033
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.126179935612918e-06,
+      "loss": 0.075,
+      "step": 23034
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.125373327171428e-06,
+      "loss": 0.0872,
+      "step": 23035
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.124566777088905e-06,
+      "loss": 0.0934,
+      "step": 23036
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.123760285373366e-06,
+      "loss": 0.1699,
+      "step": 23037
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.122953852032815e-06,
+      "loss": 0.1035,
+      "step": 23038
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.133,
+      "step": 23039
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.121341160508738e-06,
+      "loss": 0.0997,
+      "step": 23040
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.120534902341227e-06,
+      "loss": 0.1302,
+      "step": 23041
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.119728702580748e-06,
+      "loss": 0.0921,
+      "step": 23042
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.118922561235315e-06,
+      "loss": 0.1556,
+      "step": 23043
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.11811647831293e-06,
+      "loss": 0.108,
+      "step": 23044
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.117310453821602e-06,
+      "loss": 0.1078,
+      "step": 23045
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.116504487769339e-06,
+      "loss": 0.1024,
+      "step": 23046
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1156985801641454e-06,
+      "loss": 0.0736,
+      "step": 23047
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.114892731014033e-06,
+      "loss": 0.1201,
+      "step": 23048
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.114086940327004e-06,
+      "loss": 0.0856,
+      "step": 23049
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.113281208111056e-06,
+      "loss": 0.1381,
+      "step": 23050
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.112475534374202e-06,
+      "loss": 0.1023,
+      "step": 23051
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.111669919124446e-06,
+      "loss": 0.0766,
+      "step": 23052
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.110864362369784e-06,
+      "loss": 0.1238,
+      "step": 23053
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.110058864118227e-06,
+      "loss": 0.0887,
+      "step": 23054
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.109253424377773e-06,
+      "loss": 0.1108,
+      "step": 23055
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1084480431564176e-06,
+      "loss": 0.0812,
+      "step": 23056
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.107642720462172e-06,
+      "loss": 0.0929,
+      "step": 23057
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.106837456303031e-06,
+      "loss": 0.1133,
+      "step": 23058
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1060322506869895e-06,
+      "loss": 0.0953,
+      "step": 23059
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.105227103622052e-06,
+      "loss": 0.1029,
+      "step": 23060
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.10442201511622e-06,
+      "loss": 0.1084,
+      "step": 23061
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.103616985177483e-06,
+      "loss": 0.1004,
+      "step": 23062
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.102812013813847e-06,
+      "loss": 0.1039,
+      "step": 23063
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.102007101033304e-06,
+      "loss": 0.1233,
+      "step": 23064
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.101202246843846e-06,
+      "loss": 0.1073,
+      "step": 23065
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.100397451253478e-06,
+      "loss": 0.1181,
+      "step": 23066
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.099592714270185e-06,
+      "loss": 0.1207,
+      "step": 23067
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.09878803590197e-06,
+      "loss": 0.0693,
+      "step": 23068
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.09798341615682e-06,
+      "loss": 0.105,
+      "step": 23069
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.097178855042732e-06,
+      "loss": 0.1072,
+      "step": 23070
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.096374352567697e-06,
+      "loss": 0.0971,
+      "step": 23071
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0955699087397095e-06,
+      "loss": 0.1168,
+      "step": 23072
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.09476552356676e-06,
+      "loss": 0.0634,
+      "step": 23073
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.093961197056832e-06,
+      "loss": 0.131,
+      "step": 23074
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.093156929217929e-06,
+      "loss": 0.101,
+      "step": 23075
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.092352720058027e-06,
+      "loss": 0.123,
+      "step": 23076
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.091548569585123e-06,
+      "loss": 0.1067,
+      "step": 23077
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.090744477807208e-06,
+      "loss": 0.1372,
+      "step": 23078
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.089940444732265e-06,
+      "loss": 0.0987,
+      "step": 23079
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0891364703682805e-06,
+      "loss": 0.1358,
+      "step": 23080
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0883325547232455e-06,
+      "loss": 0.0955,
+      "step": 23081
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.087528697805144e-06,
+      "loss": 0.1069,
+      "step": 23082
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.086724899621958e-06,
+      "loss": 0.1262,
+      "step": 23083
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.08592116018168e-06,
+      "loss": 0.1065,
+      "step": 23084
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.085117479492285e-06,
+      "loss": 0.1171,
+      "step": 23085
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.084313857561764e-06,
+      "loss": 0.0886,
+      "step": 23086
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.083510294398101e-06,
+      "loss": 0.1036,
+      "step": 23087
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.082706790009275e-06,
+      "loss": 0.1271,
+      "step": 23088
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.081903344403267e-06,
+      "loss": 0.1524,
+      "step": 23089
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.081099957588064e-06,
+      "loss": 0.0947,
+      "step": 23090
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.08029662957164e-06,
+      "loss": 0.1295,
+      "step": 23091
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.079493360361981e-06,
+      "loss": 0.1144,
+      "step": 23092
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.078690149967065e-06,
+      "loss": 0.0916,
+      "step": 23093
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.077886998394867e-06,
+      "loss": 0.1622,
+      "step": 23094
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.07708390565337e-06,
+      "loss": 0.123,
+      "step": 23095
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.076280871750554e-06,
+      "loss": 0.0819,
+      "step": 23096
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.075477896694394e-06,
+      "loss": 0.1255,
+      "step": 23097
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0746749804928635e-06,
+      "loss": 0.083,
+      "step": 23098
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.073872123153944e-06,
+      "loss": 0.0943,
+      "step": 23099
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0730693246856055e-06,
+      "loss": 0.1185,
+      "step": 23100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.072266585095827e-06,
+      "loss": 0.1474,
+      "step": 23101
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.071463904392588e-06,
+      "loss": 0.0801,
+      "step": 23102
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.070661282583851e-06,
+      "loss": 0.0763,
+      "step": 23103
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.069858719677594e-06,
+      "loss": 0.1058,
+      "step": 23104
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.069056215681794e-06,
+      "loss": 0.0828,
+      "step": 23105
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.068253770604418e-06,
+      "loss": 0.0742,
+      "step": 23106
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0674513844534416e-06,
+      "loss": 0.1344,
+      "step": 23107
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0666490572368335e-06,
+      "loss": 0.1144,
+      "step": 23108
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.065846788962561e-06,
+      "loss": 0.1128,
+      "step": 23109
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.065044579638598e-06,
+      "loss": 0.0903,
+      "step": 23110
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.064242429272918e-06,
+      "loss": 0.1314,
+      "step": 23111
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.063440337873477e-06,
+      "loss": 0.1167,
+      "step": 23112
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.062638305448252e-06,
+      "loss": 0.1234,
+      "step": 23113
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0618363320052125e-06,
+      "loss": 0.1144,
+      "step": 23114
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0610344175523175e-06,
+      "loss": 0.0796,
+      "step": 23115
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.060232562097542e-06,
+      "loss": 0.0908,
+      "step": 23116
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.059430765648845e-06,
+      "loss": 0.1122,
+      "step": 23117
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.058629028214191e-06,
+      "loss": 0.101,
+      "step": 23118
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.057827349801548e-06,
+      "loss": 0.0767,
+      "step": 23119
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.057025730418882e-06,
+      "loss": 0.1391,
+      "step": 23120
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.056224170074155e-06,
+      "loss": 0.1188,
+      "step": 23121
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.055422668775323e-06,
+      "loss": 0.1152,
+      "step": 23122
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.054621226530358e-06,
+      "loss": 0.086,
+      "step": 23123
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.053819843347213e-06,
+      "loss": 0.0964,
+      "step": 23124
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.053018519233853e-06,
+      "loss": 0.0783,
+      "step": 23125
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.052217254198246e-06,
+      "loss": 0.0955,
+      "step": 23126
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.051416048248338e-06,
+      "loss": 0.108,
+      "step": 23127
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.050614901392094e-06,
+      "loss": 0.0635,
+      "step": 23128
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.049813813637475e-06,
+      "loss": 0.1053,
+      "step": 23129
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.049012784992436e-06,
+      "loss": 0.0954,
+      "step": 23130
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.048211815464939e-06,
+      "loss": 0.1223,
+      "step": 23131
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.047410905062936e-06,
+      "loss": 0.1087,
+      "step": 23132
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.046610053794383e-06,
+      "loss": 0.1164,
+      "step": 23133
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.045809261667241e-06,
+      "loss": 0.1124,
+      "step": 23134
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.045008528689457e-06,
+      "loss": 0.0841,
+      "step": 23135
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.044207854868996e-06,
+      "loss": 0.0905,
+      "step": 23136
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0434072402138006e-06,
+      "loss": 0.1339,
+      "step": 23137
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.042606684731835e-06,
+      "loss": 0.1005,
+      "step": 23138
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0418061884310415e-06,
+      "loss": 0.1258,
+      "step": 23139
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.041005751319382e-06,
+      "loss": 0.163,
+      "step": 23140
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0402053734048044e-06,
+      "loss": 0.1102,
+      "step": 23141
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0394050546952545e-06,
+      "loss": 0.1022,
+      "step": 23142
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0386047951986905e-06,
+      "loss": 0.1008,
+      "step": 23143
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.037804594923057e-06,
+      "loss": 0.1073,
+      "step": 23144
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.037004453876308e-06,
+      "loss": 0.1125,
+      "step": 23145
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.036204372066384e-06,
+      "loss": 0.1168,
+      "step": 23146
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.035404349501245e-06,
+      "loss": 0.097,
+      "step": 23147
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.034604386188826e-06,
+      "loss": 0.1427,
+      "step": 23148
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.033804482137085e-06,
+      "loss": 0.1299,
+      "step": 23149
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.033004637353962e-06,
+      "loss": 0.1208,
+      "step": 23150
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0322048518474014e-06,
+      "loss": 0.1312,
+      "step": 23151
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.031405125625354e-06,
+      "loss": 0.0988,
+      "step": 23152
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.030605458695756e-06,
+      "loss": 0.1555,
+      "step": 23153
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.029805851066558e-06,
+      "loss": 0.1058,
+      "step": 23154
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.029006302745706e-06,
+      "loss": 0.1258,
+      "step": 23155
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.028206813741138e-06,
+      "loss": 0.0999,
+      "step": 23156
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.027407384060792e-06,
+      "loss": 0.1257,
+      "step": 23157
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.02660801371262e-06,
+      "loss": 0.0835,
+      "step": 23158
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.025808702704551e-06,
+      "loss": 0.1022,
+      "step": 23159
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.025009451044538e-06,
+      "loss": 0.1134,
+      "step": 23160
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.024210258740513e-06,
+      "loss": 0.1402,
+      "step": 23161
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.023411125800414e-06,
+      "loss": 0.1145,
+      "step": 23162
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.022612052232182e-06,
+      "loss": 0.0948,
+      "step": 23163
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.021813038043759e-06,
+      "loss": 0.1194,
+      "step": 23164
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.02101408324308e-06,
+      "loss": 0.0941,
+      "step": 23165
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.020215187838077e-06,
+      "loss": 0.0988,
+      "step": 23166
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.019416351836694e-06,
+      "loss": 0.107,
+      "step": 23167
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.018617575246858e-06,
+      "loss": 0.0912,
+      "step": 23168
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.017818858076513e-06,
+      "loss": 0.081,
+      "step": 23169
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.01702020033359e-06,
+      "loss": 0.1051,
+      "step": 23170
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.016221602026018e-06,
+      "loss": 0.1254,
+      "step": 23171
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.015423063161735e-06,
+      "loss": 0.1371,
+      "step": 23172
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.014624583748678e-06,
+      "loss": 0.1355,
+      "step": 23173
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.013826163794774e-06,
+      "loss": 0.1114,
+      "step": 23174
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.01302780330795e-06,
+      "loss": 0.0752,
+      "step": 23175
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.012229502296149e-06,
+      "loss": 0.1002,
+      "step": 23176
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.011431260767289e-06,
+      "loss": 0.1157,
+      "step": 23177
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.010633078729306e-06,
+      "loss": 0.1394,
+      "step": 23178
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.009834956190134e-06,
+      "loss": 0.1174,
+      "step": 23179
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.00903689315769e-06,
+      "loss": 0.1052,
+      "step": 23180
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.00823888963991e-06,
+      "loss": 0.0926,
+      "step": 23181
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0074409456447205e-06,
+      "loss": 0.1072,
+      "step": 23182
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.006643061180045e-06,
+      "loss": 0.1261,
+      "step": 23183
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.005845236253816e-06,
+      "loss": 0.1059,
+      "step": 23184
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.005047470873954e-06,
+      "loss": 0.1048,
+      "step": 23185
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.004249765048382e-06,
+      "loss": 0.1274,
+      "step": 23186
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.003452118785029e-06,
+      "loss": 0.0922,
+      "step": 23187
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0026545320918205e-06,
+      "loss": 0.1118,
+      "step": 23188
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.001857004976678e-06,
+      "loss": 0.0941,
+      "step": 23189
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0010595374475195e-06,
+      "loss": 0.0912,
+      "step": 23190
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.000262129512273e-06,
+      "loss": 0.1061,
+      "step": 23191
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.999464781178855e-06,
+      "loss": 0.1539,
+      "step": 23192
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.998667492455189e-06,
+      "loss": 0.0899,
+      "step": 23193
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9978702633492026e-06,
+      "loss": 0.0796,
+      "step": 23194
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.997073093868802e-06,
+      "loss": 0.0912,
+      "step": 23195
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.996275984021912e-06,
+      "loss": 0.1158,
+      "step": 23196
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.995478933816456e-06,
+      "loss": 0.0882,
+      "step": 23197
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.994681943260348e-06,
+      "loss": 0.1014,
+      "step": 23198
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9938850123615e-06,
+      "loss": 0.125,
+      "step": 23199
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.99308814112784e-06,
+      "loss": 0.0939,
+      "step": 23200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.992291329567274e-06,
+      "loss": 0.1029,
+      "step": 23201
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.991494577687721e-06,
+      "loss": 0.0793,
+      "step": 23202
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.990697885497105e-06,
+      "loss": 0.1185,
+      "step": 23203
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.989901253003323e-06,
+      "loss": 0.112,
+      "step": 23204
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9891046802143e-06,
+      "loss": 0.1189,
+      "step": 23205
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9883081671379496e-06,
+      "loss": 0.0672,
+      "step": 23206
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98751171378218e-06,
+      "loss": 0.1435,
+      "step": 23207
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.986715320154907e-06,
+      "loss": 0.1048,
+      "step": 23208
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.985918986264042e-06,
+      "loss": 0.1032,
+      "step": 23209
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98512271211749e-06,
+      "loss": 0.1069,
+      "step": 23210
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9843264977231645e-06,
+      "loss": 0.1332,
+      "step": 23211
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98353034308898e-06,
+      "loss": 0.1411,
+      "step": 23212
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.982734248222841e-06,
+      "loss": 0.1046,
+      "step": 23213
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.981938213132654e-06,
+      "loss": 0.1267,
+      "step": 23214
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.981142237826332e-06,
+      "loss": 0.1154,
+      "step": 23215
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.980346322311777e-06,
+      "loss": 0.1582,
+      "step": 23216
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.979550466596902e-06,
+      "loss": 0.1114,
+      "step": 23217
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9787546706896085e-06,
+      "loss": 0.1575,
+      "step": 23218
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9779589345978e-06,
+      "loss": 0.116,
+      "step": 23219
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.977163258329385e-06,
+      "loss": 0.1191,
+      "step": 23220
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.976367641892269e-06,
+      "loss": 0.1132,
+      "step": 23221
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.975572085294356e-06,
+      "loss": 0.101,
+      "step": 23222
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.974776588543542e-06,
+      "loss": 0.1344,
+      "step": 23223
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9739811516477374e-06,
+      "loss": 0.1365,
+      "step": 23224
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.97318577461484e-06,
+      "loss": 0.1141,
+      "step": 23225
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.972390457452754e-06,
+      "loss": 0.1354,
+      "step": 23226
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.971595200169379e-06,
+      "loss": 0.0856,
+      "step": 23227
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.970800002772611e-06,
+      "loss": 0.0956,
+      "step": 23228
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.970004865270353e-06,
+      "loss": 0.1576,
+      "step": 23229
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.969209787670508e-06,
+      "loss": 0.1147,
+      "step": 23230
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.968414769980967e-06,
+      "loss": 0.1054,
+      "step": 23231
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.967619812209634e-06,
+      "loss": 0.1174,
+      "step": 23232
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.966824914364406e-06,
+      "loss": 0.0935,
+      "step": 23233
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.966030076453171e-06,
+      "loss": 0.0745,
+      "step": 23234
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.965235298483835e-06,
+      "loss": 0.1228,
+      "step": 23235
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.964440580464286e-06,
+      "loss": 0.0993,
+      "step": 23236
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.963645922402426e-06,
+      "loss": 0.0831,
+      "step": 23237
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.96285132430614e-06,
+      "loss": 0.1357,
+      "step": 23238
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.962056786183334e-06,
+      "loss": 0.112,
+      "step": 23239
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.961262308041888e-06,
+      "loss": 0.0779,
+      "step": 23240
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.960467889889704e-06,
+      "loss": 0.1062,
+      "step": 23241
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.95967353173467e-06,
+      "loss": 0.0859,
+      "step": 23242
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9588792335846745e-06,
+      "loss": 0.1013,
+      "step": 23243
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.958084995447614e-06,
+      "loss": 0.0825,
+      "step": 23244
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.957290817331373e-06,
+      "loss": 0.1167,
+      "step": 23245
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.956496699243842e-06,
+      "loss": 0.1108,
+      "step": 23246
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.955702641192916e-06,
+      "loss": 0.1419,
+      "step": 23247
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.954908643186478e-06,
+      "loss": 0.1276,
+      "step": 23248
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9541147052324126e-06,
+      "loss": 0.1206,
+      "step": 23249
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9533208273386135e-06,
+      "loss": 0.1181,
+      "step": 23250
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.952527009512964e-06,
+      "loss": 0.0927,
+      "step": 23251
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.951733251763346e-06,
+      "loss": 0.0797,
+      "step": 23252
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9509395540976534e-06,
+      "loss": 0.1004,
+      "step": 23253
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.950145916523762e-06,
+      "loss": 0.0887,
+      "step": 23254
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.949352339049561e-06,
+      "loss": 0.1062,
+      "step": 23255
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.948558821682935e-06,
+      "loss": 0.124,
+      "step": 23256
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.947765364431765e-06,
+      "loss": 0.1182,
+      "step": 23257
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.94697196730393e-06,
+      "loss": 0.1367,
+      "step": 23258
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9461786303073165e-06,
+      "loss": 0.1231,
+      "step": 23259
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9453853534498025e-06,
+      "loss": 0.0777,
+      "step": 23260
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9445921367392715e-06,
+      "loss": 0.1052,
+      "step": 23261
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9437989801836015e-06,
+      "loss": 0.13,
+      "step": 23262
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.943005883790669e-06,
+      "loss": 0.091,
+      "step": 23263
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9422128475683556e-06,
+      "loss": 0.1319,
+      "step": 23264
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.941419871524542e-06,
+      "loss": 0.1434,
+      "step": 23265
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.940626955667104e-06,
+      "loss": 0.1044,
+      "step": 23266
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.939834100003913e-06,
+      "loss": 0.1032,
+      "step": 23267
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.939041304542855e-06,
+      "loss": 0.1416,
+      "step": 23268
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9382485692917945e-06,
+      "loss": 0.1211,
+      "step": 23269
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.937455894258614e-06,
+      "loss": 0.1023,
+      "step": 23270
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.936663279451193e-06,
+      "loss": 0.0972,
+      "step": 23271
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.935870724877392e-06,
+      "loss": 0.0958,
+      "step": 23272
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.935078230545091e-06,
+      "loss": 0.0872,
+      "step": 23273
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.934285796462166e-06,
+      "loss": 0.0798,
+      "step": 23274
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.933493422636483e-06,
+      "loss": 0.077,
+      "step": 23275
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9327011090759206e-06,
+      "loss": 0.1009,
+      "step": 23276
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.931908855788345e-06,
+      "loss": 0.1391,
+      "step": 23277
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9311166627816225e-06,
+      "loss": 0.0979,
+      "step": 23278
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.930324530063628e-06,
+      "loss": 0.1097,
+      "step": 23279
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.929532457642238e-06,
+      "loss": 0.0835,
+      "step": 23280
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9287404455253055e-06,
+      "loss": 0.0991,
+      "step": 23281
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9279484937207045e-06,
+      "loss": 0.0971,
+      "step": 23282
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.927156602236308e-06,
+      "loss": 0.1129,
+      "step": 23283
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.926364771079976e-06,
+      "loss": 0.0848,
+      "step": 23284
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.92557300025958e-06,
+      "loss": 0.0914,
+      "step": 23285
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9247812897829815e-06,
+      "loss": 0.1014,
+      "step": 23286
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.923989639658043e-06,
+      "loss": 0.0793,
+      "step": 23287
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.923198049892634e-06,
+      "loss": 0.09,
+      "step": 23288
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.922406520494618e-06,
+      "loss": 0.1198,
+      "step": 23289
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.921615051471858e-06,
+      "loss": 0.0861,
+      "step": 23290
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.920823642832211e-06,
+      "loss": 0.0962,
+      "step": 23291
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.920032294583547e-06,
+      "loss": 0.1154,
+      "step": 23292
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.919241006733718e-06,
+      "loss": 0.1522,
+      "step": 23293
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.918449779290595e-06,
+      "loss": 0.142,
+      "step": 23294
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.917658612262033e-06,
+      "loss": 0.1117,
+      "step": 23295
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.916867505655888e-06,
+      "loss": 0.1185,
+      "step": 23296
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.916076459480023e-06,
+      "loss": 0.1161,
+      "step": 23297
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.915285473742299e-06,
+      "loss": 0.1506,
+      "step": 23298
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.914494548450566e-06,
+      "loss": 0.1081,
+      "step": 23299
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9137036836126905e-06,
+      "loss": 0.1216,
+      "step": 23300
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.912912879236523e-06,
+      "loss": 0.1127,
+      "step": 23301
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.912122135329919e-06,
+      "loss": 0.0986,
+      "step": 23302
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.911331451900737e-06,
+      "loss": 0.0917,
+      "step": 23303
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.910540828956827e-06,
+      "loss": 0.1233,
+      "step": 23304
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9097502665060506e-06,
+      "loss": 0.1302,
+      "step": 23305
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.908959764556253e-06,
+      "loss": 0.1363,
+      "step": 23306
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.908169323115294e-06,
+      "loss": 0.0925,
+      "step": 23307
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.907378942191019e-06,
+      "loss": 0.0985,
+      "step": 23308
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.906588621791288e-06,
+      "loss": 0.1636,
+      "step": 23309
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.905798361923947e-06,
+      "loss": 0.083,
+      "step": 23310
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.905008162596845e-06,
+      "loss": 0.1196,
+      "step": 23311
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.904218023817836e-06,
+      "loss": 0.1328,
+      "step": 23312
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.903427945594765e-06,
+      "loss": 0.0846,
+      "step": 23313
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.902637927935487e-06,
+      "loss": 0.1137,
+      "step": 23314
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.901847970847842e-06,
+      "loss": 0.0984,
+      "step": 23315
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.901058074339684e-06,
+      "loss": 0.1121,
+      "step": 23316
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.900268238418856e-06,
+      "loss": 0.0964,
+      "step": 23317
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.899478463093208e-06,
+      "loss": 0.1092,
+      "step": 23318
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.898688748370585e-06,
+      "loss": 0.0923,
+      "step": 23319
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.897899094258826e-06,
+      "loss": 0.1492,
+      "step": 23320
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8971095007657835e-06,
+      "loss": 0.1367,
+      "step": 23321
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8963199678992944e-06,
+      "loss": 0.1227,
+      "step": 23322
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8955304956672066e-06,
+      "loss": 0.1162,
+      "step": 23323
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.894741084077365e-06,
+      "loss": 0.1073,
+      "step": 23324
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.893951733137608e-06,
+      "loss": 0.0988,
+      "step": 23325
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.893162442855774e-06,
+      "loss": 0.1137,
+      "step": 23326
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.892373213239711e-06,
+      "loss": 0.1096,
+      "step": 23327
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.891584044297252e-06,
+      "loss": 0.1106,
+      "step": 23328
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.890794936036243e-06,
+      "loss": 0.1538,
+      "step": 23329
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.890005888464521e-06,
+      "loss": 0.1251,
+      "step": 23330
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8892169015899205e-06,
+      "loss": 0.0678,
+      "step": 23331
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.888427975420283e-06,
+      "loss": 0.1431,
+      "step": 23332
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8876391099634485e-06,
+      "loss": 0.0973,
+      "step": 23333
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.88685030522725e-06,
+      "loss": 0.0973,
+      "step": 23334
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.886061561219521e-06,
+      "loss": 0.096,
+      "step": 23335
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.885272877948103e-06,
+      "loss": 0.0993,
+      "step": 23336
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8844842554208254e-06,
+      "loss": 0.1146,
+      "step": 23337
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.883695693645527e-06,
+      "loss": 0.1171,
+      "step": 23338
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.882907192630041e-06,
+      "loss": 0.1297,
+      "step": 23339
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.882118752382195e-06,
+      "loss": 0.0928,
+      "step": 23340
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8813303729098236e-06,
+      "loss": 0.0793,
+      "step": 23341
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.880542054220765e-06,
+      "loss": 0.0921,
+      "step": 23342
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.879753796322845e-06,
+      "loss": 0.1459,
+      "step": 23343
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.878965599223892e-06,
+      "loss": 0.1041,
+      "step": 23344
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8781774629317415e-06,
+      "loss": 0.1099,
+      "step": 23345
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.877389387454216e-06,
+      "loss": 0.0953,
+      "step": 23346
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.876601372799152e-06,
+      "loss": 0.1065,
+      "step": 23347
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.875813418974378e-06,
+      "loss": 0.098,
+      "step": 23348
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.875025525987711e-06,
+      "loss": 0.1118,
+      "step": 23349
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.874237693846986e-06,
+      "loss": 0.1103,
+      "step": 23350
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.873449922560031e-06,
+      "loss": 0.1003,
+      "step": 23351
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8726622121346655e-06,
+      "loss": 0.1196,
+      "step": 23352
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.871874562578722e-06,
+      "loss": 0.108,
+      "step": 23353
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.871086973900021e-06,
+      "loss": 0.0938,
+      "step": 23354
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.870299446106382e-06,
+      "loss": 0.1216,
+      "step": 23355
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8695119792056345e-06,
+      "loss": 0.1183,
+      "step": 23356
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.868724573205604e-06,
+      "loss": 0.0747,
+      "step": 23357
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.867937228114107e-06,
+      "loss": 0.0869,
+      "step": 23358
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8671499439389625e-06,
+      "loss": 0.115,
+      "step": 23359
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.866362720687999e-06,
+      "loss": 0.1211,
+      "step": 23360
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8655755583690315e-06,
+      "loss": 0.1168,
+      "step": 23361
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8647884569898794e-06,
+      "loss": 0.1365,
+      "step": 23362
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.864001416558372e-06,
+      "loss": 0.1145,
+      "step": 23363
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.863214437082313e-06,
+      "loss": 0.0932,
+      "step": 23364
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.862427518569526e-06,
+      "loss": 0.0789,
+      "step": 23365
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.861640661027833e-06,
+      "loss": 0.1179,
+      "step": 23366
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8608538644650474e-06,
+      "loss": 0.0903,
+      "step": 23367
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.860067128888981e-06,
+      "loss": 0.1175,
+      "step": 23368
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.859280454307458e-06,
+      "loss": 0.0866,
+      "step": 23369
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.858493840728283e-06,
+      "loss": 0.1289,
+      "step": 23370
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.857707288159277e-06,
+      "loss": 0.1185,
+      "step": 23371
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.85692079660826e-06,
+      "loss": 0.1412,
+      "step": 23372
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8561343660830285e-06,
+      "loss": 0.1302,
+      "step": 23373
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.855347996591405e-06,
+      "loss": 0.1167,
+      "step": 23374
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.854561688141205e-06,
+      "loss": 0.1219,
+      "step": 23375
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.853775440740231e-06,
+      "loss": 0.1019,
+      "step": 23376
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8529892543963e-06,
+      "loss": 0.091,
+      "step": 23377
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.85220312911722e-06,
+      "loss": 0.104,
+      "step": 23378
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.851417064910798e-06,
+      "loss": 0.1364,
+      "step": 23379
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.850631061784843e-06,
+      "loss": 0.0765,
+      "step": 23380
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.84984511974717e-06,
+      "loss": 0.1057,
+      "step": 23381
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.849059238805581e-06,
+      "loss": 0.1041,
+      "step": 23382
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.848273418967881e-06,
+      "loss": 0.1192,
+      "step": 23383
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.847487660241882e-06,
+      "loss": 0.1078,
+      "step": 23384
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.846701962635383e-06,
+      "loss": 0.1352,
+      "step": 23385
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.845916326156198e-06,
+      "loss": 0.0982,
+      "step": 23386
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.845130750812126e-06,
+      "loss": 0.0979,
+      "step": 23387
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.844345236610968e-06,
+      "loss": 0.1016,
+      "step": 23388
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.843559783560531e-06,
+      "loss": 0.1353,
+      "step": 23389
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.842774391668622e-06,
+      "loss": 0.0997,
+      "step": 23390
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.841989060943038e-06,
+      "loss": 0.1079,
+      "step": 23391
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.841203791391579e-06,
+      "loss": 0.1268,
+      "step": 23392
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.840418583022051e-06,
+      "loss": 0.1128,
+      "step": 23393
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.839633435842248e-06,
+      "loss": 0.1249,
+      "step": 23394
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.838848349859977e-06,
+      "loss": 0.0985,
+      "step": 23395
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.838063325083034e-06,
+      "loss": 0.0995,
+      "step": 23396
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.837278361519214e-06,
+      "loss": 0.0858,
+      "step": 23397
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8364934591763155e-06,
+      "loss": 0.1017,
+      "step": 23398
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8357086180621435e-06,
+      "loss": 0.0954,
+      "step": 23399
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.834923838184487e-06,
+      "loss": 0.1367,
+      "step": 23400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.834139119551146e-06,
+      "loss": 0.0856,
+      "step": 23401
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.833354462169914e-06,
+      "loss": 0.1308,
+      "step": 23402
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.832569866048583e-06,
+      "loss": 0.1184,
+      "step": 23403
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8317853311949545e-06,
+      "loss": 0.0859,
+      "step": 23404
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.831000857616815e-06,
+      "loss": 0.104,
+      "step": 23405
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.830216445321963e-06,
+      "loss": 0.1407,
+      "step": 23406
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.829432094318185e-06,
+      "loss": 0.1345,
+      "step": 23407
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.82864780461328e-06,
+      "loss": 0.1025,
+      "step": 23408
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.827863576215031e-06,
+      "loss": 0.1728,
+      "step": 23409
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.827079409131239e-06,
+      "loss": 0.0973,
+      "step": 23410
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8262953033696855e-06,
+      "loss": 0.0929,
+      "step": 23411
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.82551125893816e-06,
+      "loss": 0.1321,
+      "step": 23412
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.824727275844458e-06,
+      "loss": 0.1169,
+      "step": 23413
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.823943354096359e-06,
+      "loss": 0.1153,
+      "step": 23414
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.823159493701654e-06,
+      "loss": 0.159,
+      "step": 23415
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.822375694668135e-06,
+      "loss": 0.112,
+      "step": 23416
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.821591957003583e-06,
+      "loss": 0.0996,
+      "step": 23417
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.820808280715782e-06,
+      "loss": 0.1402,
+      "step": 23418
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.820024665812524e-06,
+      "loss": 0.1703,
+      "step": 23419
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8192411123015885e-06,
+      "loss": 0.1232,
+      "step": 23420
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.818457620190758e-06,
+      "loss": 0.129,
+      "step": 23421
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.817674189487819e-06,
+      "loss": 0.1222,
+      "step": 23422
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.816890820200552e-06,
+      "loss": 0.0812,
+      "step": 23423
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.816107512336739e-06,
+      "loss": 0.1315,
+      "step": 23424
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.815324265904169e-06,
+      "loss": 0.1204,
+      "step": 23425
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.814541080910611e-06,
+      "loss": 0.1038,
+      "step": 23426
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8137579573638494e-06,
+      "loss": 0.1099,
+      "step": 23427
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8129748952716696e-06,
+      "loss": 0.1087,
+      "step": 23428
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.812191894641841e-06,
+      "loss": 0.1228,
+      "step": 23429
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.811408955482152e-06,
+      "loss": 0.1259,
+      "step": 23430
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8106260778003754e-06,
+      "loss": 0.1375,
+      "step": 23431
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.809843261604285e-06,
+      "loss": 0.0726,
+      "step": 23432
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8090605069016596e-06,
+      "loss": 0.1316,
+      "step": 23433
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8082778137002806e-06,
+      "loss": 0.0851,
+      "step": 23434
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8074951820079186e-06,
+      "loss": 0.1046,
+      "step": 23435
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.806712611832345e-06,
+      "loss": 0.1238,
+      "step": 23436
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8059301031813424e-06,
+      "loss": 0.116,
+      "step": 23437
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8051476560626744e-06,
+      "loss": 0.0929,
+      "step": 23438
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8043652704841195e-06,
+      "loss": 0.0892,
+      "step": 23439
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8035829464534566e-06,
+      "loss": 0.113,
+      "step": 23440
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8028006839784436e-06,
+      "loss": 0.1326,
+      "step": 23441
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8020184830668584e-06,
+      "loss": 0.1416,
+      "step": 23442
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8012363437264743e-06,
+      "loss": 0.0752,
+      "step": 23443
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8004542659650554e-06,
+      "loss": 0.1217,
+      "step": 23444
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.799672249790376e-06,
+      "loss": 0.0832,
+      "step": 23445
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7988902952102025e-06,
+      "loss": 0.1188,
+      "step": 23446
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7981084022323e-06,
+      "loss": 0.1318,
+      "step": 23447
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.797326570864438e-06,
+      "loss": 0.087,
+      "step": 23448
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7965448011143903e-06,
+      "loss": 0.096,
+      "step": 23449
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79576309298991e-06,
+      "loss": 0.1321,
+      "step": 23450
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79498144649877e-06,
+      "loss": 0.1026,
+      "step": 23451
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7941998616487373e-06,
+      "loss": 0.1447,
+      "step": 23452
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79341833844757e-06,
+      "loss": 0.0893,
+      "step": 23453
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7926368769030386e-06,
+      "loss": 0.0988,
+      "step": 23454
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.791855477022903e-06,
+      "loss": 0.1299,
+      "step": 23455
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.791074138814922e-06,
+      "loss": 0.1105,
+      "step": 23456
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7902928622868605e-06,
+      "loss": 0.0969,
+      "step": 23457
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7895116474464845e-06,
+      "loss": 0.0985,
+      "step": 23458
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.78873049430155e-06,
+      "loss": 0.1151,
+      "step": 23459
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.787949402859814e-06,
+      "loss": 0.1197,
+      "step": 23460
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.787168373129042e-06,
+      "loss": 0.0623,
+      "step": 23461
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.786387405116988e-06,
+      "loss": 0.1178,
+      "step": 23462
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7856064988314144e-06,
+      "loss": 0.112,
+      "step": 23463
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7848256542800776e-06,
+      "loss": 0.1319,
+      "step": 23464
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.78404487147073e-06,
+      "loss": 0.0916,
+      "step": 23465
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7832641504111325e-06,
+      "loss": 0.1714,
+      "step": 23466
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7824834911090425e-06,
+      "loss": 0.1298,
+      "step": 23467
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.781702893572209e-06,
+      "loss": 0.1091,
+      "step": 23468
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.780922357808393e-06,
+      "loss": 0.1118,
+      "step": 23469
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.780141883825347e-06,
+      "loss": 0.108,
+      "step": 23470
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.779361471630818e-06,
+      "loss": 0.1143,
+      "step": 23471
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7785811212325674e-06,
+      "loss": 0.0765,
+      "step": 23472
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.77780083263834e-06,
+      "loss": 0.1054,
+      "step": 23473
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7770206058558943e-06,
+      "loss": 0.1475,
+      "step": 23474
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.776240440892973e-06,
+      "loss": 0.0909,
+      "step": 23475
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7754603377573352e-06,
+      "loss": 0.1143,
+      "step": 23476
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.774680296456722e-06,
+      "loss": 0.1278,
+      "step": 23477
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7739003169988897e-06,
+      "loss": 0.0824,
+      "step": 23478
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7731203993915832e-06,
+      "loss": 0.0879,
+      "step": 23479
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7723405436425474e-06,
+      "loss": 0.0841,
+      "step": 23480
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.771560749759535e-06,
+      "loss": 0.0807,
+      "step": 23481
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.770781017750288e-06,
+      "loss": 0.0926,
+      "step": 23482
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.770001347622557e-06,
+      "loss": 0.1441,
+      "step": 23483
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.769221739384081e-06,
+      "loss": 0.1129,
+      "step": 23484
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.768442193042612e-06,
+      "loss": 0.0993,
+      "step": 23485
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.767662708605887e-06,
+      "loss": 0.1142,
+      "step": 23486
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.766883286081656e-06,
+      "loss": 0.1117,
+      "step": 23487
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.766103925477659e-06,
+      "loss": 0.089,
+      "step": 23488
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.765324626801634e-06,
+      "loss": 0.0921,
+      "step": 23489
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7645453900613305e-06,
+      "loss": 0.1111,
+      "step": 23490
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7637662152644816e-06,
+      "loss": 0.1072,
+      "step": 23491
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7629871024188313e-06,
+      "loss": 0.1561,
+      "step": 23492
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7622080515321237e-06,
+      "loss": 0.0933,
+      "step": 23493
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.761429062612093e-06,
+      "loss": 0.1446,
+      "step": 23494
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.760650135666476e-06,
+      "loss": 0.1287,
+      "step": 23495
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.759871270703016e-06,
+      "loss": 0.1022,
+      "step": 23496
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.759092467729445e-06,
+      "loss": 0.0938,
+      "step": 23497
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7583137267535054e-06,
+      "loss": 0.0896,
+      "step": 23498
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.757535047782929e-06,
+      "loss": 0.1538,
+      "step": 23499
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.75675643082545e-06,
+      "loss": 0.1425,
+      "step": 23500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7559778758888057e-06,
+      "loss": 0.0915,
+      "step": 23501
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.755199382980732e-06,
+      "loss": 0.0679,
+      "step": 23502
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7544209521089625e-06,
+      "loss": 0.1096,
+      "step": 23503
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.753642583281224e-06,
+      "loss": 0.1016,
+      "step": 23504
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.752864276505257e-06,
+      "loss": 0.0874,
+      "step": 23505
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.752086031788786e-06,
+      "loss": 0.0869,
+      "step": 23506
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7513078491395495e-06,
+      "loss": 0.1105,
+      "step": 23507
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.750529728565273e-06,
+      "loss": 0.0836,
+      "step": 23508
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7497516700736846e-06,
+      "loss": 0.1216,
+      "step": 23509
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.748973673672518e-06,
+      "loss": 0.1157,
+      "step": 23510
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7481957393695023e-06,
+      "loss": 0.0887,
+      "step": 23511
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7474178671723636e-06,
+      "loss": 0.0834,
+      "step": 23512
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7466400570888273e-06,
+      "loss": 0.1237,
+      "step": 23513
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7458623091266246e-06,
+      "loss": 0.1067,
+      "step": 23514
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7450846232934768e-06,
+      "loss": 0.0707,
+      "step": 23515
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7443069995971116e-06,
+      "loss": 0.1207,
+      "step": 23516
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7435294380452613e-06,
+      "loss": 0.1442,
+      "step": 23517
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.742751938645637e-06,
+      "loss": 0.1097,
+      "step": 23518
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7419745014059694e-06,
+      "loss": 0.1129,
+      "step": 23519
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.741197126333983e-06,
+      "loss": 0.1553,
+      "step": 23520
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7404198134373972e-06,
+      "loss": 0.1188,
+      "step": 23521
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7396425627239364e-06,
+      "loss": 0.1332,
+      "step": 23522
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7388653742013214e-06,
+      "loss": 0.1071,
+      "step": 23523
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7380882478772683e-06,
+      "loss": 0.1377,
+      "step": 23524
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.737311183759502e-06,
+      "loss": 0.09,
+      "step": 23525
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7365341818557433e-06,
+      "loss": 0.1175,
+      "step": 23526
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7357572421737087e-06,
+      "loss": 0.0985,
+      "step": 23527
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7349803647211123e-06,
+      "loss": 0.0868,
+      "step": 23528
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7342035495056793e-06,
+      "loss": 0.0908,
+      "step": 23529
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.733426796535119e-06,
+      "loss": 0.1059,
+      "step": 23530
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7326501058171518e-06,
+      "loss": 0.1159,
+      "step": 23531
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7318734773595e-06,
+      "loss": 0.1239,
+      "step": 23532
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.731096911169865e-06,
+      "loss": 0.0915,
+      "step": 23533
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7303204072559684e-06,
+      "loss": 0.1125,
+      "step": 23534
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.729543965625526e-06,
+      "loss": 0.0884,
+      "step": 23535
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7287675862862503e-06,
+      "loss": 0.1144,
+      "step": 23536
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7279912692458476e-06,
+      "loss": 0.0958,
+      "step": 23537
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7272150145120387e-06,
+      "loss": 0.1286,
+      "step": 23538
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7264388220925286e-06,
+      "loss": 0.1137,
+      "step": 23539
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.725662691995029e-06,
+      "loss": 0.0719,
+      "step": 23540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7248866242272584e-06,
+      "loss": 0.1192,
+      "step": 23541
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.724110618796912e-06,
+      "loss": 0.1474,
+      "step": 23542
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7233346757117073e-06,
+      "loss": 0.1263,
+      "step": 23543
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.722558794979354e-06,
+      "loss": 0.1484,
+      "step": 23544
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.721782976607554e-06,
+      "loss": 0.0929,
+      "step": 23545
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7210072206040204e-06,
+      "loss": 0.1258,
+      "step": 23546
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.720231526976457e-06,
+      "loss": 0.0823,
+      "step": 23547
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7194558957325657e-06,
+      "loss": 0.0804,
+      "step": 23548
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7186803268800554e-06,
+      "loss": 0.1004,
+      "step": 23549
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7179048204266333e-06,
+      "loss": 0.1636,
+      "step": 23550
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.717129376380001e-06,
+      "loss": 0.1605,
+      "step": 23551
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7163539947478567e-06,
+      "loss": 0.1294,
+      "step": 23552
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7155786755379132e-06,
+      "loss": 0.1407,
+      "step": 23553
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.714803418757862e-06,
+      "loss": 0.1199,
+      "step": 23554
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7140282244154136e-06,
+      "loss": 0.1391,
+      "step": 23555
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.713253092518264e-06,
+      "loss": 0.152,
+      "step": 23556
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.712478023074112e-06,
+      "loss": 0.1375,
+      "step": 23557
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.71170301609066e-06,
+      "loss": 0.0761,
+      "step": 23558
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.710928071575608e-06,
+      "loss": 0.0644,
+      "step": 23559
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.710153189536654e-06,
+      "loss": 0.0916,
+      "step": 23560
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.709378369981491e-06,
+      "loss": 0.0982,
+      "step": 23561
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.708603612917823e-06,
+      "loss": 0.073,
+      "step": 23562
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.70782891835334e-06,
+      "loss": 0.102,
+      "step": 23563
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7070542862957437e-06,
+      "loss": 0.0985,
+      "step": 23564
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7062797167527274e-06,
+      "loss": 0.1273,
+      "step": 23565
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.705505209731981e-06,
+      "loss": 0.121,
+      "step": 23566
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7047307652412036e-06,
+      "loss": 0.1146,
+      "step": 23567
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7039563832880898e-06,
+      "loss": 0.1183,
+      "step": 23568
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7031820638803275e-06,
+      "loss": 0.0975,
+      "step": 23569
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7024078070256142e-06,
+      "loss": 0.102,
+      "step": 23570
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.701633612731639e-06,
+      "loss": 0.0908,
+      "step": 23571
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7008594810060884e-06,
+      "loss": 0.0884,
+      "step": 23572
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.700085411856661e-06,
+      "loss": 0.1193,
+      "step": 23573
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6993114052910385e-06,
+      "loss": 0.1084,
+      "step": 23574
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6985374613169166e-06,
+      "loss": 0.1093,
+      "step": 23575
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6977635799419774e-06,
+      "loss": 0.1395,
+      "step": 23576
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6969897611739157e-06,
+      "loss": 0.0681,
+      "step": 23577
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.696216005020411e-06,
+      "loss": 0.086,
+      "step": 23578
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.695442311489158e-06,
+      "loss": 0.0987,
+      "step": 23579
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6946686805878386e-06,
+      "loss": 0.1194,
+      "step": 23580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.693895112324134e-06,
+      "loss": 0.1034,
+      "step": 23581
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.693121606705736e-06,
+      "loss": 0.0972,
+      "step": 23582
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6923481637403234e-06,
+      "loss": 0.1067,
+      "step": 23583
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.691574783435581e-06,
+      "loss": 0.1097,
+      "step": 23584
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6908014657991985e-06,
+      "loss": 0.1127,
+      "step": 23585
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6900282108388462e-06,
+      "loss": 0.107,
+      "step": 23586
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.689255018562211e-06,
+      "loss": 0.0787,
+      "step": 23587
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.688481888976978e-06,
+      "loss": 0.0929,
+      "step": 23588
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.687708822090823e-06,
+      "loss": 0.0967,
+      "step": 23589
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6869358179114236e-06,
+      "loss": 0.1184,
+      "step": 23590
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.686162876446465e-06,
+      "loss": 0.098,
+      "step": 23591
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.68538999770362e-06,
+      "loss": 0.1275,
+      "step": 23592
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6846171816905684e-06,
+      "loss": 0.1195,
+      "step": 23593
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6838444284149945e-06,
+      "loss": 0.124,
+      "step": 23594
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6830717378845614e-06,
+      "loss": 0.0999,
+      "step": 23595
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6822991101069517e-06,
+      "loss": 0.1128,
+      "step": 23596
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.681526545089845e-06,
+      "loss": 0.1303,
+      "step": 23597
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6807540428409084e-06,
+      "loss": 0.0904,
+      "step": 23598
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6799816033678226e-06,
+      "loss": 0.0998,
+      "step": 23599
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6792092266782587e-06,
+      "loss": 0.1212,
+      "step": 23600
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.678436912779886e-06,
+      "loss": 0.1266,
+      "step": 23601
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.677664661680378e-06,
+      "loss": 0.1093,
+      "step": 23602
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.676892473387412e-06,
+      "loss": 0.1126,
+      "step": 23603
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.676120347908656e-06,
+      "loss": 0.0794,
+      "step": 23604
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.675348285251774e-06,
+      "loss": 0.0913,
+      "step": 23605
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.674576285424446e-06,
+      "loss": 0.1067,
+      "step": 23606
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6738043484343322e-06,
+      "loss": 0.1071,
+      "step": 23607
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6730324742891043e-06,
+      "loss": 0.1036,
+      "step": 23608
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6722606629964388e-06,
+      "loss": 0.1072,
+      "step": 23609
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6714889145639877e-06,
+      "loss": 0.1004,
+      "step": 23610
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.670717228999424e-06,
+      "loss": 0.1189,
+      "step": 23611
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6699456063104188e-06,
+      "loss": 0.1084,
+      "step": 23612
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.66917404650463e-06,
+      "loss": 0.1108,
+      "step": 23613
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6684025495897278e-06,
+      "loss": 0.1075,
+      "step": 23614
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6676311155733745e-06,
+      "loss": 0.096,
+      "step": 23615
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6668597444632305e-06,
+      "loss": 0.0859,
+      "step": 23616
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.66608843626696e-06,
+      "loss": 0.0548,
+      "step": 23617
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.665317190992235e-06,
+      "loss": 0.1248,
+      "step": 23618
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.664546008646701e-06,
+      "loss": 0.1017,
+      "step": 23619
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6637748892380267e-06,
+      "loss": 0.1304,
+      "step": 23620
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.663003832773875e-06,
+      "loss": 0.1256,
+      "step": 23621
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6622328392619e-06,
+      "loss": 0.1261,
+      "step": 23622
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.661461908709767e-06,
+      "loss": 0.0728,
+      "step": 23623
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6606910411251317e-06,
+      "loss": 0.1148,
+      "step": 23624
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6599202365156483e-06,
+      "loss": 0.1189,
+      "step": 23625
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.659149494888976e-06,
+      "loss": 0.1133,
+      "step": 23626
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.658378816252777e-06,
+      "loss": 0.1147,
+      "step": 23627
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.657608200614703e-06,
+      "loss": 0.0801,
+      "step": 23628
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.656837647982405e-06,
+      "loss": 0.1042,
+      "step": 23629
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6560671583635467e-06,
+      "loss": 0.1225,
+      "step": 23630
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6552967317657738e-06,
+      "loss": 0.1072,
+      "step": 23631
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.654526368196747e-06,
+      "loss": 0.1001,
+      "step": 23632
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.653756067664115e-06,
+      "loss": 0.1126,
+      "step": 23633
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6529858301755273e-06,
+      "loss": 0.1129,
+      "step": 23634
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.652215655738639e-06,
+      "loss": 0.1165,
+      "step": 23635
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.651445544361104e-06,
+      "loss": 0.1363,
+      "step": 23636
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6506754960505663e-06,
+      "loss": 0.0787,
+      "step": 23637
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.649905510814683e-06,
+      "loss": 0.1371,
+      "step": 23638
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.649135588661098e-06,
+      "loss": 0.1,
+      "step": 23639
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6483657295974583e-06,
+      "loss": 0.0955,
+      "step": 23640
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.647595933631417e-06,
+      "loss": 0.1129,
+      "step": 23641
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6468262007706155e-06,
+      "loss": 0.104,
+      "step": 23642
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.646056531022707e-06,
+      "loss": 0.0955,
+      "step": 23643
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6452869243953302e-06,
+      "loss": 0.0947,
+      "step": 23644
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6445173808961375e-06,
+      "loss": 0.0935,
+      "step": 23645
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.643747900532767e-06,
+      "loss": 0.1137,
+      "step": 23646
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.642978483312869e-06,
+      "loss": 0.1011,
+      "step": 23647
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.642209129244084e-06,
+      "loss": 0.122,
+      "step": 23648
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6414398383340523e-06,
+      "loss": 0.0956,
+      "step": 23649
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.640670610590421e-06,
+      "loss": 0.08,
+      "step": 23650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.639901446020826e-06,
+      "loss": 0.1044,
+      "step": 23651
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6391323446329153e-06,
+      "loss": 0.0946,
+      "step": 23652
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6383633064343215e-06,
+      "loss": 0.1524,
+      "step": 23653
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.637594331432691e-06,
+      "loss": 0.1095,
+      "step": 23654
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6368254196356576e-06,
+      "loss": 0.113,
+      "step": 23655
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6360565710508655e-06,
+      "loss": 0.0926,
+      "step": 23656
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.635287785685949e-06,
+      "loss": 0.0881,
+      "step": 23657
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.634519063548542e-06,
+      "loss": 0.1206,
+      "step": 23658
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.633750404646287e-06,
+      "loss": 0.088,
+      "step": 23659
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6329818089868153e-06,
+      "loss": 0.1085,
+      "step": 23660
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.632213276577763e-06,
+      "loss": 0.1019,
+      "step": 23661
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6314448074267704e-06,
+      "loss": 0.0828,
+      "step": 23662
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6306764015414663e-06,
+      "loss": 0.1313,
+      "step": 23663
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6299080589294823e-06,
+      "loss": 0.0833,
+      "step": 23664
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6291397795984564e-06,
+      "loss": 0.1025,
+      "step": 23665
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.628371563556016e-06,
+      "loss": 0.1141,
+      "step": 23666
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6276034108097966e-06,
+      "loss": 0.091,
+      "step": 23667
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.626835321367428e-06,
+      "loss": 0.0984,
+      "step": 23668
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.626067295236536e-06,
+      "loss": 0.1034,
+      "step": 23669
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6252993324247544e-06,
+      "loss": 0.1504,
+      "step": 23670
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.624531432939715e-06,
+      "loss": 0.125,
+      "step": 23671
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.623763596789043e-06,
+      "loss": 0.1063,
+      "step": 23672
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.622995823980362e-06,
+      "loss": 0.1055,
+      "step": 23673
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.622228114521307e-06,
+      "loss": 0.1343,
+      "step": 23674
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6214604684194974e-06,
+      "loss": 0.1149,
+      "step": 23675
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6206928856825652e-06,
+      "loss": 0.1053,
+      "step": 23676
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6199253663181335e-06,
+      "loss": 0.1168,
+      "step": 23677
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6191579103338216e-06,
+      "loss": 0.0865,
+      "step": 23678
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.618390517737258e-06,
+      "loss": 0.0967,
+      "step": 23679
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6176231885360704e-06,
+      "loss": 0.1102,
+      "step": 23680
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6168559227378763e-06,
+      "loss": 0.13,
+      "step": 23681
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.616088720350295e-06,
+      "loss": 0.1256,
+      "step": 23682
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6153215813809562e-06,
+      "loss": 0.0841,
+      "step": 23683
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.614554505837471e-06,
+      "loss": 0.0995,
+      "step": 23684
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6137874937274652e-06,
+      "loss": 0.0895,
+      "step": 23685
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6130205450585654e-06,
+      "loss": 0.1042,
+      "step": 23686
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6122536598383753e-06,
+      "loss": 0.1091,
+      "step": 23687
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6114868380745207e-06,
+      "loss": 0.1255,
+      "step": 23688
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6107200797746223e-06,
+      "loss": 0.0828,
+      "step": 23689
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.609953384946292e-06,
+      "loss": 0.1159,
+      "step": 23690
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6091867535971504e-06,
+      "loss": 0.1328,
+      "step": 23691
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6084201857348124e-06,
+      "loss": 0.077,
+      "step": 23692
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6076536813668883e-06,
+      "loss": 0.0963,
+      "step": 23693
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.606887240500996e-06,
+      "loss": 0.1161,
+      "step": 23694
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.606120863144753e-06,
+      "loss": 0.1022,
+      "step": 23695
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6053545493057696e-06,
+      "loss": 0.1189,
+      "step": 23696
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6045882989916535e-06,
+      "loss": 0.1136,
+      "step": 23697
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.603822112210026e-06,
+      "loss": 0.0887,
+      "step": 23698
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.60305598896849e-06,
+      "loss": 0.1064,
+      "step": 23699
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6022899292746593e-06,
+      "loss": 0.0781,
+      "step": 23700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.601523933136153e-06,
+      "loss": 0.1055,
+      "step": 23701
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6007580005605634e-06,
+      "loss": 0.1516,
+      "step": 23702
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.59999213155551e-06,
+      "loss": 0.1086,
+      "step": 23703
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.599226326128601e-06,
+      "loss": 0.1167,
+      "step": 23704
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.598460584287442e-06,
+      "loss": 0.151,
+      "step": 23705
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.597694906039637e-06,
+      "loss": 0.126,
+      "step": 23706
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5969292913927987e-06,
+      "loss": 0.0906,
+      "step": 23707
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.596163740354526e-06,
+      "loss": 0.1251,
+      "step": 23708
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5953982529324273e-06,
+      "loss": 0.1398,
+      "step": 23709
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5946328291341135e-06,
+      "loss": 0.1093,
+      "step": 23710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.593867468967176e-06,
+      "loss": 0.1385,
+      "step": 23711
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5931021724392236e-06,
+      "loss": 0.0988,
+      "step": 23712
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.592336939557861e-06,
+      "loss": 0.0901,
+      "step": 23713
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5915717703306862e-06,
+      "loss": 0.088,
+      "step": 23714
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.590806664765305e-06,
+      "loss": 0.0875,
+      "step": 23715
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5900416228693158e-06,
+      "loss": 0.0943,
+      "step": 23716
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5892766446503147e-06,
+      "loss": 0.0713,
+      "step": 23717
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5885117301159046e-06,
+      "loss": 0.1101,
+      "step": 23718
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.587746879273688e-06,
+      "loss": 0.1007,
+      "step": 23719
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5869820921312583e-06,
+      "loss": 0.0879,
+      "step": 23720
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.586217368696211e-06,
+      "loss": 0.0914,
+      "step": 23721
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.585452708976148e-06,
+      "loss": 0.1106,
+      "step": 23722
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5846881129786604e-06,
+      "loss": 0.0922,
+      "step": 23723
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5839235807113502e-06,
+      "loss": 0.1184,
+      "step": 23724
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5831591121818086e-06,
+      "loss": 0.1128,
+      "step": 23725
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5823947073976262e-06,
+      "loss": 0.1088,
+      "step": 23726
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.581630366366401e-06,
+      "loss": 0.0919,
+      "step": 23727
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5808660890957283e-06,
+      "loss": 0.1238,
+      "step": 23728
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.580101875593197e-06,
+      "loss": 0.1101,
+      "step": 23729
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5793377258663963e-06,
+      "loss": 0.1052,
+      "step": 23730
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5785736399229243e-06,
+      "loss": 0.1258,
+      "step": 23731
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5778096177703645e-06,
+      "loss": 0.13,
+      "step": 23732
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.577045659416313e-06,
+      "loss": 0.104,
+      "step": 23733
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5762817648683556e-06,
+      "loss": 0.0894,
+      "step": 23734
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5755179341340772e-06,
+      "loss": 0.1184,
+      "step": 23735
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5747541672210717e-06,
+      "loss": 0.1122,
+      "step": 23736
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.573990464136926e-06,
+      "loss": 0.1469,
+      "step": 23737
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5732268248892222e-06,
+      "loss": 0.1082,
+      "step": 23738
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5724632494855537e-06,
+      "loss": 0.0864,
+      "step": 23739
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5716997379335004e-06,
+      "loss": 0.1112,
+      "step": 23740
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5709362902406464e-06,
+      "loss": 0.1227,
+      "step": 23741
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5701729064145798e-06,
+      "loss": 0.1063,
+      "step": 23742
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5694095864628784e-06,
+      "loss": 0.1078,
+      "step": 23743
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5686463303931327e-06,
+      "loss": 0.092,
+      "step": 23744
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.567883138212922e-06,
+      "loss": 0.1155,
+      "step": 23745
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.567120009929822e-06,
+      "loss": 0.1153,
+      "step": 23746
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.566356945551419e-06,
+      "loss": 0.1231,
+      "step": 23747
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.565593945085296e-06,
+      "loss": 0.1095,
+      "step": 23748
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.56483100853903e-06,
+      "loss": 0.0836,
+      "step": 23749
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5640681359201954e-06,
+      "loss": 0.118,
+      "step": 23750
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.563305327236379e-06,
+      "loss": 0.1296,
+      "step": 23751
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5625425824951508e-06,
+      "loss": 0.1045,
+      "step": 23752
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.561779901704092e-06,
+      "loss": 0.1224,
+      "step": 23753
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5610172848707858e-06,
+      "loss": 0.0856,
+      "step": 23754
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5602547320027937e-06,
+      "loss": 0.0885,
+      "step": 23755
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5594922431076984e-06,
+      "loss": 0.0912,
+      "step": 23756
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5587298181930772e-06,
+      "loss": 0.0951,
+      "step": 23757
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5579674572665025e-06,
+      "loss": 0.1271,
+      "step": 23758
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.557205160335543e-06,
+      "loss": 0.0947,
+      "step": 23759
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5564429274077772e-06,
+      "loss": 0.1152,
+      "step": 23760
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5556807584907717e-06,
+      "loss": 0.0719,
+      "step": 23761
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.554918653592101e-06,
+      "loss": 0.1248,
+      "step": 23762
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.554156612719344e-06,
+      "loss": 0.0849,
+      "step": 23763
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5533946358800552e-06,
+      "loss": 0.1333,
+      "step": 23764
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5526327230818114e-06,
+      "loss": 0.138,
+      "step": 23765
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.551870874332185e-06,
+      "loss": 0.0871,
+      "step": 23766
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.551109089638739e-06,
+      "loss": 0.125,
+      "step": 23767
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5503473690090462e-06,
+      "loss": 0.097,
+      "step": 23768
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5495857124506704e-06,
+      "loss": 0.1075,
+      "step": 23769
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.548824119971174e-06,
+      "loss": 0.0846,
+      "step": 23770
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5480625915781274e-06,
+      "loss": 0.0738,
+      "step": 23771
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5473011272790993e-06,
+      "loss": 0.1401,
+      "step": 23772
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5465397270816494e-06,
+      "loss": 0.1135,
+      "step": 23773
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5457783909933373e-06,
+      "loss": 0.1449,
+      "step": 23774
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5450171190217364e-06,
+      "loss": 0.1062,
+      "step": 23775
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.544255911174399e-06,
+      "loss": 0.114,
+      "step": 23776
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5434947674588926e-06,
+      "loss": 0.0662,
+      "step": 23777
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5427336878827836e-06,
+      "loss": 0.0993,
+      "step": 23778
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5419726724536197e-06,
+      "loss": 0.1252,
+      "step": 23779
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.541211721178969e-06,
+      "loss": 0.0877,
+      "step": 23780
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5404508340663925e-06,
+      "loss": 0.0896,
+      "step": 23781
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5396900111234424e-06,
+      "loss": 0.0872,
+      "step": 23782
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5389292523576833e-06,
+      "loss": 0.1354,
+      "step": 23783
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5381685577766712e-06,
+      "loss": 0.1148,
+      "step": 23784
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.537407927387957e-06,
+      "loss": 0.1146,
+      "step": 23785
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5366473611991014e-06,
+      "loss": 0.1271,
+      "step": 23786
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5358868592176665e-06,
+      "loss": 0.1178,
+      "step": 23787
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.535126421451194e-06,
+      "loss": 0.099,
+      "step": 23788
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.534366047907245e-06,
+      "loss": 0.1043,
+      "step": 23789
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.533605738593375e-06,
+      "loss": 0.0717,
+      "step": 23790
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.532845493517133e-06,
+      "loss": 0.1166,
+      "step": 23791
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5320853126860756e-06,
+      "loss": 0.076,
+      "step": 23792
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5313251961077522e-06,
+      "loss": 0.1208,
+      "step": 23793
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.530565143789709e-06,
+      "loss": 0.0944,
+      "step": 23794
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.529805155739502e-06,
+      "loss": 0.0973,
+      "step": 23795
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.529045231964684e-06,
+      "loss": 0.1213,
+      "step": 23796
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5282853724727994e-06,
+      "loss": 0.14,
+      "step": 23797
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5275255772713945e-06,
+      "loss": 0.146,
+      "step": 23798
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.526765846368022e-06,
+      "loss": 0.1287,
+      "step": 23799
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.526006179770226e-06,
+      "loss": 0.106,
+      "step": 23800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.525246577485556e-06,
+      "loss": 0.1252,
+      "step": 23801
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5244870395215567e-06,
+      "loss": 0.1302,
+      "step": 23802
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5237275658857705e-06,
+      "loss": 0.097,
+      "step": 23803
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5229681565857433e-06,
+      "loss": 0.099,
+      "step": 23804
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5222088116290243e-06,
+      "loss": 0.1173,
+      "step": 23805
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.52144953102315e-06,
+      "loss": 0.0685,
+      "step": 23806
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.520690314775669e-06,
+      "loss": 0.1388,
+      "step": 23807
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.51993116289412e-06,
+      "loss": 0.1153,
+      "step": 23808
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5191720753860413e-06,
+      "loss": 0.1172,
+      "step": 23809
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.518413052258982e-06,
+      "loss": 0.113,
+      "step": 23810
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.517654093520474e-06,
+      "loss": 0.0778,
+      "step": 23811
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.516895199178063e-06,
+      "loss": 0.1329,
+      "step": 23812
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5161363692392824e-06,
+      "loss": 0.1161,
+      "step": 23813
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5153776037116784e-06,
+      "loss": 0.1071,
+      "step": 23814
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5146189026027787e-06,
+      "loss": 0.1155,
+      "step": 23815
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5138602659201295e-06,
+      "loss": 0.1128,
+      "step": 23816
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5131016936712626e-06,
+      "loss": 0.1124,
+      "step": 23817
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5123431858637103e-06,
+      "loss": 0.0768,
+      "step": 23818
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5115847425050153e-06,
+      "loss": 0.1348,
+      "step": 23819
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.510826363602705e-06,
+      "loss": 0.1206,
+      "step": 23820
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.510068049164319e-06,
+      "loss": 0.1096,
+      "step": 23821
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.509309799197386e-06,
+      "loss": 0.1061,
+      "step": 23822
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.508551613709442e-06,
+      "loss": 0.1263,
+      "step": 23823
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5077934927080147e-06,
+      "loss": 0.1034,
+      "step": 23824
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.507035436200642e-06,
+      "loss": 0.1134,
+      "step": 23825
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.506277444194849e-06,
+      "loss": 0.1631,
+      "step": 23826
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.505519516698165e-06,
+      "loss": 0.1046,
+      "step": 23827
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.504761653718123e-06,
+      "loss": 0.1149,
+      "step": 23828
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.504003855262248e-06,
+      "loss": 0.1702,
+      "step": 23829
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5032461213380687e-06,
+      "loss": 0.0885,
+      "step": 23830
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5024884519531175e-06,
+      "loss": 0.1257,
+      "step": 23831
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.501730847114918e-06,
+      "loss": 0.1207,
+      "step": 23832
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5009733068309926e-06,
+      "loss": 0.1362,
+      "step": 23833
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.500215831108873e-06,
+      "loss": 0.0951,
+      "step": 23834
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4994584199560767e-06,
+      "loss": 0.1294,
+      "step": 23835
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.498701073380136e-06,
+      "loss": 0.1156,
+      "step": 23836
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4979437913885706e-06,
+      "loss": 0.0883,
+      "step": 23837
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4971865739889e-06,
+      "loss": 0.1145,
+      "step": 23838
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4964294211886495e-06,
+      "loss": 0.1129,
+      "step": 23839
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4956723329953444e-06,
+      "loss": 0.0902,
+      "step": 23840
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4949153094165023e-06,
+      "loss": 0.1072,
+      "step": 23841
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.494158350459641e-06,
+      "loss": 0.116,
+      "step": 23842
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.493401456132285e-06,
+      "loss": 0.0964,
+      "step": 23843
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4926446264419476e-06,
+      "loss": 0.0879,
+      "step": 23844
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.491887861396154e-06,
+      "loss": 0.0909,
+      "step": 23845
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.491131161002418e-06,
+      "loss": 0.0869,
+      "step": 23846
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4903745252682552e-06,
+      "loss": 0.1344,
+      "step": 23847
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.489617954201183e-06,
+      "loss": 0.1151,
+      "step": 23848
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4888614478087223e-06,
+      "loss": 0.1033,
+      "step": 23849
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4881050060983844e-06,
+      "loss": 0.0849,
+      "step": 23850
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4873486290776803e-06,
+      "loss": 0.1166,
+      "step": 23851
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4865923167541307e-06,
+      "loss": 0.0889,
+      "step": 23852
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4858360691352424e-06,
+      "loss": 0.0947,
+      "step": 23853
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4850798862285307e-06,
+      "loss": 0.1309,
+      "step": 23854
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4843237680415153e-06,
+      "loss": 0.1256,
+      "step": 23855
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.483567714581694e-06,
+      "loss": 0.1274,
+      "step": 23856
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.482811725856583e-06,
+      "loss": 0.098,
+      "step": 23857
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4820558018736973e-06,
+      "loss": 0.1105,
+      "step": 23858
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.481299942640538e-06,
+      "loss": 0.0857,
+      "step": 23859
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.480544148164621e-06,
+      "loss": 0.1158,
+      "step": 23860
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4797884184534526e-06,
+      "loss": 0.0882,
+      "step": 23861
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.479032753514535e-06,
+      "loss": 0.104,
+      "step": 23862
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4782771533553793e-06,
+      "loss": 0.1299,
+      "step": 23863
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4775216179834947e-06,
+      "loss": 0.0863,
+      "step": 23864
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.476766147406384e-06,
+      "loss": 0.1042,
+      "step": 23865
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.476010741631548e-06,
+      "loss": 0.0897,
+      "step": 23866
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.475255400666497e-06,
+      "loss": 0.1527,
+      "step": 23867
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4745001245187306e-06,
+      "loss": 0.085,
+      "step": 23868
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4737449131957523e-06,
+      "loss": 0.0957,
+      "step": 23869
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4729897667050727e-06,
+      "loss": 0.1365,
+      "step": 23870
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4722346850541787e-06,
+      "loss": 0.0987,
+      "step": 23871
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4714796682505804e-06,
+      "loss": 0.1328,
+      "step": 23872
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4707247163017787e-06,
+      "loss": 0.0714,
+      "step": 23873
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.469969829215273e-06,
+      "loss": 0.0806,
+      "step": 23874
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4692150069985563e-06,
+      "loss": 0.0832,
+      "step": 23875
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.468460249659135e-06,
+      "loss": 0.1506,
+      "step": 23876
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.467705557204499e-06,
+      "loss": 0.1096,
+      "step": 23877
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4669509296421524e-06,
+      "loss": 0.12,
+      "step": 23878
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.466196366979594e-06,
+      "loss": 0.1252,
+      "step": 23879
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4654418692243088e-06,
+      "loss": 0.1394,
+      "step": 23880
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.464687436383798e-06,
+      "loss": 0.072,
+      "step": 23881
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.463933068465559e-06,
+      "loss": 0.0741,
+      "step": 23882
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.463178765477081e-06,
+      "loss": 0.104,
+      "step": 23883
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4624245274258626e-06,
+      "loss": 0.0916,
+      "step": 23884
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.461670354319393e-06,
+      "loss": 0.1142,
+      "step": 23885
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4609162461651613e-06,
+      "loss": 0.1048,
+      "step": 23886
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4601622029706615e-06,
+      "loss": 0.0993,
+      "step": 23887
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.459408224743388e-06,
+      "loss": 0.1151,
+      "step": 23888
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.458654311490829e-06,
+      "loss": 0.1411,
+      "step": 23889
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4579004632204683e-06,
+      "loss": 0.106,
+      "step": 23890
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4571466799398025e-06,
+      "loss": 0.1002,
+      "step": 23891
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.456392961656313e-06,
+      "loss": 0.087,
+      "step": 23892
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4556393083774932e-06,
+      "loss": 0.1198,
+      "step": 23893
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4548857201108275e-06,
+      "loss": 0.1072,
+      "step": 23894
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4541321968637995e-06,
+      "loss": 0.1274,
+      "step": 23895
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4533787386438967e-06,
+      "loss": 0.1107,
+      "step": 23896
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.452625345458608e-06,
+      "loss": 0.0938,
+      "step": 23897
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4518720173154143e-06,
+      "loss": 0.1052,
+      "step": 23898
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.451118754221795e-06,
+      "loss": 0.1354,
+      "step": 23899
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.450365556185241e-06,
+      "loss": 0.1304,
+      "step": 23900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.449612423213227e-06,
+      "loss": 0.1334,
+      "step": 23901
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4488593553132423e-06,
+      "loss": 0.0541,
+      "step": 23902
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4481063524927638e-06,
+      "loss": 0.0845,
+      "step": 23903
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.447353414759268e-06,
+      "loss": 0.1357,
+      "step": 23904
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.446600542120243e-06,
+      "loss": 0.0806,
+      "step": 23905
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.44584773458316e-06,
+      "loss": 0.1034,
+      "step": 23906
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.445094992155501e-06,
+      "loss": 0.0773,
+      "step": 23907
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4443423148447474e-06,
+      "loss": 0.1101,
+      "step": 23908
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.443589702658373e-06,
+      "loss": 0.147,
+      "step": 23909
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4428371556038507e-06,
+      "loss": 0.1258,
+      "step": 23910
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.442084673688664e-06,
+      "loss": 0.1227,
+      "step": 23911
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4413322569202802e-06,
+      "loss": 0.0961,
+      "step": 23912
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4405799053061804e-06,
+      "loss": 0.163,
+      "step": 23913
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.439827618853836e-06,
+      "loss": 0.0968,
+      "step": 23914
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.439075397570716e-06,
+      "loss": 0.1352,
+      "step": 23915
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4383232414642974e-06,
+      "loss": 0.1112,
+      "step": 23916
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.437571150542055e-06,
+      "loss": 0.0956,
+      "step": 23917
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4368191248114578e-06,
+      "loss": 0.1211,
+      "step": 23918
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4360671642799714e-06,
+      "loss": 0.1166,
+      "step": 23919
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4353152689550727e-06,
+      "loss": 0.1145,
+      "step": 23920
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.434563438844226e-06,
+      "loss": 0.0854,
+      "step": 23921
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433811673954902e-06,
+      "loss": 0.1542,
+      "step": 23922
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433059974294577e-06,
+      "loss": 0.1168,
+      "step": 23923
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4323083398707025e-06,
+      "loss": 0.0901,
+      "step": 23924
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.431556770690755e-06,
+      "loss": 0.1044,
+      "step": 23925
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4308052667622006e-06,
+      "loss": 0.1355,
+      "step": 23926
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4300538280925043e-06,
+      "loss": 0.1247,
+      "step": 23927
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4293024546891263e-06,
+      "loss": 0.102,
+      "step": 23928
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.428551146559538e-06,
+      "loss": 0.0882,
+      "step": 23929
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4277999037111966e-06,
+      "loss": 0.1145,
+      "step": 23930
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4270487261515663e-06,
+      "loss": 0.1026,
+      "step": 23931
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.426297613888119e-06,
+      "loss": 0.1019,
+      "step": 23932
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4255465669283005e-06,
+      "loss": 0.1121,
+      "step": 23933
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.424795585279579e-06,
+      "loss": 0.1852,
+      "step": 23934
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.42404466894942e-06,
+      "loss": 0.0955,
+      "step": 23935
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4232938179452733e-06,
+      "loss": 0.1321,
+      "step": 23936
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4225430322746077e-06,
+      "loss": 0.1156,
+      "step": 23937
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.421792311944876e-06,
+      "loss": 0.1334,
+      "step": 23938
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.421041656963534e-06,
+      "loss": 0.0782,
+      "step": 23939
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4202910673380407e-06,
+      "loss": 0.0942,
+      "step": 23940
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4195405430758576e-06,
+      "loss": 0.1448,
+      "step": 23941
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4187900841844357e-06,
+      "loss": 0.1286,
+      "step": 23942
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4180396906712265e-06,
+      "loss": 0.0903,
+      "step": 23943
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.417289362543693e-06,
+      "loss": 0.1078,
+      "step": 23944
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4165390998092806e-06,
+      "loss": 0.1587,
+      "step": 23945
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4157889024754465e-06,
+      "loss": 0.1097,
+      "step": 23946
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.41503877054965e-06,
+      "loss": 0.0915,
+      "step": 23947
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.414288704039329e-06,
+      "loss": 0.1074,
+      "step": 23948
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.413538702951942e-06,
+      "loss": 0.1159,
+      "step": 23949
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4127887672949434e-06,
+      "loss": 0.0945,
+      "step": 23950
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4120388970757744e-06,
+      "loss": 0.1031,
+      "step": 23951
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4112890923018936e-06,
+      "loss": 0.1187,
+      "step": 23952
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.410539352980744e-06,
+      "loss": 0.0911,
+      "step": 23953
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4097896791197716e-06,
+      "loss": 0.0953,
+      "step": 23954
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.409040070726427e-06,
+      "loss": 0.0872,
+      "step": 23955
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4082905278081623e-06,
+      "loss": 0.1133,
+      "step": 23956
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4075410503724116e-06,
+      "loss": 0.0851,
+      "step": 23957
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4067916384266264e-06,
+      "loss": 0.1119,
+      "step": 23958
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4060422919782556e-06,
+      "loss": 0.1088,
+      "step": 23959
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.405293011034734e-06,
+      "loss": 0.1072,
+      "step": 23960
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4045437956035143e-06,
+      "loss": 0.1116,
+      "step": 23961
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4037946456920347e-06,
+      "loss": 0.0895,
+      "step": 23962
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.403045561307735e-06,
+      "loss": 0.1287,
+      "step": 23963
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.402296542458059e-06,
+      "loss": 0.1059,
+      "step": 23964
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4015475891504514e-06,
+      "loss": 0.1366,
+      "step": 23965
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.400798701392348e-06,
+      "loss": 0.1137,
+      "step": 23966
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.400049879191186e-06,
+      "loss": 0.1793,
+      "step": 23967
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3993011225544104e-06,
+      "loss": 0.09,
+      "step": 23968
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3985524314894536e-06,
+      "loss": 0.1185,
+      "step": 23969
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3978038060037586e-06,
+      "loss": 0.0843,
+      "step": 23970
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.397055246104761e-06,
+      "loss": 0.1048,
+      "step": 23971
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3963067517998914e-06,
+      "loss": 0.0666,
+      "step": 23972
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3955583230965896e-06,
+      "loss": 0.1108,
+      "step": 23973
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.394809960002293e-06,
+      "loss": 0.1136,
+      "step": 23974
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3940616625244314e-06,
+      "loss": 0.0951,
+      "step": 23975
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.393313430670443e-06,
+      "loss": 0.1,
+      "step": 23976
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.392565264447759e-06,
+      "loss": 0.1172,
+      "step": 23977
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3918171638638076e-06,
+      "loss": 0.1261,
+      "step": 23978
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3910691289260267e-06,
+      "loss": 0.1163,
+      "step": 23979
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.390321159641842e-06,
+      "loss": 0.1195,
+      "step": 23980
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.389573256018689e-06,
+      "loss": 0.1237,
+      "step": 23981
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3888254180639922e-06,
+      "loss": 0.1017,
+      "step": 23982
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.388077645785186e-06,
+      "loss": 0.0862,
+      "step": 23983
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.387329939189693e-06,
+      "loss": 0.0618,
+      "step": 23984
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3865822982849485e-06,
+      "loss": 0.1013,
+      "step": 23985
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.385834723078374e-06,
+      "loss": 0.1086,
+      "step": 23986
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3850872135773937e-06,
+      "loss": 0.1492,
+      "step": 23987
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3843397697894407e-06,
+      "loss": 0.0842,
+      "step": 23988
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.383592391721934e-06,
+      "loss": 0.1049,
+      "step": 23989
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.382845079382303e-06,
+      "loss": 0.1028,
+      "step": 23990
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3820978327779654e-06,
+      "loss": 0.1235,
+      "step": 23991
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3813506519163518e-06,
+      "loss": 0.1269,
+      "step": 23992
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3806035368048774e-06,
+      "loss": 0.1108,
+      "step": 23993
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.379856487450972e-06,
+      "loss": 0.1008,
+      "step": 23994
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3791095038620515e-06,
+      "loss": 0.1382,
+      "step": 23995
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.378362586045535e-06,
+      "loss": 0.0889,
+      "step": 23996
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.377615734008848e-06,
+      "loss": 0.12,
+      "step": 23997
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.376868947759403e-06,
+      "loss": 0.1052,
+      "step": 23998
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3761222273046223e-06,
+      "loss": 0.1098,
+      "step": 23999
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3753755726519276e-06,
+      "loss": 0.072,
+      "step": 24000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.374628983808732e-06,
+      "loss": 0.0804,
+      "step": 24001
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3738824607824504e-06,
+      "loss": 0.1139,
+      "step": 24002
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3731360035805036e-06,
+      "loss": 0.1258,
+      "step": 24003
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3723896122103027e-06,
+      "loss": 0.1161,
+      "step": 24004
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.371643286679266e-06,
+      "loss": 0.1421,
+      "step": 24005
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.370897026994806e-06,
+      "loss": 0.0833,
+      "step": 24006
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.370150833164333e-06,
+      "loss": 0.1095,
+      "step": 24007
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3694047051952626e-06,
+      "loss": 0.1054,
+      "step": 24008
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.368658643095011e-06,
+      "loss": 0.0845,
+      "step": 24009
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3679126468709844e-06,
+      "loss": 0.0972,
+      "step": 24010
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.367166716530591e-06,
+      "loss": 0.1044,
+      "step": 24011
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.366420852081249e-06,
+      "loss": 0.089,
+      "step": 24012
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3656750535303594e-06,
+      "loss": 0.0711,
+      "step": 24013
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3649293208853394e-06,
+      "loss": 0.1092,
+      "step": 24014
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.364183654153592e-06,
+      "loss": 0.0856,
+      "step": 24015
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3634380533425225e-06,
+      "loss": 0.1356,
+      "step": 24016
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3626925184595414e-06,
+      "loss": 0.1199,
+      "step": 24017
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3619470495120575e-06,
+      "loss": 0.1053,
+      "step": 24018
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3612016465074726e-06,
+      "loss": 0.0824,
+      "step": 24019
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3604563094531894e-06,
+      "loss": 0.1049,
+      "step": 24020
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.359711038356619e-06,
+      "loss": 0.1493,
+      "step": 24021
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.358965833225156e-06,
+      "loss": 0.11,
+      "step": 24022
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.358220694066209e-06,
+      "loss": 0.1111,
+      "step": 24023
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3574756208871862e-06,
+      "loss": 0.1246,
+      "step": 24024
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.356730613695476e-06,
+      "loss": 0.095,
+      "step": 24025
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.355985672498485e-06,
+      "loss": 0.1094,
+      "step": 24026
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3552407973036183e-06,
+      "loss": 0.1212,
+      "step": 24027
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3544959881182683e-06,
+      "loss": 0.0812,
+      "step": 24028
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.353751244949841e-06,
+      "loss": 0.1215,
+      "step": 24029
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.35300656780573e-06,
+      "loss": 0.1078,
+      "step": 24030
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3522619566933313e-06,
+      "loss": 0.1243,
+      "step": 24031
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3515174116200444e-06,
+      "loss": 0.1195,
+      "step": 24032
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.35077293259327e-06,
+      "loss": 0.0915,
+      "step": 24033
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3500285196203986e-06,
+      "loss": 0.1238,
+      "step": 24034
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.349284172708823e-06,
+      "loss": 0.1174,
+      "step": 24035
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3485398918659453e-06,
+      "loss": 0.1065,
+      "step": 24036
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3477956770991514e-06,
+      "loss": 0.1033,
+      "step": 24037
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.347051528415837e-06,
+      "loss": 0.1318,
+      "step": 24038
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3463074458234023e-06,
+      "loss": 0.1115,
+      "step": 24039
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.345563429329226e-06,
+      "loss": 0.1289,
+      "step": 24040
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3448194789407053e-06,
+      "loss": 0.106,
+      "step": 24041
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3440755946652335e-06,
+      "loss": 0.0977,
+      "step": 24042
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.343331776510198e-06,
+      "loss": 0.0949,
+      "step": 24043
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3425880244829843e-06,
+      "loss": 0.1078,
+      "step": 24044
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.341844338590987e-06,
+      "loss": 0.0907,
+      "step": 24045
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.341100718841589e-06,
+      "loss": 0.1063,
+      "step": 24046
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3403571652421783e-06,
+      "loss": 0.1161,
+      "step": 24047
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.33961367780015e-06,
+      "loss": 0.1115,
+      "step": 24048
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.338870256522876e-06,
+      "loss": 0.0853,
+      "step": 24049
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3381269014177477e-06,
+      "loss": 0.0968,
+      "step": 24050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.337383612492153e-06,
+      "loss": 0.1179,
+      "step": 24051
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3366403897534704e-06,
+      "loss": 0.158,
+      "step": 24052
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.335897233209089e-06,
+      "loss": 0.1454,
+      "step": 24053
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.335154142866387e-06,
+      "loss": 0.0985,
+      "step": 24054
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.334411118732744e-06,
+      "loss": 0.1379,
+      "step": 24055
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3336681608155463e-06,
+      "loss": 0.0769,
+      "step": 24056
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3329252691221747e-06,
+      "loss": 0.1111,
+      "step": 24057
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.332182443660007e-06,
+      "loss": 0.1206,
+      "step": 24058
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3314396844364193e-06,
+      "loss": 0.0863,
+      "step": 24059
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.330696991458797e-06,
+      "loss": 0.1432,
+      "step": 24060
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.329954364734512e-06,
+      "loss": 0.1076,
+      "step": 24061
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3292118042709475e-06,
+      "loss": 0.1128,
+      "step": 24062
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.328469310075476e-06,
+      "loss": 0.0909,
+      "step": 24063
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.327726882155472e-06,
+      "loss": 0.1097,
+      "step": 24064
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.326984520518313e-06,
+      "loss": 0.1383,
+      "step": 24065
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3262422251713766e-06,
+      "loss": 0.1214,
+      "step": 24066
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3254999961220347e-06,
+      "loss": 0.0925,
+      "step": 24067
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3247578333776574e-06,
+      "loss": 0.0871,
+      "step": 24068
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3240157369456227e-06,
+      "loss": 0.1191,
+      "step": 24069
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3232737068332963e-06,
+      "loss": 0.1199,
+      "step": 24070
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3225317430480564e-06,
+      "loss": 0.1161,
+      "step": 24071
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3217898455972717e-06,
+      "loss": 0.0988,
+      "step": 24072
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3210480144883072e-06,
+      "loss": 0.1257,
+      "step": 24073
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3203062497285397e-06,
+      "loss": 0.0975,
+      "step": 24074
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3195645513253316e-06,
+      "loss": 0.0866,
+      "step": 24075
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3188229192860533e-06,
+      "loss": 0.1154,
+      "step": 24076
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3180813536180766e-06,
+      "loss": 0.1378,
+      "step": 24077
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.317339854328764e-06,
+      "loss": 0.0937,
+      "step": 24078
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3165984214254797e-06,
+      "loss": 0.0972,
+      "step": 24079
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3158570549155945e-06,
+      "loss": 0.0825,
+      "step": 24080
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.315115754806466e-06,
+      "loss": 0.0857,
+      "step": 24081
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.314374521105467e-06,
+      "loss": 0.1009,
+      "step": 24082
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.313633353819956e-06,
+      "loss": 0.1119,
+      "step": 24083
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3128922529572937e-06,
+      "loss": 0.1107,
+      "step": 24084
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3121512185248437e-06,
+      "loss": 0.1086,
+      "step": 24085
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3114102505299727e-06,
+      "loss": 0.0798,
+      "step": 24086
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.310669348980038e-06,
+      "loss": 0.1339,
+      "step": 24087
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3099285138823956e-06,
+      "loss": 0.1143,
+      "step": 24088
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.309187745244412e-06,
+      "loss": 0.0804,
+      "step": 24089
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3084470430734395e-06,
+      "loss": 0.0986,
+      "step": 24090
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3077064073768396e-06,
+      "loss": 0.1215,
+      "step": 24091
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3069658381619762e-06,
+      "loss": 0.1168,
+      "step": 24092
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3062253354361926e-06,
+      "loss": 0.1264,
+      "step": 24093
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3054848992068534e-06,
+      "loss": 0.1055,
+      "step": 24094
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.304744529481315e-06,
+      "loss": 0.1237,
+      "step": 24095
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3040042262669313e-06,
+      "loss": 0.1072,
+      "step": 24096
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3032639895710517e-06,
+      "loss": 0.0919,
+      "step": 24097
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.302523819401037e-06,
+      "loss": 0.1152,
+      "step": 24098
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.301783715764232e-06,
+      "loss": 0.097,
+      "step": 24099
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3010436786679944e-06,
+      "loss": 0.131,
+      "step": 24100
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3003037081196822e-06,
+      "loss": 0.1386,
+      "step": 24101
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2995638041266307e-06,
+      "loss": 0.1048,
+      "step": 24102
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.298823966696201e-06,
+      "loss": 0.1301,
+      "step": 24103
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2980841958357413e-06,
+      "loss": 0.0978,
+      "step": 24104
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2973444915525966e-06,
+      "loss": 0.1031,
+      "step": 24105
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.296604853854122e-06,
+      "loss": 0.1282,
+      "step": 24106
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.29586528274766e-06,
+      "loss": 0.0955,
+      "step": 24107
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2951257782405566e-06,
+      "loss": 0.0772,
+      "step": 24108
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2943863403401597e-06,
+      "loss": 0.0905,
+      "step": 24109
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2936469690538185e-06,
+      "loss": 0.102,
+      "step": 24110
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2929076643888746e-06,
+      "loss": 0.1138,
+      "step": 24111
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.29216842635267e-06,
+      "loss": 0.1218,
+      "step": 24112
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2914292549525537e-06,
+      "loss": 0.1358,
+      "step": 24113
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2906901501958634e-06,
+      "loss": 0.1112,
+      "step": 24114
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.289951112089943e-06,
+      "loss": 0.1069,
+      "step": 24115
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.289212140642142e-06,
+      "loss": 0.1705,
+      "step": 24116
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.288473235859787e-06,
+      "loss": 0.0941,
+      "step": 24117
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2877343977502264e-06,
+      "loss": 0.1188,
+      "step": 24118
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2869956263208036e-06,
+      "loss": 0.1076,
+      "step": 24119
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.286256921578849e-06,
+      "loss": 0.1466,
+      "step": 24120
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.285518283531708e-06,
+      "loss": 0.1024,
+      "step": 24121
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2847797121867165e-06,
+      "loss": 0.1167,
+      "step": 24122
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.284041207551206e-06,
+      "loss": 0.0956,
+      "step": 24123
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.283302769632519e-06,
+      "loss": 0.1234,
+      "step": 24124
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.282564398437995e-06,
+      "loss": 0.1104,
+      "step": 24125
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.281826093974958e-06,
+      "loss": 0.0845,
+      "step": 24126
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2810878562507474e-06,
+      "loss": 0.1,
+      "step": 24127
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2803496852727013e-06,
+      "loss": 0.1097,
+      "step": 24128
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2796115810481444e-06,
+      "loss": 0.1396,
+      "step": 24129
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2788735435844175e-06,
+      "loss": 0.1296,
+      "step": 24130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2781355728888487e-06,
+      "loss": 0.1081,
+      "step": 24131
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2773976689687657e-06,
+      "loss": 0.117,
+      "step": 24132
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2766598318315013e-06,
+      "loss": 0.0849,
+      "step": 24133
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.27592206148439e-06,
+      "loss": 0.1115,
+      "step": 24134
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2751843579347562e-06,
+      "loss": 0.1165,
+      "step": 24135
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2744467211899254e-06,
+      "loss": 0.1134,
+      "step": 24136
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.273709151257233e-06,
+      "loss": 0.0759,
+      "step": 24137
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2729716481439976e-06,
+      "loss": 0.1015,
+      "step": 24138
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2722342118575547e-06,
+      "loss": 0.1279,
+      "step": 24139
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.271496842405224e-06,
+      "loss": 0.104,
+      "step": 24140
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.270759539794328e-06,
+      "loss": 0.0956,
+      "step": 24141
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2700223040321953e-06,
+      "loss": 0.1016,
+      "step": 24142
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2692851351261534e-06,
+      "loss": 0.1278,
+      "step": 24143
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2685480330835172e-06,
+      "loss": 0.1015,
+      "step": 24144
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.267810997911617e-06,
+      "loss": 0.1308,
+      "step": 24145
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2670740296177705e-06,
+      "loss": 0.0987,
+      "step": 24146
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.266337128209296e-06,
+      "loss": 0.118,
+      "step": 24147
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2656002936935203e-06,
+      "loss": 0.1441,
+      "step": 24148
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2648635260777563e-06,
+      "loss": 0.0515,
+      "step": 24149
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2641268253693305e-06,
+      "loss": 0.0832,
+      "step": 24150
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.263390191575554e-06,
+      "loss": 0.1014,
+      "step": 24151
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.262653624703751e-06,
+      "loss": 0.1278,
+      "step": 24152
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2619171247612327e-06,
+      "loss": 0.1145,
+      "step": 24153
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2611806917553215e-06,
+      "loss": 0.1511,
+      "step": 24154
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2604443256933307e-06,
+      "loss": 0.0787,
+      "step": 24155
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.259708026582571e-06,
+      "loss": 0.1117,
+      "step": 24156
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2589717944303643e-06,
+      "loss": 0.1125,
+      "step": 24157
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2582356292440174e-06,
+      "loss": 0.1063,
+      "step": 24158
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.257499531030851e-06,
+      "loss": 0.1135,
+      "step": 24159
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.256763499798169e-06,
+      "loss": 0.1126,
+      "step": 24160
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2560275355532924e-06,
+      "loss": 0.1041,
+      "step": 24161
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2552916383035228e-06,
+      "loss": 0.0745,
+      "step": 24162
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.254555808056179e-06,
+      "loss": 0.0993,
+      "step": 24163
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2538200448185675e-06,
+      "loss": 0.1078,
+      "step": 24164
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2530843485979936e-06,
+      "loss": 0.1072,
+      "step": 24165
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2523487194017723e-06,
+      "loss": 0.1089,
+      "step": 24166
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2516131572372058e-06,
+      "loss": 0.0951,
+      "step": 24167
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2508776621116035e-06,
+      "loss": 0.1232,
+      "step": 24168
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.250142234032275e-06,
+      "loss": 0.0983,
+      "step": 24169
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2494068730065243e-06,
+      "loss": 0.0912,
+      "step": 24170
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2486715790416514e-06,
+      "loss": 0.0957,
+      "step": 24171
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.247936352144968e-06,
+      "loss": 0.0968,
+      "step": 24172
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.247201192323772e-06,
+      "loss": 0.1212,
+      "step": 24173
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2464660995853723e-06,
+      "loss": 0.1021,
+      "step": 24174
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2457310739370684e-06,
+      "loss": 0.104,
+      "step": 24175
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2449961153861588e-06,
+      "loss": 0.0965,
+      "step": 24176
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2442612239399474e-06,
+      "loss": 0.1013,
+      "step": 24177
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2435263996057388e-06,
+      "loss": 0.0922,
+      "step": 24178
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.242791642390829e-06,
+      "loss": 0.1065,
+      "step": 24179
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2420569523025135e-06,
+      "loss": 0.131,
+      "step": 24180
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.241322329348099e-06,
+      "loss": 0.0846,
+      "step": 24181
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2405877735348746e-06,
+      "loss": 0.1081,
+      "step": 24182
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2398532848701447e-06,
+      "loss": 0.1082,
+      "step": 24183
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2391188633612036e-06,
+      "loss": 0.1376,
+      "step": 24184
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2383845090153434e-06,
+      "loss": 0.0889,
+      "step": 24185
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.237650221839861e-06,
+      "loss": 0.0912,
+      "step": 24186
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.236916001842055e-06,
+      "loss": 0.0943,
+      "step": 24187
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2361818490292164e-06,
+      "loss": 0.095,
+      "step": 24188
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2354477634086346e-06,
+      "loss": 0.084,
+      "step": 24189
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.23471374498761e-06,
+      "loss": 0.1144,
+      "step": 24190
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.233979793773425e-06,
+      "loss": 0.1143,
+      "step": 24191
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.233245909773376e-06,
+      "loss": 0.1115,
+      "step": 24192
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2325120929947595e-06,
+      "loss": 0.0908,
+      "step": 24193
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2317783434448537e-06,
+      "loss": 0.0986,
+      "step": 24194
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.231044661130951e-06,
+      "loss": 0.1145,
+      "step": 24195
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.230311046060346e-06,
+      "loss": 0.095,
+      "step": 24196
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.229577498240319e-06,
+      "loss": 0.1103,
+      "step": 24197
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.228844017678163e-06,
+      "loss": 0.074,
+      "step": 24198
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2281106043811627e-06,
+      "loss": 0.1018,
+      "step": 24199
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.227377258356599e-06,
+      "loss": 0.08,
+      "step": 24200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.226643979611761e-06,
+      "loss": 0.1258,
+      "step": 24201
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.225910768153936e-06,
+      "loss": 0.0858,
+      "step": 24202
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2251776239904054e-06,
+      "loss": 0.1112,
+      "step": 24203
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.224444547128449e-06,
+      "loss": 0.1289,
+      "step": 24204
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2237115375753547e-06,
+      "loss": 0.082,
+      "step": 24205
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2229785953383984e-06,
+      "loss": 0.1127,
+      "step": 24206
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.222245720424868e-06,
+      "loss": 0.0874,
+      "step": 24207
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2215129128420395e-06,
+      "loss": 0.1141,
+      "step": 24208
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.220780172597191e-06,
+      "loss": 0.1196,
+      "step": 24209
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2200474996976038e-06,
+      "loss": 0.0956,
+      "step": 24210
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2193148941505604e-06,
+      "loss": 0.1047,
+      "step": 24211
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2185823559633344e-06,
+      "loss": 0.1157,
+      "step": 24212
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2178498851431993e-06,
+      "loss": 0.0911,
+      "step": 24213
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.217117481697439e-06,
+      "loss": 0.1034,
+      "step": 24214
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.216385145633324e-06,
+      "loss": 0.0918,
+      "step": 24215
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2156528769581295e-06,
+      "loss": 0.091,
+      "step": 24216
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2149206756791384e-06,
+      "loss": 0.1499,
+      "step": 24217
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.214188541803611e-06,
+      "loss": 0.0919,
+      "step": 24218
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2134564753388276e-06,
+      "loss": 0.0999,
+      "step": 24219
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.212724476292063e-06,
+      "loss": 0.1094,
+      "step": 24220
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2119925446705824e-06,
+      "loss": 0.1216,
+      "step": 24221
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2112606804816625e-06,
+      "loss": 0.117,
+      "step": 24222
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2105288837325732e-06,
+      "loss": 0.0906,
+      "step": 24223
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2097971544305784e-06,
+      "loss": 0.134,
+      "step": 24224
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.209065492582951e-06,
+      "loss": 0.0869,
+      "step": 24225
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2083338981969637e-06,
+      "loss": 0.1346,
+      "step": 24226
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.20760237127988e-06,
+      "loss": 0.0905,
+      "step": 24227
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.206870911838964e-06,
+      "loss": 0.0766,
+      "step": 24228
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.206139519881488e-06,
+      "loss": 0.0586,
+      "step": 24229
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.205408195414712e-06,
+      "loss": 0.1351,
+      "step": 24230
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2046769384459073e-06,
+      "loss": 0.1132,
+      "step": 24231
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2039457489823357e-06,
+      "loss": 0.1013,
+      "step": 24232
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2032146270312546e-06,
+      "loss": 0.0898,
+      "step": 24233
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.202483572599938e-06,
+      "loss": 0.1376,
+      "step": 24234
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.201752585695638e-06,
+      "loss": 0.0982,
+      "step": 24235
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2010216663256257e-06,
+      "loss": 0.1295,
+      "step": 24236
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2002908144971524e-06,
+      "loss": 0.1018,
+      "step": 24237
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.199560030217488e-06,
+      "loss": 0.1308,
+      "step": 24238
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1988293134938843e-06,
+      "loss": 0.1803,
+      "step": 24239
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1980986643336066e-06,
+      "loss": 0.1137,
+      "step": 24240
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.197368082743909e-06,
+      "loss": 0.1098,
+      "step": 24241
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1966375687320494e-06,
+      "loss": 0.0939,
+      "step": 24242
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.195907122305287e-06,
+      "loss": 0.092,
+      "step": 24243
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.195176743470875e-06,
+      "loss": 0.1212,
+      "step": 24244
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1944464322360704e-06,
+      "loss": 0.1481,
+      "step": 24245
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1937161886081327e-06,
+      "loss": 0.1627,
+      "step": 24246
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.192986012594311e-06,
+      "loss": 0.0975,
+      "step": 24247
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1922559042018573e-06,
+      "loss": 0.0805,
+      "step": 24248
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.191525863438031e-06,
+      "loss": 0.0843,
+      "step": 24249
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1907958903100777e-06,
+      "loss": 0.113,
+      "step": 24250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1900659848252545e-06,
+      "loss": 0.1274,
+      "step": 24251
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.18933614699081e-06,
+      "loss": 0.1102,
+      "step": 24252
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1886063768139907e-06,
+      "loss": 0.1026,
+      "step": 24253
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1878766743020506e-06,
+      "loss": 0.1068,
+      "step": 24254
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1871470394622407e-06,
+      "loss": 0.1228,
+      "step": 24255
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.186417472301807e-06,
+      "loss": 0.1023,
+      "step": 24256
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1856879728279923e-06,
+      "loss": 0.1067,
+      "step": 24257
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.184958541048051e-06,
+      "loss": 0.0938,
+      "step": 24258
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1842291769692237e-06,
+      "loss": 0.1126,
+      "step": 24259
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.183499880598757e-06,
+      "loss": 0.1089,
+      "step": 24260
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.182770651943905e-06,
+      "loss": 0.0782,
+      "step": 24261
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1820414910118958e-06,
+      "loss": 0.1248,
+      "step": 24262
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1813123978099824e-06,
+      "loss": 0.0909,
+      "step": 24263
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1805833723454094e-06,
+      "loss": 0.0968,
+      "step": 24264
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.179854414625415e-06,
+      "loss": 0.0845,
+      "step": 24265
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1791255246572396e-06,
+      "loss": 0.094,
+      "step": 24266
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.178396702448128e-06,
+      "loss": 0.1215,
+      "step": 24267
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1776679480053163e-06,
+      "loss": 0.1125,
+      "step": 24268
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.176939261336045e-06,
+      "loss": 0.1047,
+      "step": 24269
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1762106424475612e-06,
+      "loss": 0.1045,
+      "step": 24270
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.17548209134709e-06,
+      "loss": 0.1067,
+      "step": 24271
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.174753608041875e-06,
+      "loss": 0.1041,
+      "step": 24272
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.174025192539155e-06,
+      "loss": 0.0978,
+      "step": 24273
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.173296844846161e-06,
+      "loss": 0.1356,
+      "step": 24274
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1725685649701354e-06,
+      "loss": 0.1056,
+      "step": 24275
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1718403529183083e-06,
+      "loss": 0.0895,
+      "step": 24276
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1711122086979118e-06,
+      "loss": 0.1087,
+      "step": 24277
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.170384132316181e-06,
+      "loss": 0.0926,
+      "step": 24278
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1696561237803538e-06,
+      "loss": 0.1242,
+      "step": 24279
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1689281830976584e-06,
+      "loss": 0.1519,
+      "step": 24280
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.168200310275322e-06,
+      "loss": 0.1041,
+      "step": 24281
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1674725053205836e-06,
+      "loss": 0.1023,
+      "step": 24282
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.166744768240665e-06,
+      "loss": 0.1306,
+      "step": 24283
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1660170990427994e-06,
+      "loss": 0.1349,
+      "step": 24284
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.165289497734223e-06,
+      "loss": 0.1048,
+      "step": 24285
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.16456196432215e-06,
+      "loss": 0.0814,
+      "step": 24286
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.163834498813815e-06,
+      "loss": 0.1299,
+      "step": 24287
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1631071012164462e-06,
+      "loss": 0.0825,
+      "step": 24288
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.162379771537264e-06,
+      "loss": 0.074,
+      "step": 24289
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1616525097835014e-06,
+      "loss": 0.1026,
+      "step": 24290
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1609253159623786e-06,
+      "loss": 0.1165,
+      "step": 24291
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1601981900811164e-06,
+      "loss": 0.1026,
+      "step": 24292
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.159471132146942e-06,
+      "loss": 0.1022,
+      "step": 24293
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1587441421670838e-06,
+      "loss": 0.0881,
+      "step": 24294
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.158017220148751e-06,
+      "loss": 0.1004,
+      "step": 24295
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1572903660991717e-06,
+      "loss": 0.1158,
+      "step": 24296
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1565635800255702e-06,
+      "loss": 0.0779,
+      "step": 24297
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.155836861935159e-06,
+      "loss": 0.1113,
+      "step": 24298
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.155110211835163e-06,
+      "loss": 0.0943,
+      "step": 24299
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1543836297328e-06,
+      "loss": 0.0977,
+      "step": 24300
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1536571156352823e-06,
+      "loss": 0.1256,
+      "step": 24301
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.152930669549833e-06,
+      "loss": 0.1089,
+      "step": 24302
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1522042914836703e-06,
+      "loss": 0.088,
+      "step": 24303
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1514779814440057e-06,
+      "loss": 0.1188,
+      "step": 24304
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1507517394380536e-06,
+      "loss": 0.0992,
+      "step": 24305
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.150025565473034e-06,
+      "loss": 0.1442,
+      "step": 24306
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1492994595561533e-06,
+      "loss": 0.1516,
+      "step": 24307
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1485734216946317e-06,
+      "loss": 0.133,
+      "step": 24308
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.14784745189568e-06,
+      "loss": 0.0821,
+      "step": 24309
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1471215501665054e-06,
+      "loss": 0.1054,
+      "step": 24310
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.146395716514322e-06,
+      "loss": 0.1052,
+      "step": 24311
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.145669950946344e-06,
+      "loss": 0.1065,
+      "step": 24312
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.144944253469775e-06,
+      "loss": 0.1893,
+      "step": 24313
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1442186240918303e-06,
+      "loss": 0.1172,
+      "step": 24314
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1434930628197145e-06,
+      "loss": 0.1269,
+      "step": 24315
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1427675696606342e-06,
+      "loss": 0.0957,
+      "step": 24316
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1420421446218007e-06,
+      "loss": 0.1057,
+      "step": 24317
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1413167877104158e-06,
+      "loss": 0.0597,
+      "step": 24318
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1405914989336894e-06,
+      "loss": 0.1174,
+      "step": 24319
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.139866278298822e-06,
+      "loss": 0.1268,
+      "step": 24320
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1391411258130243e-06,
+      "loss": 0.1353,
+      "step": 24321
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138416041483493e-06,
+      "loss": 0.1038,
+      "step": 24322
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1376910253174374e-06,
+      "loss": 0.1186,
+      "step": 24323
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1369660773220566e-06,
+      "loss": 0.0565,
+      "step": 24324
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1362411975045504e-06,
+      "loss": 0.1171,
+      "step": 24325
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135516385872125e-06,
+      "loss": 0.0894,
+      "step": 24326
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1347916424319745e-06,
+      "loss": 0.1089,
+      "step": 24327
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134066967191306e-06,
+      "loss": 0.0935,
+      "step": 24328
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.133342360157311e-06,
+      "loss": 0.1022,
+      "step": 24329
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1326178213371937e-06,
+      "loss": 0.1133,
+      "step": 24330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1318933507381454e-06,
+      "loss": 0.0926,
+      "step": 24331
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131168948367371e-06,
+      "loss": 0.1187,
+      "step": 24332
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.130444614232062e-06,
+      "loss": 0.1004,
+      "step": 24333
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1297203483394124e-06,
+      "loss": 0.1019,
+      "step": 24334
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1289961506966217e-06,
+      "loss": 0.1229,
+      "step": 24335
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1282720213108787e-06,
+      "loss": 0.1006,
+      "step": 24336
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.12754796018938e-06,
+      "loss": 0.1019,
+      "step": 24337
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.126823967339322e-06,
+      "loss": 0.1044,
+      "step": 24338
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1261000427678933e-06,
+      "loss": 0.1072,
+      "step": 24339
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.125376186482282e-06,
+      "loss": 0.1515,
+      "step": 24340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.124652398489685e-06,
+      "loss": 0.1297,
+      "step": 24341
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.123928678797288e-06,
+      "loss": 0.0957,
+      "step": 24342
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1232050274122837e-06,
+      "loss": 0.0898,
+      "step": 24343
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1224814443418614e-06,
+      "loss": 0.0829,
+      "step": 24344
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.121757929593203e-06,
+      "loss": 0.1151,
+      "step": 24345
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1210344831735017e-06,
+      "loss": 0.1095,
+      "step": 24346
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1203111050899447e-06,
+      "loss": 0.1009,
+      "step": 24347
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1195877953497167e-06,
+      "loss": 0.1326,
+      "step": 24348
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1188645539600004e-06,
+      "loss": 0.1256,
+      "step": 24349
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1181413809279848e-06,
+      "loss": 0.1223,
+      "step": 24350
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1174182762608495e-06,
+      "loss": 0.0918,
+      "step": 24351
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1166952399657835e-06,
+      "loss": 0.1028,
+      "step": 24352
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.115972272049966e-06,
+      "loss": 0.1293,
+      "step": 24353
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.115249372520577e-06,
+      "loss": 0.1098,
+      "step": 24354
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.114526541384799e-06,
+      "loss": 0.0953,
+      "step": 24355
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1138037786498188e-06,
+      "loss": 0.0982,
+      "step": 24356
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.11308108432281e-06,
+      "loss": 0.092,
+      "step": 24357
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1123584584109502e-06,
+      "loss": 0.1404,
+      "step": 24358
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1116359009214247e-06,
+      "loss": 0.1175,
+      "step": 24359
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1109134118614048e-06,
+      "loss": 0.1061,
+      "step": 24360
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1101909912380712e-06,
+      "loss": 0.0747,
+      "step": 24361
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.109468639058606e-06,
+      "loss": 0.1188,
+      "step": 24362
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.108746355330173e-06,
+      "loss": 0.0795,
+      "step": 24363
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1080241400599543e-06,
+      "loss": 0.091,
+      "step": 24364
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.107301993255126e-06,
+      "loss": 0.1055,
+      "step": 24365
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.106579914922857e-06,
+      "loss": 0.1122,
+      "step": 24366
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1058579050703265e-06,
+      "loss": 0.1435,
+      "step": 24367
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.105135963704704e-06,
+      "loss": 0.0977,
+      "step": 24368
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1044140908331577e-06,
+      "loss": 0.1256,
+      "step": 24369
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.103692286462862e-06,
+      "loss": 0.1206,
+      "step": 24370
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.10297055060099e-06,
+      "loss": 0.1102,
+      "step": 24371
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1022488832547093e-06,
+      "loss": 0.0798,
+      "step": 24372
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.101527284431186e-06,
+      "loss": 0.1016,
+      "step": 24373
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1008057541375937e-06,
+      "loss": 0.1141,
+      "step": 24374
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.100084292381095e-06,
+      "loss": 0.0705,
+      "step": 24375
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.099362899168863e-06,
+      "loss": 0.0953,
+      "step": 24376
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0986415745080602e-06,
+      "loss": 0.087,
+      "step": 24377
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.097920318405848e-06,
+      "loss": 0.1288,
+      "step": 24378
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.097199130869398e-06,
+      "loss": 0.0796,
+      "step": 24379
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0964780119058758e-06,
+      "loss": 0.0869,
+      "step": 24380
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0957569615224403e-06,
+      "loss": 0.1099,
+      "step": 24381
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0950359797262543e-06,
+      "loss": 0.1282,
+      "step": 24382
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.094315066524485e-06,
+      "loss": 0.0997,
+      "step": 24383
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0935942219242866e-06,
+      "loss": 0.1077,
+      "step": 24384
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.092873445932825e-06,
+      "loss": 0.0851,
+      "step": 24385
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.092152738557266e-06,
+      "loss": 0.1022,
+      "step": 24386
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0914320998047565e-06,
+      "loss": 0.0968,
+      "step": 24387
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0907115296824618e-06,
+      "loss": 0.1207,
+      "step": 24388
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.089991028197543e-06,
+      "loss": 0.1115,
+      "step": 24389
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.089270595357151e-06,
+      "loss": 0.0908,
+      "step": 24390
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.088550231168451e-06,
+      "loss": 0.1218,
+      "step": 24391
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.087829935638593e-06,
+      "loss": 0.0984,
+      "step": 24392
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0871097087747316e-06,
+      "loss": 0.1278,
+      "step": 24393
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.086389550584027e-06,
+      "loss": 0.0838,
+      "step": 24394
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.085669461073626e-06,
+      "loss": 0.1195,
+      "step": 24395
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0849494402506897e-06,
+      "loss": 0.1183,
+      "step": 24396
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0842294881223644e-06,
+      "loss": 0.149,
+      "step": 24397
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.083509604695809e-06,
+      "loss": 0.1135,
+      "step": 24398
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.082789789978167e-06,
+      "loss": 0.0894,
+      "step": 24399
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.082070043976597e-06,
+      "loss": 0.1226,
+      "step": 24400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0813503666982436e-06,
+      "loss": 0.1245,
+      "step": 24401
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0806307581502558e-06,
+      "loss": 0.0936,
+      "step": 24402
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.079911218339787e-06,
+      "loss": 0.0728,
+      "step": 24403
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0791917472739786e-06,
+      "loss": 0.1077,
+      "step": 24404
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.078472344959985e-06,
+      "loss": 0.131,
+      "step": 24405
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.077753011404946e-06,
+      "loss": 0.1138,
+      "step": 24406
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0770337466160138e-06,
+      "loss": 0.0999,
+      "step": 24407
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.076314550600328e-06,
+      "loss": 0.1252,
+      "step": 24408
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.075595423365039e-06,
+      "loss": 0.0929,
+      "step": 24409
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0748763649172865e-06,
+      "loss": 0.0781,
+      "step": 24410
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.074157375264212e-06,
+      "loss": 0.0878,
+      "step": 24411
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0734384544129637e-06,
+      "loss": 0.1266,
+      "step": 24412
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0727196023706773e-06,
+      "loss": 0.1478,
+      "step": 24413
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0720008191444963e-06,
+      "loss": 0.1293,
+      "step": 24414
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0712821047415655e-06,
+      "loss": 0.1768,
+      "step": 24415
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0705634591690216e-06,
+      "loss": 0.0738,
+      "step": 24416
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0698448824339987e-06,
+      "loss": 0.1271,
+      "step": 24417
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.069126374543643e-06,
+      "loss": 0.1016,
+      "step": 24418
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0684079355050855e-06,
+      "loss": 0.1186,
+      "step": 24419
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0676895653254703e-06,
+      "loss": 0.0906,
+      "step": 24420
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0669712640119287e-06,
+      "loss": 0.0938,
+      "step": 24421
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0662530315715954e-06,
+      "loss": 0.1111,
+      "step": 24422
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.065534868011607e-06,
+      "loss": 0.1119,
+      "step": 24423
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.064816773339102e-06,
+      "loss": 0.1022,
+      "step": 24424
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0640987475612104e-06,
+      "loss": 0.1062,
+      "step": 24425
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.063380790685061e-06,
+      "loss": 0.0765,
+      "step": 24426
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0626629027177946e-06,
+      "loss": 0.1305,
+      "step": 24427
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0619450836665345e-06,
+      "loss": 0.1311,
+      "step": 24428
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0612273335384148e-06,
+      "loss": 0.0957,
+      "step": 24429
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.060509652340572e-06,
+      "loss": 0.1079,
+      "step": 24430
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.059792040080125e-06,
+      "loss": 0.095,
+      "step": 24431
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.059074496764206e-06,
+      "loss": 0.0993,
+      "step": 24432
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.058357022399948e-06,
+      "loss": 0.0748,
+      "step": 24433
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0576396169944754e-06,
+      "loss": 0.1234,
+      "step": 24434
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0569222805549114e-06,
+      "loss": 0.1004,
+      "step": 24435
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.056205013088388e-06,
+      "loss": 0.0996,
+      "step": 24436
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.055487814602025e-06,
+      "loss": 0.128,
+      "step": 24437
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054770685102949e-06,
+      "loss": 0.116,
+      "step": 24438
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054053624598292e-06,
+      "loss": 0.088,
+      "step": 24439
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.053336633095164e-06,
+      "loss": 0.0923,
+      "step": 24440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0526197106006937e-06,
+      "loss": 0.1292,
+      "step": 24441
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0519028571220067e-06,
+      "loss": 0.0913,
+      "step": 24442
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.051186072666217e-06,
+      "loss": 0.1198,
+      "step": 24443
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.050469357240452e-06,
+      "loss": 0.1129,
+      "step": 24444
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0497527108518286e-06,
+      "loss": 0.1147,
+      "step": 24445
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.049036133507464e-06,
+      "loss": 0.1013,
+      "step": 24446
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0483196252144777e-06,
+      "loss": 0.1693,
+      "step": 24447
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0476031859799916e-06,
+      "loss": 0.09,
+      "step": 24448
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.046886815811121e-06,
+      "loss": 0.1294,
+      "step": 24449
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.046170514714977e-06,
+      "loss": 0.1402,
+      "step": 24450
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0454542826986835e-06,
+      "loss": 0.092,
+      "step": 24451
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.044738119769347e-06,
+      "loss": 0.1402,
+      "step": 24452
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.044022025934088e-06,
+      "loss": 0.0962,
+      "step": 24453
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.043306001200025e-06,
+      "loss": 0.0847,
+      "step": 24454
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0425900455742584e-06,
+      "loss": 0.1152,
+      "step": 24455
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.041874159063908e-06,
+      "loss": 0.088,
+      "step": 24456
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0411583416760858e-06,
+      "loss": 0.0976,
+      "step": 24457
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0404425934178993e-06,
+      "loss": 0.1365,
+      "step": 24458
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.039726914296465e-06,
+      "loss": 0.1481,
+      "step": 24459
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0390113043188875e-06,
+      "loss": 0.1269,
+      "step": 24460
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0382957634922737e-06,
+      "loss": 0.1348,
+      "step": 24461
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.037580291823735e-06,
+      "loss": 0.1301,
+      "step": 24462
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0368648893203857e-06,
+      "loss": 0.1088,
+      "step": 24463
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0361495559893183e-06,
+      "loss": 0.1277,
+      "step": 24464
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0354342918376467e-06,
+      "loss": 0.0919,
+      "step": 24465
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0347190968724793e-06,
+      "loss": 0.1398,
+      "step": 24466
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.034003971100915e-06,
+      "loss": 0.1234,
+      "step": 24467
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0332889145300635e-06,
+      "loss": 0.1153,
+      "step": 24468
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0325739271670264e-06,
+      "loss": 0.1186,
+      "step": 24469
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0318590090189003e-06,
+      "loss": 0.0897,
+      "step": 24470
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.031144160092794e-06,
+      "loss": 0.0696,
+      "step": 24471
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0304293803958097e-06,
+      "loss": 0.0765,
+      "step": 24472
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0297146699350453e-06,
+      "loss": 0.1139,
+      "step": 24473
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.029000028717598e-06,
+      "loss": 0.0768,
+      "step": 24474
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0282854567505737e-06,
+      "loss": 0.1112,
+      "step": 24475
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0275709540410636e-06,
+      "loss": 0.0882,
+      "step": 24476
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.026856520596174e-06,
+      "loss": 0.1276,
+      "step": 24477
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0261421564229966e-06,
+      "loss": 0.0936,
+      "step": 24478
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.025427861528626e-06,
+      "loss": 0.1114,
+      "step": 24479
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.024713635920161e-06,
+      "loss": 0.1321,
+      "step": 24480
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0239994796047e-06,
+      "loss": 0.0828,
+      "step": 24481
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.023285392589331e-06,
+      "loss": 0.0874,
+      "step": 24482
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0225713748811546e-06,
+      "loss": 0.104,
+      "step": 24483
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.02185742648726e-06,
+      "loss": 0.1112,
+      "step": 24484
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0211435474147364e-06,
+      "loss": 0.113,
+      "step": 24485
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0204297376706838e-06,
+      "loss": 0.1106,
+      "step": 24486
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0197159972621838e-06,
+      "loss": 0.1564,
+      "step": 24487
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0190023261963354e-06,
+      "loss": 0.1054,
+      "step": 24488
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0182887244802204e-06,
+      "loss": 0.1205,
+      "step": 24489
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0175751921209363e-06,
+      "loss": 0.109,
+      "step": 24490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.016861729125563e-06,
+      "loss": 0.0919,
+      "step": 24491
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0161483355011944e-06,
+      "loss": 0.0769,
+      "step": 24492
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.015435011254916e-06,
+      "loss": 0.0993,
+      "step": 24493
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.014721756393809e-06,
+      "loss": 0.1135,
+      "step": 24494
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0140085709249666e-06,
+      "loss": 0.1087,
+      "step": 24495
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.013295454855466e-06,
+      "loss": 0.1093,
+      "step": 24496
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.012582408192399e-06,
+      "loss": 0.0864,
+      "step": 24497
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.011869430942842e-06,
+      "loss": 0.1482,
+      "step": 24498
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.011156523113885e-06,
+      "loss": 0.1076,
+      "step": 24499
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0104436847126017e-06,
+      "loss": 0.1067,
+      "step": 24500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.009730915746082e-06,
+      "loss": 0.1001,
+      "step": 24501
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.009018216221402e-06,
+      "loss": 0.1155,
+      "step": 24502
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.008305586145639e-06,
+      "loss": 0.0783,
+      "step": 24503
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.007593025525879e-06,
+      "loss": 0.1191,
+      "step": 24504
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0068805343691944e-06,
+      "loss": 0.0988,
+      "step": 24505
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0061681126826646e-06,
+      "loss": 0.1344,
+      "step": 24506
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0054557604733727e-06,
+      "loss": 0.1026,
+      "step": 24507
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.00474347774839e-06,
+      "loss": 0.0993,
+      "step": 24508
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0040312645147896e-06,
+      "loss": 0.1169,
+      "step": 24509
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.003319120779654e-06,
+      "loss": 0.0925,
+      "step": 24510
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.002607046550051e-06,
+      "loss": 0.0765,
+      "step": 24511
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0018950418330596e-06,
+      "loss": 0.0841,
+      "step": 24512
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0011831066357512e-06,
+      "loss": 0.1045,
+      "step": 24513
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0004712409651947e-06,
+      "loss": 0.1149,
+      "step": 24514
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9997594448284638e-06,
+      "loss": 0.0969,
+      "step": 24515
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.999047718232634e-06,
+      "loss": 0.1191,
+      "step": 24516
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.998336061184772e-06,
+      "loss": 0.0996,
+      "step": 24517
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9976244736919447e-06,
+      "loss": 0.1601,
+      "step": 24518
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9969129557612265e-06,
+      "loss": 0.0988,
+      "step": 24519
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.996201507399681e-06,
+      "loss": 0.111,
+      "step": 24520
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9954901286143802e-06,
+      "loss": 0.1183,
+      "step": 24521
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.994778819412388e-06,
+      "loss": 0.1054,
+      "step": 24522
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.994067579800769e-06,
+      "loss": 0.1046,
+      "step": 24523
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.993356409786591e-06,
+      "loss": 0.0791,
+      "step": 24524
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9926453093769215e-06,
+      "loss": 0.0792,
+      "step": 24525
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.991934278578822e-06,
+      "loss": 0.1006,
+      "step": 24526
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.991223317399352e-06,
+      "loss": 0.1041,
+      "step": 24527
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.990512425845582e-06,
+      "loss": 0.0991,
+      "step": 24528
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.989801603924567e-06,
+      "loss": 0.1156,
+      "step": 24529
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9890908516433713e-06,
+      "loss": 0.1053,
+      "step": 24530
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.988380169009062e-06,
+      "loss": 0.1328,
+      "step": 24531
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.987669556028687e-06,
+      "loss": 0.1016,
+      "step": 24532
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.986959012709311e-06,
+      "loss": 0.1217,
+      "step": 24533
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9862485390579954e-06,
+      "loss": 0.079,
+      "step": 24534
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9855381350817937e-06,
+      "loss": 0.1224,
+      "step": 24535
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.984827800787767e-06,
+      "loss": 0.1014,
+      "step": 24536
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.984117536182971e-06,
+      "loss": 0.0949,
+      "step": 24537
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.983407341274457e-06,
+      "loss": 0.1363,
+      "step": 24538
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9826972160692825e-06,
+      "loss": 0.1215,
+      "step": 24539
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9819871605745077e-06,
+      "loss": 0.0979,
+      "step": 24540
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.98127717479718e-06,
+      "loss": 0.0918,
+      "step": 24541
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9805672587443526e-06,
+      "loss": 0.1406,
+      "step": 24542
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9798574124230804e-06,
+      "loss": 0.0988,
+      "step": 24543
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9791476358404126e-06,
+      "loss": 0.0913,
+      "step": 24544
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9784379290034038e-06,
+      "loss": 0.1065,
+      "step": 24545
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.977728291919103e-06,
+      "loss": 0.108,
+      "step": 24546
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9770187245945546e-06,
+      "loss": 0.0926,
+      "step": 24547
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.976309227036812e-06,
+      "loss": 0.1356,
+      "step": 24548
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9755997992529263e-06,
+      "loss": 0.1096,
+      "step": 24549
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9748904412499425e-06,
+      "loss": 0.1188,
+      "step": 24550
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.974181153034903e-06,
+      "loss": 0.1336,
+      "step": 24551
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.97347193461486e-06,
+      "loss": 0.145,
+      "step": 24552
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9727627859968546e-06,
+      "loss": 0.1231,
+      "step": 24553
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9720537071879365e-06,
+      "loss": 0.1465,
+      "step": 24554
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9713446981951466e-06,
+      "loss": 0.129,
+      "step": 24555
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9706357590255243e-06,
+      "loss": 0.1034,
+      "step": 24556
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9699268896861176e-06,
+      "loss": 0.0718,
+      "step": 24557
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9692180901839697e-06,
+      "loss": 0.1078,
+      "step": 24558
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.968509360526116e-06,
+      "loss": 0.1157,
+      "step": 24559
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.967800700719605e-06,
+      "loss": 0.1266,
+      "step": 24560
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.967092110771471e-06,
+      "loss": 0.1219,
+      "step": 24561
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9663835906887505e-06,
+      "loss": 0.1143,
+      "step": 24562
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.96567514047849e-06,
+      "loss": 0.1161,
+      "step": 24563
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9649667601477193e-06,
+      "loss": 0.1082,
+      "step": 24564
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.964258449703482e-06,
+      "loss": 0.0966,
+      "step": 24565
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.963550209152809e-06,
+      "loss": 0.0756,
+      "step": 24566
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9628420385027425e-06,
+      "loss": 0.1228,
+      "step": 24567
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9621339377603096e-06,
+      "loss": 0.088,
+      "step": 24568
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9614259069325522e-06,
+      "loss": 0.0733,
+      "step": 24569
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9607179460265002e-06,
+      "loss": 0.1137,
+      "step": 24570
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.960010055049184e-06,
+      "loss": 0.1036,
+      "step": 24571
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.959302234007643e-06,
+      "loss": 0.0711,
+      "step": 24572
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9585944829089e-06,
+      "loss": 0.0905,
+      "step": 24573
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9578868017599927e-06,
+      "loss": 0.116,
+      "step": 24574
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.957179190567947e-06,
+      "loss": 0.093,
+      "step": 24575
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.956471649339796e-06,
+      "loss": 0.1492,
+      "step": 24576
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.955764178082564e-06,
+      "loss": 0.1218,
+      "step": 24577
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9550567768032856e-06,
+      "loss": 0.0974,
+      "step": 24578
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9543494455089837e-06,
+      "loss": 0.1377,
+      "step": 24579
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9536421842066833e-06,
+      "loss": 0.1228,
+      "step": 24580
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9529349929034144e-06,
+      "loss": 0.1054,
+      "step": 24581
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.952227871606198e-06,
+      "loss": 0.1132,
+      "step": 24582
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.951520820322061e-06,
+      "loss": 0.1,
+      "step": 24583
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.950813839058031e-06,
+      "loss": 0.0944,
+      "step": 24584
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.950106927821128e-06,
+      "loss": 0.0938,
+      "step": 24585
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9494000866183715e-06,
+      "loss": 0.0968,
+      "step": 24586
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9486933154567877e-06,
+      "loss": 0.1078,
+      "step": 24587
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9479866143433943e-06,
+      "loss": 0.1256,
+      "step": 24588
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.947279983285216e-06,
+      "loss": 0.0722,
+      "step": 24589
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9465734222892696e-06,
+      "loss": 0.1284,
+      "step": 24590
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.945866931362572e-06,
+      "loss": 0.069,
+      "step": 24591
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.945160510512144e-06,
+      "loss": 0.1399,
+      "step": 24592
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.944454159745007e-06,
+      "loss": 0.0716,
+      "step": 24593
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.943747879068174e-06,
+      "loss": 0.0829,
+      "step": 24594
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9430416684886575e-06,
+      "loss": 0.1132,
+      "step": 24595
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9423355280134813e-06,
+      "loss": 0.0826,
+      "step": 24596
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9416294576496518e-06,
+      "loss": 0.1112,
+      "step": 24597
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.940923457404188e-06,
+      "loss": 0.1,
+      "step": 24598
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9402175272841092e-06,
+      "loss": 0.1232,
+      "step": 24599
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9395116672964154e-06,
+      "loss": 0.1338,
+      "step": 24600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9388058774481242e-06,
+      "loss": 0.1067,
+      "step": 24601
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9381001577462508e-06,
+      "loss": 0.1308,
+      "step": 24602
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.937394508197804e-06,
+      "loss": 0.1178,
+      "step": 24603
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.936688928809788e-06,
+      "loss": 0.0922,
+      "step": 24604
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.93598341958922e-06,
+      "loss": 0.0938,
+      "step": 24605
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.935277980543103e-06,
+      "loss": 0.0931,
+      "step": 24606
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.934572611678447e-06,
+      "loss": 0.0794,
+      "step": 24607
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9338673130022644e-06,
+      "loss": 0.0876,
+      "step": 24608
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9331620845215515e-06,
+      "loss": 0.1106,
+      "step": 24609
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.932456926243319e-06,
+      "loss": 0.072,
+      "step": 24610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.931751838174575e-06,
+      "loss": 0.0989,
+      "step": 24611
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9310468203223176e-06,
+      "loss": 0.1243,
+      "step": 24612
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9303418726935573e-06,
+      "loss": 0.08,
+      "step": 24613
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.929636995295293e-06,
+      "loss": 0.1482,
+      "step": 24614
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.1034,
+      "step": 24615
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9282274512182583e-06,
+      "loss": 0.0939,
+      "step": 24616
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9275227845534958e-06,
+      "loss": 0.093,
+      "step": 24617
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.926818188147236e-06,
+      "loss": 0.1198,
+      "step": 24618
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9261136620064735e-06,
+      "loss": 0.1203,
+      "step": 24619
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9254092061382143e-06,
+      "loss": 0.1023,
+      "step": 24620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9247048205494498e-06,
+      "loss": 0.1003,
+      "step": 24621
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.924000505247182e-06,
+      "loss": 0.1114,
+      "step": 24622
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.923296260238412e-06,
+      "loss": 0.125,
+      "step": 24623
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.922592085530124e-06,
+      "loss": 0.123,
+      "step": 24624
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9218879811293198e-06,
+      "loss": 0.085,
+      "step": 24625
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.921183947042997e-06,
+      "loss": 0.0977,
+      "step": 24626
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9204799832781438e-06,
+      "loss": 0.0973,
+      "step": 24627
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9197760898417593e-06,
+      "loss": 0.1448,
+      "step": 24628
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.919072266740832e-06,
+      "loss": 0.0794,
+      "step": 24629
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9183685139823527e-06,
+      "loss": 0.1216,
+      "step": 24630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9176648315733126e-06,
+      "loss": 0.0977,
+      "step": 24631
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9169612195207108e-06,
+      "loss": 0.0948,
+      "step": 24632
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9162576778315242e-06,
+      "loss": 0.092,
+      "step": 24633
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9155542065127474e-06,
+      "loss": 0.1476,
+      "step": 24634
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9148508055713733e-06,
+      "loss": 0.0706,
+      "step": 24635
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9141474750143806e-06,
+      "loss": 0.1186,
+      "step": 24636
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.913444214848765e-06,
+      "loss": 0.1123,
+      "step": 24637
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.912741025081508e-06,
+      "loss": 0.1288,
+      "step": 24638
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9120379057195934e-06,
+      "loss": 0.1083,
+      "step": 24639
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9113348567700082e-06,
+      "loss": 0.0809,
+      "step": 24640
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9106318782397403e-06,
+      "loss": 0.1053,
+      "step": 24641
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.909928970135769e-06,
+      "loss": 0.1342,
+      "step": 24642
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.909226132465076e-06,
+      "loss": 0.1236,
+      "step": 24643
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.908523365234647e-06,
+      "loss": 0.1079,
+      "step": 24644
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9078206684514575e-06,
+      "loss": 0.1206,
+      "step": 24645
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9071180421224963e-06,
+      "loss": 0.1196,
+      "step": 24646
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.906415486254739e-06,
+      "loss": 0.1062,
+      "step": 24647
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9057130008551616e-06,
+      "loss": 0.1096,
+      "step": 24648
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9050105859307464e-06,
+      "loss": 0.1276,
+      "step": 24649
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9043082414884725e-06,
+      "loss": 0.0991,
+      "step": 24650
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9036059675353136e-06,
+      "loss": 0.1646,
+      "step": 24651
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.902903764078251e-06,
+      "loss": 0.099,
+      "step": 24652
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9022016311242564e-06,
+      "loss": 0.1068,
+      "step": 24653
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9014995686803028e-06,
+      "loss": 0.1013,
+      "step": 24654
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9007975767533714e-06,
+      "loss": 0.1304,
+      "step": 24655
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9000956553504277e-06,
+      "loss": 0.1652,
+      "step": 24656
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.899393804478453e-06,
+      "loss": 0.1023,
+      "step": 24657
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8986920241444117e-06,
+      "loss": 0.1147,
+      "step": 24658
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8979903143552823e-06,
+      "loss": 0.1367,
+      "step": 24659
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8972886751180284e-06,
+      "loss": 0.0876,
+      "step": 24660
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.896587106439629e-06,
+      "loss": 0.1034,
+      "step": 24661
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.895885608327047e-06,
+      "loss": 0.1019,
+      "step": 24662
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8951841807872507e-06,
+      "loss": 0.1058,
+      "step": 24663
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.894482823827213e-06,
+      "loss": 0.1248,
+      "step": 24664
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.893781537453895e-06,
+      "loss": 0.1173,
+      "step": 24665
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.89308032167427e-06,
+      "loss": 0.1013,
+      "step": 24666
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.892379176495298e-06,
+      "loss": 0.0829,
+      "step": 24667
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8916781019239493e-06,
+      "loss": 0.0976,
+      "step": 24668
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8909770979671835e-06,
+      "loss": 0.1127,
+      "step": 24669
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.89027616463197e-06,
+      "loss": 0.1213,
+      "step": 24670
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.889575301925268e-06,
+      "loss": 0.14,
+      "step": 24671
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8888745098540393e-06,
+      "loss": 0.0843,
+      "step": 24672
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.888173788425249e-06,
+      "loss": 0.1093,
+      "step": 24673
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8874731376458543e-06,
+      "loss": 0.0947,
+      "step": 24674
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8867725575228166e-06,
+      "loss": 0.1179,
+      "step": 24675
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8860720480630997e-06,
+      "loss": 0.1302,
+      "step": 24676
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8853716092736596e-06,
+      "loss": 0.0947,
+      "step": 24677
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.88467124116145e-06,
+      "loss": 0.0693,
+      "step": 24678
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.883970943733436e-06,
+      "loss": 0.1195,
+      "step": 24679
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8832707169965678e-06,
+      "loss": 0.1289,
+      "step": 24680
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.882570560957808e-06,
+      "loss": 0.1015,
+      "step": 24681
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8818704756241087e-06,
+      "loss": 0.1046,
+      "step": 24682
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8811704610024216e-06,
+      "loss": 0.0672,
+      "step": 24683
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.880470517099703e-06,
+      "loss": 0.1036,
+      "step": 24684
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.87977064392291e-06,
+      "loss": 0.1652,
+      "step": 24685
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8790708414789926e-06,
+      "loss": 0.1261,
+      "step": 24686
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.878371109774898e-06,
+      "loss": 0.1454,
+      "step": 24687
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.877671448817585e-06,
+      "loss": 0.1161,
+      "step": 24688
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8769718586139984e-06,
+      "loss": 0.1337,
+      "step": 24689
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8762723391710922e-06,
+      "loss": 0.1138,
+      "step": 24690
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8755728904958135e-06,
+      "loss": 0.0871,
+      "step": 24691
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.874873512595108e-06,
+      "loss": 0.1034,
+      "step": 24692
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.874174205475925e-06,
+      "loss": 0.1207,
+      "step": 24693
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8734749691452156e-06,
+      "loss": 0.0934,
+      "step": 24694
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.872775803609924e-06,
+      "loss": 0.0977,
+      "step": 24695
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8720767088769906e-06,
+      "loss": 0.0671,
+      "step": 24696
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8713776849533683e-06,
+      "loss": 0.0965,
+      "step": 24697
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8706787318459927e-06,
+      "loss": 0.1053,
+      "step": 24698
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8699798495618125e-06,
+      "loss": 0.0818,
+      "step": 24699
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8692810381077763e-06,
+      "loss": 0.1076,
+      "step": 24700
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.868582297490813e-06,
+      "loss": 0.1378,
+      "step": 24701
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.867883627717871e-06,
+      "loss": 0.0771,
+      "step": 24702
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8671850287958935e-06,
+      "loss": 0.1133,
+      "step": 24703
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.866486500731814e-06,
+      "loss": 0.0878,
+      "step": 24704
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8657880435325793e-06,
+      "loss": 0.0945,
+      "step": 24705
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8650896572051225e-06,
+      "loss": 0.0875,
+      "step": 24706
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8643913417563795e-06,
+      "loss": 0.0978,
+      "step": 24707
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.863693097193291e-06,
+      "loss": 0.1011,
+      "step": 24708
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.862994923522796e-06,
+      "loss": 0.1455,
+      "step": 24709
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8622968207518277e-06,
+      "loss": 0.1029,
+      "step": 24710
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8615987888873175e-06,
+      "loss": 0.1213,
+      "step": 24711
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.860900827936205e-06,
+      "loss": 0.0794,
+      "step": 24712
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.86020293790542e-06,
+      "loss": 0.1159,
+      "step": 24713
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8595051188018984e-06,
+      "loss": 0.1181,
+      "step": 24714
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8588073706325726e-06,
+      "loss": 0.1244,
+      "step": 24715
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8581096934043686e-06,
+      "loss": 0.103,
+      "step": 24716
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8574120871242206e-06,
+      "loss": 0.1149,
+      "step": 24717
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8567145517990626e-06,
+      "loss": 0.0642,
+      "step": 24718
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8560170874358206e-06,
+      "loss": 0.0948,
+      "step": 24719
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.855319694041421e-06,
+      "loss": 0.1179,
+      "step": 24720
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.854622371622795e-06,
+      "loss": 0.1128,
+      "step": 24721
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.853925120186867e-06,
+      "loss": 0.1083,
+      "step": 24722
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8532279397405684e-06,
+      "loss": 0.099,
+      "step": 24723
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8525308302908216e-06,
+      "loss": 0.0818,
+      "step": 24724
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8518337918445484e-06,
+      "loss": 0.137,
+      "step": 24725
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.851136824408678e-06,
+      "loss": 0.1029,
+      "step": 24726
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8504399279901353e-06,
+      "loss": 0.1099,
+      "step": 24727
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.849743102595839e-06,
+      "loss": 0.0983,
+      "step": 24728
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8490463482327157e-06,
+      "loss": 0.1225,
+      "step": 24729
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.848349664907685e-06,
+      "loss": 0.1237,
+      "step": 24730
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8476530526276648e-06,
+      "loss": 0.0612,
+      "step": 24731
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.84695651139958e-06,
+      "loss": 0.0958,
+      "step": 24732
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8462600412303463e-06,
+      "loss": 0.0967,
+      "step": 24733
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.845563642126887e-06,
+      "loss": 0.0981,
+      "step": 24734
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.844867314096115e-06,
+      "loss": 0.12,
+      "step": 24735
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.844171057144952e-06,
+      "loss": 0.1085,
+      "step": 24736
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8434748712803107e-06,
+      "loss": 0.0976,
+      "step": 24737
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8427787565091127e-06,
+      "loss": 0.1124,
+      "step": 24738
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.842082712838269e-06,
+      "loss": 0.0703,
+      "step": 24739
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.841386740274692e-06,
+      "loss": 0.1129,
+      "step": 24740
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8406908388253017e-06,
+      "loss": 0.0946,
+      "step": 24741
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.839995008497005e-06,
+      "loss": 0.1003,
+      "step": 24742
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.839299249296721e-06,
+      "loss": 0.1081,
+      "step": 24743
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8386035612313534e-06,
+      "loss": 0.086,
+      "step": 24744
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8379079443078215e-06,
+      "loss": 0.124,
+      "step": 24745
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.837212398533029e-06,
+      "loss": 0.1353,
+      "step": 24746
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.83651692391389e-06,
+      "loss": 0.0892,
+      "step": 24747
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8358215204573124e-06,
+      "loss": 0.0917,
+      "step": 24748
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.835126188170201e-06,
+      "loss": 0.1284,
+      "step": 24749
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8344309270594685e-06,
+      "loss": 0.0935,
+      "step": 24750
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.833735737132015e-06,
+      "loss": 0.1095,
+      "step": 24751
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8330406183947525e-06,
+      "loss": 0.1246,
+      "step": 24752
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8323455708545854e-06,
+      "loss": 0.129,
+      "step": 24753
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8316505945184194e-06,
+      "loss": 0.1006,
+      "step": 24754
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8309556893931522e-06,
+      "loss": 0.1342,
+      "step": 24755
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.830260855485695e-06,
+      "loss": 0.1147,
+      "step": 24756
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.829566092802942e-06,
+      "loss": 0.1391,
+      "step": 24757
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.828871401351805e-06,
+      "loss": 0.1003,
+      "step": 24758
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.828176781139178e-06,
+      "loss": 0.1218,
+      "step": 24759
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8274822321719608e-06,
+      "loss": 0.1177,
+      "step": 24760
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8267877544570556e-06,
+      "loss": 0.0891,
+      "step": 24761
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.826093348001364e-06,
+      "loss": 0.1227,
+      "step": 24762
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8253990128117814e-06,
+      "loss": 0.0939,
+      "step": 24763
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.824704748895203e-06,
+      "loss": 0.1012,
+      "step": 24764
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8240105562585307e-06,
+      "loss": 0.1262,
+      "step": 24765
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.823316434908656e-06,
+      "loss": 0.1149,
+      "step": 24766
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8226223848524757e-06,
+      "loss": 0.1122,
+      "step": 24767
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.821928406096892e-06,
+      "loss": 0.1444,
+      "step": 24768
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.821234498648786e-06,
+      "loss": 0.1217,
+      "step": 24769
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.820540662515057e-06,
+      "loss": 0.1148,
+      "step": 24770
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8198468977026006e-06,
+      "loss": 0.0799,
+      "step": 24771
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8191532042183068e-06,
+      "loss": 0.119,
+      "step": 24772
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.818459582069063e-06,
+      "loss": 0.097,
+      "step": 24773
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.817766031261766e-06,
+      "loss": 0.1122,
+      "step": 24774
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8170725518032983e-06,
+      "loss": 0.0812,
+      "step": 24775
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.816379143700554e-06,
+      "loss": 0.0901,
+      "step": 24776
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.815685806960425e-06,
+      "loss": 0.0781,
+      "step": 24777
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8149925415897904e-06,
+      "loss": 0.1185,
+      "step": 24778
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8142993475955393e-06,
+      "loss": 0.0969,
+      "step": 24779
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.813606224984564e-06,
+      "loss": 0.1097,
+      "step": 24780
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8129131737637425e-06,
+      "loss": 0.0597,
+      "step": 24781
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8122201939399663e-06,
+      "loss": 0.1122,
+      "step": 24782
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.811527285520116e-06,
+      "loss": 0.1245,
+      "step": 24783
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8108344485110716e-06,
+      "loss": 0.0773,
+      "step": 24784
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8101416829197204e-06,
+      "loss": 0.1404,
+      "step": 24785
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8094489887529454e-06,
+      "loss": 0.113,
+      "step": 24786
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8087563660176263e-06,
+      "loss": 0.0685,
+      "step": 24787
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8080638147206396e-06,
+      "loss": 0.0997,
+      "step": 24788
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8073713348688724e-06,
+      "loss": 0.1076,
+      "step": 24789
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8066789264691975e-06,
+      "loss": 0.1054,
+      "step": 24790
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8059865895284955e-06,
+      "loss": 0.0635,
+      "step": 24791
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.80529432405365e-06,
+      "loss": 0.1379,
+      "step": 24792
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.804602130051528e-06,
+      "loss": 0.1392,
+      "step": 24793
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8039100075290093e-06,
+      "loss": 0.1045,
+      "step": 24794
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.803217956492974e-06,
+      "loss": 0.1383,
+      "step": 24795
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.802525976950291e-06,
+      "loss": 0.0914,
+      "step": 24796
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8018340689078392e-06,
+      "loss": 0.1204,
+      "step": 24797
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8011422323724913e-06,
+      "loss": 0.0844,
+      "step": 24798
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8004504673511156e-06,
+      "loss": 0.1033,
+      "step": 24799
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.799758773850586e-06,
+      "loss": 0.1401,
+      "step": 24800
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7990671518777823e-06,
+      "loss": 0.0855,
+      "step": 24801
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7983756014395615e-06,
+      "loss": 0.1023,
+      "step": 24802
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7976841225427996e-06,
+      "loss": 0.0934,
+      "step": 24803
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.79699271519437e-06,
+      "loss": 0.1184,
+      "step": 24804
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.796301379401134e-06,
+      "loss": 0.1144,
+      "step": 24805
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7956101151699655e-06,
+      "loss": 0.1043,
+      "step": 24806
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.794918922507729e-06,
+      "loss": 0.0786,
+      "step": 24807
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7942278014212877e-06,
+      "loss": 0.1093,
+      "step": 24808
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7935367519175104e-06,
+      "loss": 0.0776,
+      "step": 24809
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7928457740032645e-06,
+      "loss": 0.0721,
+      "step": 24810
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7921548676854114e-06,
+      "loss": 0.0958,
+      "step": 24811
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.791464032970812e-06,
+      "loss": 0.0808,
+      "step": 24812
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7907732698663347e-06,
+      "loss": 0.082,
+      "step": 24813
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7900825783788366e-06,
+      "loss": 0.0844,
+      "step": 24814
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.789391958515183e-06,
+      "loss": 0.1092,
+      "step": 24815
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7887014102822342e-06,
+      "loss": 0.1198,
+      "step": 24816
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.788010933686844e-06,
+      "loss": 0.0925,
+      "step": 24817
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.787320528735877e-06,
+      "loss": 0.1476,
+      "step": 24818
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7866301954361954e-06,
+      "loss": 0.1098,
+      "step": 24819
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7859399337946492e-06,
+      "loss": 0.1132,
+      "step": 24820
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7852497438181014e-06,
+      "loss": 0.1003,
+      "step": 24821
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7845596255134065e-06,
+      "loss": 0.0803,
+      "step": 24822
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.783869578887417e-06,
+      "loss": 0.0885,
+      "step": 24823
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7831796039469927e-06,
+      "loss": 0.1279,
+      "step": 24824
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.782489700698984e-06,
+      "loss": 0.1367,
+      "step": 24825
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.781799869150248e-06,
+      "loss": 0.1059,
+      "step": 24826
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7811101093076344e-06,
+      "loss": 0.1015,
+      "step": 24827
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7804204211779984e-06,
+      "loss": 0.0738,
+      "step": 24828
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.779730804768187e-06,
+      "loss": 0.1086,
+      "step": 24829
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.779041260085057e-06,
+      "loss": 0.0862,
+      "step": 24830
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.778351787135455e-06,
+      "loss": 0.0964,
+      "step": 24831
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7776623859262262e-06,
+      "loss": 0.117,
+      "step": 24832
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7769730564642274e-06,
+      "loss": 0.0773,
+      "step": 24833
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7762837987562987e-06,
+      "loss": 0.0984,
+      "step": 24834
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7755946128092935e-06,
+      "loss": 0.0938,
+      "step": 24835
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7749054986300537e-06,
+      "loss": 0.1041,
+      "step": 24836
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7742164562254294e-06,
+      "loss": 0.1087,
+      "step": 24837
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7735274856022598e-06,
+      "loss": 0.0869,
+      "step": 24838
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7728385867673957e-06,
+      "loss": 0.086,
+      "step": 24839
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.772149759727677e-06,
+      "loss": 0.1199,
+      "step": 24840
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.771461004489944e-06,
+      "loss": 0.1068,
+      "step": 24841
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770772321061045e-06,
+      "loss": 0.1367,
+      "step": 24842
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770083709447815e-06,
+      "loss": 0.113,
+      "step": 24843
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7693951696570988e-06,
+      "loss": 0.1075,
+      "step": 24844
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7687067016957413e-06,
+      "loss": 0.0869,
+      "step": 24845
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.768018305570569e-06,
+      "loss": 0.1222,
+      "step": 24846
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.767329981288429e-06,
+      "loss": 0.093,
+      "step": 24847
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.76664172885616e-06,
+      "loss": 0.096,
+      "step": 24848
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.765953548280593e-06,
+      "loss": 0.1322,
+      "step": 24849
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.765265439568573e-06,
+      "loss": 0.0955,
+      "step": 24850
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.76457740272693e-06,
+      "loss": 0.1028,
+      "step": 24851
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.763889437762497e-06,
+      "loss": 0.1188,
+      "step": 24852
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7632015446821116e-06,
+      "loss": 0.125,
+      "step": 24853
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7625137234926104e-06,
+      "loss": 0.1051,
+      "step": 24854
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7618259742008213e-06,
+      "loss": 0.1169,
+      "step": 24855
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.761138296813577e-06,
+      "loss": 0.0931,
+      "step": 24856
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7604506913377125e-06,
+      "loss": 0.1278,
+      "step": 24857
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.759763157780052e-06,
+      "loss": 0.1212,
+      "step": 24858
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.759075696147434e-06,
+      "loss": 0.0975,
+      "step": 24859
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.758388306446683e-06,
+      "loss": 0.1336,
+      "step": 24860
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.757700988684625e-06,
+      "loss": 0.1038,
+      "step": 24861
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.75701374286809e-06,
+      "loss": 0.1088,
+      "step": 24862
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.75632656900391e-06,
+      "loss": 0.0848,
+      "step": 24863
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.755639467098906e-06,
+      "loss": 0.0826,
+      "step": 24864
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7549524371599033e-06,
+      "loss": 0.1024,
+      "step": 24865
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.754265479193732e-06,
+      "loss": 0.0846,
+      "step": 24866
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.753578593207209e-06,
+      "loss": 0.0813,
+      "step": 24867
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.752891779207163e-06,
+      "loss": 0.0938,
+      "step": 24868
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7522050372004216e-06,
+      "loss": 0.1132,
+      "step": 24869
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.751518367193795e-06,
+      "loss": 0.1152,
+      "step": 24870
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.750831769194111e-06,
+      "loss": 0.0854,
+      "step": 24871
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7501452432081923e-06,
+      "loss": 0.1326,
+      "step": 24872
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.749458789242855e-06,
+      "loss": 0.0938,
+      "step": 24873
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7487724073049215e-06,
+      "loss": 0.1139,
+      "step": 24874
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.74808609740121e-06,
+      "loss": 0.1013,
+      "step": 24875
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7473998595385353e-06,
+      "loss": 0.082,
+      "step": 24876
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.746713693723715e-06,
+      "loss": 0.1301,
+      "step": 24877
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.746027599963571e-06,
+      "loss": 0.1133,
+      "step": 24878
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.745341578264915e-06,
+      "loss": 0.1078,
+      "step": 24879
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7446556286345583e-06,
+      "loss": 0.0807,
+      "step": 24880
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.743969751079323e-06,
+      "loss": 0.1199,
+      "step": 24881
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7432839456060145e-06,
+      "loss": 0.1151,
+      "step": 24882
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7425982122214535e-06,
+      "loss": 0.1328,
+      "step": 24883
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7419125509324485e-06,
+      "loss": 0.1364,
+      "step": 24884
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7412269617458074e-06,
+      "loss": 0.1176,
+      "step": 24885
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7405414446683444e-06,
+      "loss": 0.0808,
+      "step": 24886
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.739855999706873e-06,
+      "loss": 0.0792,
+      "step": 24887
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7391706268681985e-06,
+      "loss": 0.1082,
+      "step": 24888
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.738485326159126e-06,
+      "loss": 0.0911,
+      "step": 24889
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7378000975864705e-06,
+      "loss": 0.1139,
+      "step": 24890
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.737114941157033e-06,
+      "loss": 0.0844,
+      "step": 24891
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.736429856877625e-06,
+      "loss": 0.1151,
+      "step": 24892
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7357448447550506e-06,
+      "loss": 0.0843,
+      "step": 24893
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7350599047961103e-06,
+      "loss": 0.1212,
+      "step": 24894
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7343750370076127e-06,
+      "loss": 0.116,
+      "step": 24895
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.733690241396363e-06,
+      "loss": 0.1047,
+      "step": 24896
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7330055179691585e-06,
+      "loss": 0.1263,
+      "step": 24897
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7323208667328073e-06,
+      "loss": 0.1043,
+      "step": 24898
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.731636287694108e-06,
+      "loss": 0.0961,
+      "step": 24899
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7309517808598585e-06,
+      "loss": 0.0768,
+      "step": 24900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7302673462368646e-06,
+      "loss": 0.1118,
+      "step": 24901
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.729582983831919e-06,
+      "loss": 0.1502,
+      "step": 24902
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.728898693651826e-06,
+      "loss": 0.075,
+      "step": 24903
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.728214475703379e-06,
+      "loss": 0.1153,
+      "step": 24904
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7275303299933786e-06,
+      "loss": 0.1047,
+      "step": 24905
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.726846256528618e-06,
+      "loss": 0.1238,
+      "step": 24906
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.726162255315896e-06,
+      "loss": 0.1127,
+      "step": 24907
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7254783263620066e-06,
+      "loss": 0.0968,
+      "step": 24908
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7247944696737405e-06,
+      "loss": 0.1229,
+      "step": 24909
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7241106852578957e-06,
+      "loss": 0.0847,
+      "step": 24910
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7234269731212616e-06,
+      "loss": 0.0771,
+      "step": 24911
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7227433332706344e-06,
+      "loss": 0.1041,
+      "step": 24912
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7220597657128e-06,
+      "loss": 0.126,
+      "step": 24913
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7213762704545543e-06,
+      "loss": 0.1011,
+      "step": 24914
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7206928475026828e-06,
+      "loss": 0.1068,
+      "step": 24915
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.720009496863979e-06,
+      "loss": 0.1149,
+      "step": 24916
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7193262185452283e-06,
+      "loss": 0.1041,
+      "step": 24917
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7186430125532183e-06,
+      "loss": 0.0749,
+      "step": 24918
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7179598788947383e-06,
+      "loss": 0.1154,
+      "step": 24919
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7172768175765705e-06,
+      "loss": 0.0898,
+      "step": 24920
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.716593828605503e-06,
+      "loss": 0.1164,
+      "step": 24921
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7159109119883244e-06,
+      "loss": 0.1165,
+      "step": 24922
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.715228067731814e-06,
+      "loss": 0.0969,
+      "step": 24923
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7145452958427543e-06,
+      "loss": 0.0728,
+      "step": 24924
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.713862596327933e-06,
+      "loss": 0.0862,
+      "step": 24925
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.713179969194126e-06,
+      "loss": 0.1343,
+      "step": 24926
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.712497414448122e-06,
+      "loss": 0.1216,
+      "step": 24927
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.711814932096697e-06,
+      "loss": 0.0953,
+      "step": 24928
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.711132522146628e-06,
+      "loss": 0.11,
+      "step": 24929
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.710450184604697e-06,
+      "loss": 0.0758,
+      "step": 24930
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.709767919477686e-06,
+      "loss": 0.083,
+      "step": 24931
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7090857267723692e-06,
+      "loss": 0.0932,
+      "step": 24932
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.708403606495521e-06,
+      "loss": 0.0951,
+      "step": 24933
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.707721558653923e-06,
+      "loss": 0.0741,
+      "step": 24934
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.707039583254346e-06,
+      "loss": 0.104,
+      "step": 24935
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7063576803035663e-06,
+      "loss": 0.0868,
+      "step": 24936
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7056758498083648e-06,
+      "loss": 0.1115,
+      "step": 24937
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7049940917755012e-06,
+      "loss": 0.143,
+      "step": 24938
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.704312406211758e-06,
+      "loss": 0.103,
+      "step": 24939
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7036307931239057e-06,
+      "loss": 0.1055,
+      "step": 24940
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7029492525187153e-06,
+      "loss": 0.117,
+      "step": 24941
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.702267784402953e-06,
+      "loss": 0.0874,
+      "step": 24942
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.701586388783395e-06,
+      "loss": 0.1145,
+      "step": 24943
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7009050656668046e-06,
+      "loss": 0.1483,
+      "step": 24944
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7002238150599527e-06,
+      "loss": 0.1185,
+      "step": 24945
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6995426369696133e-06,
+      "loss": 0.1028,
+      "step": 24946
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.69886153140254e-06,
+      "loss": 0.1267,
+      "step": 24947
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.698180498365507e-06,
+      "loss": 0.1093,
+      "step": 24948
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6974995378652813e-06,
+      "loss": 0.1064,
+      "step": 24949
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6968186499086224e-06,
+      "loss": 0.1027,
+      "step": 24950
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.696137834502299e-06,
+      "loss": 0.1357,
+      "step": 24951
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.695457091653073e-06,
+      "loss": 0.0958,
+      "step": 24952
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.694776421367703e-06,
+      "loss": 0.0854,
+      "step": 24953
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6940958236529548e-06,
+      "loss": 0.0844,
+      "step": 24954
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6934152985155904e-06,
+      "loss": 0.129,
+      "step": 24955
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6927348459623693e-06,
+      "loss": 0.0827,
+      "step": 24956
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.692054466000047e-06,
+      "loss": 0.0741,
+      "step": 24957
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6913741586353893e-06,
+      "loss": 0.1522,
+      "step": 24958
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6906939238751472e-06,
+      "loss": 0.0814,
+      "step": 24959
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6900137617260825e-06,
+      "loss": 0.1111,
+      "step": 24960
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.689333672194957e-06,
+      "loss": 0.0895,
+      "step": 24961
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.688653655288516e-06,
+      "loss": 0.0923,
+      "step": 24962
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6879737110135195e-06,
+      "loss": 0.0855,
+      "step": 24963
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.687293839376726e-06,
+      "loss": 0.0965,
+      "step": 24964
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.686614040384883e-06,
+      "loss": 0.0945,
+      "step": 24965
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6859343140447493e-06,
+      "loss": 0.1066,
+      "step": 24966
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6852546603630745e-06,
+      "loss": 0.0876,
+      "step": 24967
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6845750793466076e-06,
+      "loss": 0.1452,
+      "step": 24968
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.683895571002103e-06,
+      "loss": 0.1145,
+      "step": 24969
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.683216135336317e-06,
+      "loss": 0.136,
+      "step": 24970
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.682536772355986e-06,
+      "loss": 0.1123,
+      "step": 24971
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6818574820678654e-06,
+      "loss": 0.0874,
+      "step": 24972
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6811782644787076e-06,
+      "loss": 0.1042,
+      "step": 24973
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6804991195952523e-06,
+      "loss": 0.0952,
+      "step": 24974
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.679820047424253e-06,
+      "loss": 0.0978,
+      "step": 24975
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.679141047972452e-06,
+      "loss": 0.1307,
+      "step": 24976
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.678462121246592e-06,
+      "loss": 0.1073,
+      "step": 24977
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.677783267253421e-06,
+      "loss": 0.1051,
+      "step": 24978
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6771044859996855e-06,
+      "loss": 0.0866,
+      "step": 24979
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.676425777492124e-06,
+      "loss": 0.0962,
+      "step": 24980
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.675747141737478e-06,
+      "loss": 0.1323,
+      "step": 24981
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6750685787424936e-06,
+      "loss": 0.0879,
+      "step": 24982
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6743900885139074e-06,
+      "loss": 0.0822,
+      "step": 24983
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6737116710584643e-06,
+      "loss": 0.1547,
+      "step": 24984
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6730333263829e-06,
+      "loss": 0.0906,
+      "step": 24985
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6723550544939526e-06,
+      "loss": 0.1027,
+      "step": 24986
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6716768553983606e-06,
+      "loss": 0.1269,
+      "step": 24987
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.670998729102866e-06,
+      "loss": 0.0784,
+      "step": 24988
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6703206756141986e-06,
+      "loss": 0.1248,
+      "step": 24989
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6696426949391e-06,
+      "loss": 0.1076,
+      "step": 24990
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6689647870843027e-06,
+      "loss": 0.112,
+      "step": 24991
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.668286952056539e-06,
+      "loss": 0.1059,
+      "step": 24992
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6676091898625466e-06,
+      "loss": 0.1158,
+      "step": 24993
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6669315005090547e-06,
+      "loss": 0.0912,
+      "step": 24994
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6662538840027984e-06,
+      "loss": 0.0772,
+      "step": 24995
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6655763403505066e-06,
+      "loss": 0.0998,
+      "step": 24996
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.664898869558914e-06,
+      "loss": 0.1305,
+      "step": 24997
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.664221471634746e-06,
+      "loss": 0.1308,
+      "step": 24998
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6635441465847355e-06,
+      "loss": 0.0715,
+      "step": 24999
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.662866894415611e-06,
+      "loss": 0.1253,
+      "step": 25000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6621897151340957e-06,
+      "loss": 0.1141,
+      "step": 25001
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6615126087469232e-06,
+      "loss": 0.0876,
+      "step": 25002
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.660835575260814e-06,
+      "loss": 0.0865,
+      "step": 25003
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.660158614682499e-06,
+      "loss": 0.0827,
+      "step": 25004
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6594817270187012e-06,
+      "loss": 0.1011,
+      "step": 25005
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6588049122761415e-06,
+      "loss": 0.1347,
+      "step": 25006
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.658128170461547e-06,
+      "loss": 0.1128,
+      "step": 25007
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.657451501581643e-06,
+      "loss": 0.0934,
+      "step": 25008
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.656774905643147e-06,
+      "loss": 0.1239,
+      "step": 25009
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6560983826527796e-06,
+      "loss": 0.1295,
+      "step": 25010
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.655421932617267e-06,
+      "loss": 0.1121,
+      "step": 25011
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.654745555543322e-06,
+      "loss": 0.0865,
+      "step": 25012
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6540692514376688e-06,
+      "loss": 0.1036,
+      "step": 25013
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.653393020307029e-06,
+      "loss": 0.1047,
+      "step": 25014
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6527168621581114e-06,
+      "loss": 0.1169,
+      "step": 25015
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6520407769976365e-06,
+      "loss": 0.1006,
+      "step": 25016
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.651364764832326e-06,
+      "loss": 0.1012,
+      "step": 25017
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6506888256688866e-06,
+      "loss": 0.1178,
+      "step": 25018
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.650012959514041e-06,
+      "loss": 0.0704,
+      "step": 25019
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6493371663745005e-06,
+      "loss": 0.0732,
+      "step": 25020
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6486614462569735e-06,
+      "loss": 0.1249,
+      "step": 25021
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.647985799168179e-06,
+      "loss": 0.1408,
+      "step": 25022
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6473102251148286e-06,
+      "loss": 0.113,
+      "step": 25023
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6466347241036315e-06,
+      "loss": 0.0802,
+      "step": 25024
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6459592961412973e-06,
+      "loss": 0.0934,
+      "step": 25025
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6452839412345387e-06,
+      "loss": 0.1112,
+      "step": 25026
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6446086593900598e-06,
+      "loss": 0.0808,
+      "step": 25027
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.643933450614575e-06,
+      "loss": 0.061,
+      "step": 25028
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6432583149147904e-06,
+      "loss": 0.1703,
+      "step": 25029
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.642583252297407e-06,
+      "loss": 0.1169,
+      "step": 25030
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.641908262769136e-06,
+      "loss": 0.0798,
+      "step": 25031
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6412333463366857e-06,
+      "loss": 0.0717,
+      "step": 25032
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.640558503006756e-06,
+      "loss": 0.1263,
+      "step": 25033
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6398837327860506e-06,
+      "loss": 0.0923,
+      "step": 25034
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.639209035681276e-06,
+      "loss": 0.1068,
+      "step": 25035
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6385344116991298e-06,
+      "loss": 0.1009,
+      "step": 25036
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.637859860846317e-06,
+      "loss": 0.1264,
+      "step": 25037
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6371853831295447e-06,
+      "loss": 0.0997,
+      "step": 25038
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.636510978555501e-06,
+      "loss": 0.1143,
+      "step": 25039
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6358366471308906e-06,
+      "loss": 0.1236,
+      "step": 25040
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6351623888624177e-06,
+      "loss": 0.106,
+      "step": 25041
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.634488203756772e-06,
+      "loss": 0.0818,
+      "step": 25042
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6338140918206577e-06,
+      "loss": 0.1557,
+      "step": 25043
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.633140053060769e-06,
+      "loss": 0.0749,
+      "step": 25044
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6324660874837982e-06,
+      "loss": 0.1032,
+      "step": 25045
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.631792195096444e-06,
+      "loss": 0.1124,
+      "step": 25046
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6311183759054036e-06,
+      "loss": 0.1192,
+      "step": 25047
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.630444629917368e-06,
+      "loss": 0.1055,
+      "step": 25048
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.629770957139027e-06,
+      "loss": 0.1544,
+      "step": 25049
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6290973575770793e-06,
+      "loss": 0.0731,
+      "step": 25050
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6284238312382105e-06,
+      "loss": 0.0535,
+      "step": 25051
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.627750378129117e-06,
+      "loss": 0.1046,
+      "step": 25052
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.627076998256486e-06,
+      "loss": 0.0787,
+      "step": 25053
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.626403691627003e-06,
+      "loss": 0.0958,
+      "step": 25054
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.625730458247362e-06,
+      "loss": 0.1409,
+      "step": 25055
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6250572981242517e-06,
+      "loss": 0.1066,
+      "step": 25056
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.624384211264358e-06,
+      "loss": 0.0883,
+      "step": 25057
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6237111976743634e-06,
+      "loss": 0.0918,
+      "step": 25058
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6230382573609604e-06,
+      "loss": 0.0977,
+      "step": 25059
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6223653903308266e-06,
+      "loss": 0.1265,
+      "step": 25060
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6216925965906535e-06,
+      "loss": 0.1138,
+      "step": 25061
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6210198761471217e-06,
+      "loss": 0.1051,
+      "step": 25062
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6203472290069108e-06,
+      "loss": 0.0888,
+      "step": 25063
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.619674655176706e-06,
+      "loss": 0.0974,
+      "step": 25064
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6190021546631915e-06,
+      "loss": 0.0923,
+      "step": 25065
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.618329727473041e-06,
+      "loss": 0.1011,
+      "step": 25066
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.617657373612943e-06,
+      "loss": 0.1347,
+      "step": 25067
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.616985093089572e-06,
+      "loss": 0.1096,
+      "step": 25068
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.616312885909603e-06,
+      "loss": 0.1328,
+      "step": 25069
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.615640752079721e-06,
+      "loss": 0.1376,
+      "step": 25070
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6149686916065975e-06,
+      "loss": 0.0912,
+      "step": 25071
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6142967044969127e-06,
+      "loss": 0.0864,
+      "step": 25072
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6136247907573374e-06,
+      "loss": 0.0638,
+      "step": 25073
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6129529503945518e-06,
+      "loss": 0.1106,
+      "step": 25074
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6122811834152253e-06,
+      "loss": 0.0932,
+      "step": 25075
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.611609489826036e-06,
+      "loss": 0.0836,
+      "step": 25076
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6109378696336542e-06,
+      "loss": 0.1143,
+      "step": 25077
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6102663228447487e-06,
+      "loss": 0.1133,
+      "step": 25078
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.609594849465997e-06,
+      "loss": 0.1093,
+      "step": 25079
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.608923449504063e-06,
+      "loss": 0.0987,
+      "step": 25080
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.608252122965622e-06,
+      "loss": 0.1068,
+      "step": 25081
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.607580869857339e-06,
+      "loss": 0.0913,
+      "step": 25082
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.606909690185886e-06,
+      "loss": 0.0948,
+      "step": 25083
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.606238583957926e-06,
+      "loss": 0.107,
+      "step": 25084
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6055675511801315e-06,
+      "loss": 0.0842,
+      "step": 25085
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6048965918591653e-06,
+      "loss": 0.1169,
+      "step": 25086
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6042257060016897e-06,
+      "loss": 0.1176,
+      "step": 25087
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.603554893614375e-06,
+      "loss": 0.156,
+      "step": 25088
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6028841547038797e-06,
+      "loss": 0.0852,
+      "step": 25089
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.60221348927687e-06,
+      "loss": 0.1061,
+      "step": 25090
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6015428973400114e-06,
+      "loss": 0.0886,
+      "step": 25091
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6008723788999623e-06,
+      "loss": 0.1047,
+      "step": 25092
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.600201933963381e-06,
+      "loss": 0.1067,
+      "step": 25093
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.599531562536933e-06,
+      "loss": 0.1264,
+      "step": 25094
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5988612646272735e-06,
+      "loss": 0.0884,
+      "step": 25095
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.598191040241066e-06,
+      "loss": 0.0923,
+      "step": 25096
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5975208893849646e-06,
+      "loss": 0.128,
+      "step": 25097
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5968508120656255e-06,
+      "loss": 0.1124,
+      "step": 25098
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5961808082897078e-06,
+      "loss": 0.1127,
+      "step": 25099
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5955108780638695e-06,
+      "loss": 0.0749,
+      "step": 25100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.594841021394763e-06,
+      "loss": 0.1053,
+      "step": 25101
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5941712382890404e-06,
+      "loss": 0.1083,
+      "step": 25102
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5935015287533615e-06,
+      "loss": 0.1001,
+      "step": 25103
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5928318927943717e-06,
+      "loss": 0.1081,
+      "step": 25104
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.592162330418728e-06,
+      "loss": 0.1118,
+      "step": 25105
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5914928416330875e-06,
+      "loss": 0.1036,
+      "step": 25106
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.590823426444088e-06,
+      "loss": 0.0946,
+      "step": 25107
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5901540848583862e-06,
+      "loss": 0.1315,
+      "step": 25108
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.589484816882635e-06,
+      "loss": 0.1151,
+      "step": 25109
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5888156225234784e-06,
+      "loss": 0.0904,
+      "step": 25110
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.588146501787563e-06,
+      "loss": 0.1281,
+      "step": 25111
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5874774546815396e-06,
+      "loss": 0.1338,
+      "step": 25112
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.586808481212051e-06,
+      "loss": 0.13,
+      "step": 25113
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5861395813857447e-06,
+      "loss": 0.1089,
+      "step": 25114
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5854707552092705e-06,
+      "loss": 0.1261,
+      "step": 25115
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.584802002689263e-06,
+      "loss": 0.1103,
+      "step": 25116
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5841333238323706e-06,
+      "loss": 0.1071,
+      "step": 25117
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5834647186452376e-06,
+      "loss": 0.0883,
+      "step": 25118
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.582796187134502e-06,
+      "loss": 0.0936,
+      "step": 25119
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5821277293068105e-06,
+      "loss": 0.0953,
+      "step": 25120
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5814593451687997e-06,
+      "loss": 0.1134,
+      "step": 25121
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.580791034727107e-06,
+      "loss": 0.0989,
+      "step": 25122
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.580122797988375e-06,
+      "loss": 0.1231,
+      "step": 25123
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5794546349592432e-06,
+      "loss": 0.0747,
+      "step": 25124
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.578786545646348e-06,
+      "loss": 0.1032,
+      "step": 25125
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5781185300563226e-06,
+      "loss": 0.1134,
+      "step": 25126
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.57745058819581e-06,
+      "loss": 0.1299,
+      "step": 25127
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.576782720071438e-06,
+      "loss": 0.0942,
+      "step": 25128
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5761149256898453e-06,
+      "loss": 0.0923,
+      "step": 25129
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.575447205057672e-06,
+      "loss": 0.1238,
+      "step": 25130
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5747795581815383e-06,
+      "loss": 0.0904,
+      "step": 25131
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.574111985068083e-06,
+      "loss": 0.096,
+      "step": 25132
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5734444857239405e-06,
+      "loss": 0.1099,
+      "step": 25133
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5727770601557367e-06,
+      "loss": 0.0842,
+      "step": 25134
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5721097083701085e-06,
+      "loss": 0.1458,
+      "step": 25135
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5714424303736796e-06,
+      "loss": 0.1393,
+      "step": 25136
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.570775226173079e-06,
+      "loss": 0.1075,
+      "step": 25137
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5701080957749356e-06,
+      "loss": 0.0846,
+      "step": 25138
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5694410391858837e-06,
+      "loss": 0.0893,
+      "step": 25139
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5687740564125374e-06,
+      "loss": 0.0972,
+      "step": 25140
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.568107147461528e-06,
+      "loss": 0.1353,
+      "step": 25141
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.567440312339485e-06,
+      "loss": 0.1173,
+      "step": 25142
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.566773551053027e-06,
+      "loss": 0.1064,
+      "step": 25143
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.566106863608783e-06,
+      "loss": 0.0723,
+      "step": 25144
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5654402500133714e-06,
+      "loss": 0.1378,
+      "step": 25145
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5647737102734126e-06,
+      "loss": 0.1091,
+      "step": 25146
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5641072443955315e-06,
+      "loss": 0.0948,
+      "step": 25147
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.563440852386353e-06,
+      "loss": 0.1063,
+      "step": 25148
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.562774534252491e-06,
+      "loss": 0.1211,
+      "step": 25149
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5621082900005635e-06,
+      "loss": 0.1492,
+      "step": 25150
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.561442119637195e-06,
+      "loss": 0.1092,
+      "step": 25151
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5607760231689984e-06,
+      "loss": 0.1321,
+      "step": 25152
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5601100006025938e-06,
+      "loss": 0.1124,
+      "step": 25153
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.559444051944596e-06,
+      "loss": 0.1362,
+      "step": 25154
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5587781772016196e-06,
+      "loss": 0.104,
+      "step": 25155
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5581123763802794e-06,
+      "loss": 0.0836,
+      "step": 25156
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5574466494871943e-06,
+      "loss": 0.1038,
+      "step": 25157
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5567809965289703e-06,
+      "loss": 0.0998,
+      "step": 25158
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5561154175122282e-06,
+      "loss": 0.0839,
+      "step": 25159
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.555449912443575e-06,
+      "loss": 0.1308,
+      "step": 25160
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5547844813296186e-06,
+      "loss": 0.123,
+      "step": 25161
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.554119124176977e-06,
+      "loss": 0.084,
+      "step": 25162
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.553453840992254e-06,
+      "loss": 0.0982,
+      "step": 25163
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.552788631782063e-06,
+      "loss": 0.0957,
+      "step": 25164
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5521234965530094e-06,
+      "loss": 0.1069,
+      "step": 25165
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.551458435311699e-06,
+      "loss": 0.1151,
+      "step": 25166
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.550793448064741e-06,
+      "loss": 0.1202,
+      "step": 25167
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5501285348187445e-06,
+      "loss": 0.0698,
+      "step": 25168
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5494636955803108e-06,
+      "loss": 0.108,
+      "step": 25169
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5487989303560433e-06,
+      "loss": 0.0789,
+      "step": 25170
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.548134239152551e-06,
+      "loss": 0.0773,
+      "step": 25171
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5474696219764295e-06,
+      "loss": 0.1152,
+      "step": 25172
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5468050788342892e-06,
+      "loss": 0.1374,
+      "step": 25173
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.546140609732728e-06,
+      "loss": 0.0977,
+      "step": 25174
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.545476214678344e-06,
+      "loss": 0.1064,
+      "step": 25175
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.54481189367774e-06,
+      "loss": 0.1295,
+      "step": 25176
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.544147646737518e-06,
+      "loss": 0.1287,
+      "step": 25177
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5434834738642743e-06,
+      "loss": 0.0958,
+      "step": 25178
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5428193750646033e-06,
+      "loss": 0.0981,
+      "step": 25179
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.54215535034511e-06,
+      "loss": 0.1055,
+      "step": 25180
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5414913997123826e-06,
+      "loss": 0.1306,
+      "step": 25181
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5408275231730208e-06,
+      "loss": 0.1091,
+      "step": 25182
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.540163720733626e-06,
+      "loss": 0.0977,
+      "step": 25183
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.53949999240078e-06,
+      "loss": 0.144,
+      "step": 25184
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.538836338181082e-06,
+      "loss": 0.1154,
+      "step": 25185
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5381727580811277e-06,
+      "loss": 0.1181,
+      "step": 25186
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.537509252107504e-06,
+      "loss": 0.1383,
+      "step": 25187
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.536845820266808e-06,
+      "loss": 0.1141,
+      "step": 25188
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.536182462565627e-06,
+      "loss": 0.1436,
+      "step": 25189
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.535519179010548e-06,
+      "loss": 0.0785,
+      "step": 25190
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5348559696081632e-06,
+      "loss": 0.1249,
+      "step": 25191
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5341928343650656e-06,
+      "loss": 0.1036,
+      "step": 25192
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5335297732878327e-06,
+      "loss": 0.0843,
+      "step": 25193
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.532866786383058e-06,
+      "loss": 0.1134,
+      "step": 25194
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5322038736573273e-06,
+      "loss": 0.0967,
+      "step": 25195
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.531541035117223e-06,
+      "loss": 0.0939,
+      "step": 25196
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.530878270769336e-06,
+      "loss": 0.0755,
+      "step": 25197
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5302155806202442e-06,
+      "loss": 0.1315,
+      "step": 25198
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5295529646765315e-06,
+      "loss": 0.1207,
+      "step": 25199
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.528890422944781e-06,
+      "loss": 0.1349,
+      "step": 25200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.528227955431578e-06,
+      "loss": 0.1254,
+      "step": 25201
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5275655621435e-06,
+      "loss": 0.1232,
+      "step": 25202
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5269032430871253e-06,
+      "loss": 0.1179,
+      "step": 25203
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5262409982690396e-06,
+      "loss": 0.1212,
+      "step": 25204
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.525578827695814e-06,
+      "loss": 0.1099,
+      "step": 25205
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5249167313740307e-06,
+      "loss": 0.098,
+      "step": 25206
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5242547093102733e-06,
+      "loss": 0.1328,
+      "step": 25207
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5235927615111054e-06,
+      "loss": 0.099,
+      "step": 25208
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5229308879831104e-06,
+      "loss": 0.1313,
+      "step": 25209
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.522269088732865e-06,
+      "loss": 0.0963,
+      "step": 25210
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.521607363766938e-06,
+      "loss": 0.1524,
+      "step": 25211
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.520945713091909e-06,
+      "loss": 0.1342,
+      "step": 25212
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5202841367143472e-06,
+      "loss": 0.0858,
+      "step": 25213
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5196226346408238e-06,
+      "loss": 0.1215,
+      "step": 25214
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.518961206877911e-06,
+      "loss": 0.0835,
+      "step": 25215
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5182998534321836e-06,
+      "loss": 0.0941,
+      "step": 25216
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5176385743102082e-06,
+      "loss": 0.0988,
+      "step": 25217
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5169773695185508e-06,
+      "loss": 0.1002,
+      "step": 25218
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5163162390637863e-06,
+      "loss": 0.0834,
+      "step": 25219
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5156551829524756e-06,
+      "loss": 0.0983,
+      "step": 25220
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5149942011911932e-06,
+      "loss": 0.075,
+      "step": 25221
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5143332937865008e-06,
+      "loss": 0.1046,
+      "step": 25222
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5136724607449623e-06,
+      "loss": 0.1094,
+      "step": 25223
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5130117020731436e-06,
+      "loss": 0.0993,
+      "step": 25224
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5123510177776143e-06,
+      "loss": 0.1065,
+      "step": 25225
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5116904078649317e-06,
+      "loss": 0.093,
+      "step": 25226
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.511029872341658e-06,
+      "loss": 0.0898,
+      "step": 25227
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5103694112143597e-06,
+      "loss": 0.0788,
+      "step": 25228
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.509709024489592e-06,
+      "loss": 0.0935,
+      "step": 25229
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5090487121739206e-06,
+      "loss": 0.0845,
+      "step": 25230
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.508388474273903e-06,
+      "loss": 0.0778,
+      "step": 25231
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5077283107960946e-06,
+      "loss": 0.0984,
+      "step": 25232
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5070682217470576e-06,
+      "loss": 0.0966,
+      "step": 25233
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.506408207133352e-06,
+      "loss": 0.1202,
+      "step": 25234
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5057482669615263e-06,
+      "loss": 0.1072,
+      "step": 25235
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5050884012381447e-06,
+      "loss": 0.1125,
+      "step": 25236
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.504428609969759e-06,
+      "loss": 0.1112,
+      "step": 25237
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.50376889316292e-06,
+      "loss": 0.1046,
+      "step": 25238
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5031092508241873e-06,
+      "loss": 0.1565,
+      "step": 25239
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5024496829601084e-06,
+      "loss": 0.076,
+      "step": 25240
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.501790189577242e-06,
+      "loss": 0.1125,
+      "step": 25241
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.501130770682132e-06,
+      "loss": 0.0842,
+      "step": 25242
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.500471426281337e-06,
+      "loss": 0.1068,
+      "step": 25243
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.499812156381398e-06,
+      "loss": 0.1364,
+      "step": 25244
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4991529609888743e-06,
+      "loss": 0.1156,
+      "step": 25245
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4984938401103087e-06,
+      "loss": 0.0996,
+      "step": 25246
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.497834793752246e-06,
+      "loss": 0.1126,
+      "step": 25247
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4971758219212392e-06,
+      "loss": 0.1015,
+      "step": 25248
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4965169246238306e-06,
+      "loss": 0.1312,
+      "step": 25249
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4958581018665694e-06,
+      "loss": 0.101,
+      "step": 25250
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4951993536559948e-06,
+      "loss": 0.0643,
+      "step": 25251
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.494540679998658e-06,
+      "loss": 0.0942,
+      "step": 25252
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4938820809010954e-06,
+      "loss": 0.1198,
+      "step": 25253
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.493223556369856e-06,
+      "loss": 0.1117,
+      "step": 25254
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4925651064114788e-06,
+      "loss": 0.1197,
+      "step": 25255
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.491906731032501e-06,
+      "loss": 0.1165,
+      "step": 25256
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.491248430239469e-06,
+      "loss": 0.0901,
+      "step": 25257
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4905902040389173e-06,
+      "loss": 0.1199,
+      "step": 25258
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.489932052437387e-06,
+      "loss": 0.0905,
+      "step": 25259
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4892739754414195e-06,
+      "loss": 0.1164,
+      "step": 25260
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.48861597305755e-06,
+      "loss": 0.0869,
+      "step": 25261
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4879580452923114e-06,
+      "loss": 0.0681,
+      "step": 25262
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4873001921522446e-06,
+      "loss": 0.1068,
+      "step": 25263
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4866424136438806e-06,
+      "loss": 0.1158,
+      "step": 25264
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.485984709773759e-06,
+      "loss": 0.1372,
+      "step": 25265
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.48532708054841e-06,
+      "loss": 0.0922,
+      "step": 25266
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4846695259743636e-06,
+      "loss": 0.0928,
+      "step": 25267
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.484012046058156e-06,
+      "loss": 0.1072,
+      "step": 25268
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4833546408063203e-06,
+      "loss": 0.1338,
+      "step": 25269
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.482697310225386e-06,
+      "loss": 0.1065,
+      "step": 25270
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.482040054321878e-06,
+      "loss": 0.0967,
+      "step": 25271
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4813828731023327e-06,
+      "loss": 0.095,
+      "step": 25272
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.480725766573273e-06,
+      "loss": 0.0851,
+      "step": 25273
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.480068734741229e-06,
+      "loss": 0.1048,
+      "step": 25274
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4794117776127337e-06,
+      "loss": 0.1123,
+      "step": 25275
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4787548951943007e-06,
+      "loss": 0.1155,
+      "step": 25276
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4780980874924632e-06,
+      "loss": 0.1008,
+      "step": 25277
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4774413545137476e-06,
+      "loss": 0.0859,
+      "step": 25278
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4767846962646746e-06,
+      "loss": 0.1178,
+      "step": 25279
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4761281127517665e-06,
+      "loss": 0.1295,
+      "step": 25280
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4754716039815497e-06,
+      "loss": 0.1319,
+      "step": 25281
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4748151699605416e-06,
+      "loss": 0.0999,
+      "step": 25282
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4741588106952652e-06,
+      "loss": 0.1137,
+      "step": 25283
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4735025261922474e-06,
+      "loss": 0.0915,
+      "step": 25284
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.472846316457995e-06,
+      "loss": 0.0968,
+      "step": 25285
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4721901814990334e-06,
+      "loss": 0.0934,
+      "step": 25286
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4715341213218848e-06,
+      "loss": 0.1103,
+      "step": 25287
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.470878135933058e-06,
+      "loss": 0.0943,
+      "step": 25288
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.470222225339077e-06,
+      "loss": 0.14,
+      "step": 25289
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.469566389546456e-06,
+      "loss": 0.1344,
+      "step": 25290
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.468910628561705e-06,
+      "loss": 0.1208,
+      "step": 25291
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4682549423913427e-06,
+      "loss": 0.1465,
+      "step": 25292
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4675993310418846e-06,
+      "loss": 0.1362,
+      "step": 25293
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4669437945198414e-06,
+      "loss": 0.1179,
+      "step": 25294
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.466288332831722e-06,
+      "loss": 0.0969,
+      "step": 25295
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4656329459840445e-06,
+      "loss": 0.0945,
+      "step": 25296
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.464977633983313e-06,
+      "loss": 0.1053,
+      "step": 25297
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4643223968360395e-06,
+      "loss": 0.1231,
+      "step": 25298
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.46366723454874e-06,
+      "loss": 0.1144,
+      "step": 25299
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4630121471279122e-06,
+      "loss": 0.0797,
+      "step": 25300
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.462357134580067e-06,
+      "loss": 0.1278,
+      "step": 25301
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.461702196911717e-06,
+      "loss": 0.1132,
+      "step": 25302
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4610473341293608e-06,
+      "loss": 0.1203,
+      "step": 25303
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4603925462395105e-06,
+      "loss": 0.1117,
+      "step": 25304
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4597378332486675e-06,
+      "loss": 0.1087,
+      "step": 25305
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.459083195163333e-06,
+      "loss": 0.1212,
+      "step": 25306
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.458428631990013e-06,
+      "loss": 0.1508,
+      "step": 25307
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.457774143735215e-06,
+      "loss": 0.1211,
+      "step": 25308
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.457119730405432e-06,
+      "loss": 0.0981,
+      "step": 25309
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.456465392007168e-06,
+      "loss": 0.1089,
+      "step": 25310
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4558111285469266e-06,
+      "loss": 0.106,
+      "step": 25311
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4551569400312025e-06,
+      "loss": 0.1258,
+      "step": 25312
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4545028264664994e-06,
+      "loss": 0.1178,
+      "step": 25313
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.453848787859312e-06,
+      "loss": 0.1091,
+      "step": 25314
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4531948242161365e-06,
+      "loss": 0.113,
+      "step": 25315
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4525409355434705e-06,
+      "loss": 0.1204,
+      "step": 25316
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4518871218478126e-06,
+      "loss": 0.1043,
+      "step": 25317
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.451233383135657e-06,
+      "loss": 0.0907,
+      "step": 25318
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.450579719413493e-06,
+      "loss": 0.1016,
+      "step": 25319
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.449926130687821e-06,
+      "loss": 0.1138,
+      "step": 25320
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.449272616965127e-06,
+      "loss": 0.1158,
+      "step": 25321
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4486191782519098e-06,
+      "loss": 0.0915,
+      "step": 25322
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.447965814554658e-06,
+      "loss": 0.0788,
+      "step": 25323
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.447312525879858e-06,
+      "loss": 0.113,
+      "step": 25324
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4466593122340043e-06,
+      "loss": 0.1069,
+      "step": 25325
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4460061736235876e-06,
+      "loss": 0.1132,
+      "step": 25326
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.445353110055091e-06,
+      "loss": 0.136,
+      "step": 25327
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4447001215350063e-06,
+      "loss": 0.1267,
+      "step": 25328
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4440472080698196e-06,
+      "loss": 0.115,
+      "step": 25329
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4433943696660135e-06,
+      "loss": 0.1131,
+      "step": 25330
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.442741606330079e-06,
+      "loss": 0.1457,
+      "step": 25331
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4420889180684937e-06,
+      "loss": 0.1248,
+      "step": 25332
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4414363048877486e-06,
+      "loss": 0.1382,
+      "step": 25333
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.440783766794325e-06,
+      "loss": 0.0966,
+      "step": 25334
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4401313037946993e-06,
+      "loss": 0.1062,
+      "step": 25335
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.439478915895358e-06,
+      "loss": 0.0983,
+      "step": 25336
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4388266031027852e-06,
+      "loss": 0.1523,
+      "step": 25337
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.438174365423457e-06,
+      "loss": 0.1386,
+      "step": 25338
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.437522202863851e-06,
+      "loss": 0.1149,
+      "step": 25339
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4368701154304507e-06,
+      "loss": 0.0963,
+      "step": 25340
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4362181031297284e-06,
+      "loss": 0.0924,
+      "step": 25341
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.435566165968167e-06,
+      "loss": 0.104,
+      "step": 25342
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4349143039522404e-06,
+      "loss": 0.111,
+      "step": 25343
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4342625170884205e-06,
+      "loss": 0.1287,
+      "step": 25344
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4336108053831863e-06,
+      "loss": 0.0746,
+      "step": 25345
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4329591688430133e-06,
+      "loss": 0.0808,
+      "step": 25346
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4323076074743733e-06,
+      "loss": 0.1069,
+      "step": 25347
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4316561212837365e-06,
+      "loss": 0.0994,
+      "step": 25348
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.431004710277579e-06,
+      "loss": 0.1505,
+      "step": 25349
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4303533744623666e-06,
+      "loss": 0.1156,
+      "step": 25350
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.429702113844574e-06,
+      "loss": 0.1077,
+      "step": 25351
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4290509284306752e-06,
+      "loss": 0.1244,
+      "step": 25352
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4283998182271275e-06,
+      "loss": 0.0948,
+      "step": 25353
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.427748783240406e-06,
+      "loss": 0.0981,
+      "step": 25354
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4270978234769794e-06,
+      "loss": 0.1654,
+      "step": 25355
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4264469389433097e-06,
+      "loss": 0.1009,
+      "step": 25356
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.425796129645869e-06,
+      "loss": 0.0816,
+      "step": 25357
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4251453955911196e-06,
+      "loss": 0.0854,
+      "step": 25358
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4244947367855208e-06,
+      "loss": 0.0807,
+      "step": 25359
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.423844153235543e-06,
+      "loss": 0.0559,
+      "step": 25360
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4231936449476513e-06,
+      "loss": 0.0863,
+      "step": 25361
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.422543211928299e-06,
+      "loss": 0.0911,
+      "step": 25362
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4218928541839503e-06,
+      "loss": 0.118,
+      "step": 25363
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.421242571721072e-06,
+      "loss": 0.0862,
+      "step": 25364
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4205923645461173e-06,
+      "loss": 0.123,
+      "step": 25365
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4199422326655496e-06,
+      "loss": 0.0935,
+      "step": 25366
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4192921760858267e-06,
+      "loss": 0.0901,
+      "step": 25367
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.418642194813402e-06,
+      "loss": 0.0839,
+      "step": 25368
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.417992288854736e-06,
+      "loss": 0.0927,
+      "step": 25369
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4173424582162875e-06,
+      "loss": 0.0924,
+      "step": 25370
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4166927029045097e-06,
+      "loss": 0.1212,
+      "step": 25371
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4160430229258526e-06,
+      "loss": 0.1032,
+      "step": 25372
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4153934182867777e-06,
+      "loss": 0.0953,
+      "step": 25373
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4147438889937326e-06,
+      "loss": 0.0681,
+      "step": 25374
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4140944350531714e-06,
+      "loss": 0.1136,
+      "step": 25375
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.413445056471553e-06,
+      "loss": 0.1217,
+      "step": 25376
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.412795753255315e-06,
+      "loss": 0.0915,
+      "step": 25377
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4121465254109146e-06,
+      "loss": 0.1131,
+      "step": 25378
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.411497372944804e-06,
+      "loss": 0.1007,
+      "step": 25379
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.410848295863427e-06,
+      "loss": 0.0788,
+      "step": 25380
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4101992941732356e-06,
+      "loss": 0.1327,
+      "step": 25381
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4095503678806752e-06,
+      "loss": 0.1224,
+      "step": 25382
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4089015169921893e-06,
+      "loss": 0.1466,
+      "step": 25383
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4082527415142264e-06,
+      "loss": 0.0969,
+      "step": 25384
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4076040414532352e-06,
+      "loss": 0.0886,
+      "step": 25385
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.406955416815656e-06,
+      "loss": 0.0686,
+      "step": 25386
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4063068676079293e-06,
+      "loss": 0.0955,
+      "step": 25387
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.405658393836504e-06,
+      "loss": 0.1033,
+      "step": 25388
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4050099955078164e-06,
+      "loss": 0.1043,
+      "step": 25389
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4043616726283135e-06,
+      "loss": 0.1268,
+      "step": 25390
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4037134252044325e-06,
+      "loss": 0.0998,
+      "step": 25391
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.40306525324261e-06,
+      "loss": 0.0755,
+      "step": 25392
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4024171567492905e-06,
+      "loss": 0.1001,
+      "step": 25393
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4017691357309115e-06,
+      "loss": 0.1295,
+      "step": 25394
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.401121190193909e-06,
+      "loss": 0.1398,
+      "step": 25395
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.400473320144717e-06,
+      "loss": 0.1094,
+      "step": 25396
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3998255255897785e-06,
+      "loss": 0.0822,
+      "step": 25397
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.399177806535521e-06,
+      "loss": 0.0845,
+      "step": 25398
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.398530162988385e-06,
+      "loss": 0.064,
+      "step": 25399
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3978825949548023e-06,
+      "loss": 0.1013,
+      "step": 25400
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3972351024412023e-06,
+      "loss": 0.1111,
+      "step": 25401
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3965876854540204e-06,
+      "loss": 0.1172,
+      "step": 25402
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.395940343999691e-06,
+      "loss": 0.1138,
+      "step": 25403
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3952930780846396e-06,
+      "loss": 0.1474,
+      "step": 25404
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.394645887715301e-06,
+      "loss": 0.0719,
+      "step": 25405
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.393998772898102e-06,
+      "loss": 0.1126,
+      "step": 25406
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3933517336394675e-06,
+      "loss": 0.1014,
+      "step": 25407
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.392704769945833e-06,
+      "loss": 0.1066,
+      "step": 25408
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3920578818236162e-06,
+      "loss": 0.0901,
+      "step": 25409
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3914110692792527e-06,
+      "loss": 0.1219,
+      "step": 25410
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.390764332319161e-06,
+      "loss": 0.0775,
+      "step": 25411
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.390117670949771e-06,
+      "loss": 0.0855,
+      "step": 25412
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3894710851775006e-06,
+      "loss": 0.0977,
+      "step": 25413
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.38882457500878e-06,
+      "loss": 0.0852,
+      "step": 25414
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3881781404500284e-06,
+      "loss": 0.086,
+      "step": 25415
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.387531781507665e-06,
+      "loss": 0.1139,
+      "step": 25416
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.386885498188115e-06,
+      "loss": 0.1326,
+      "step": 25417
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3862392904977938e-06,
+      "loss": 0.1354,
+      "step": 25418
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.385593158443128e-06,
+      "loss": 0.0893,
+      "step": 25419
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.384947102030528e-06,
+      "loss": 0.0746,
+      "step": 25420
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.38430112126642e-06,
+      "loss": 0.0892,
+      "step": 25421
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3836552161572135e-06,
+      "loss": 0.1032,
+      "step": 25422
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.383009386709332e-06,
+      "loss": 0.128,
+      "step": 25423
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3823636329291876e-06,
+      "loss": 0.092,
+      "step": 25424
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.381717954823193e-06,
+      "loss": 0.1006,
+      "step": 25425
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3810723523977676e-06,
+      "loss": 0.1015,
+      "step": 25426
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.380426825659319e-06,
+      "loss": 0.1341,
+      "step": 25427
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.379781374614263e-06,
+      "loss": 0.1262,
+      "step": 25428
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3791359992690156e-06,
+      "loss": 0.1452,
+      "step": 25429
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.378490699629983e-06,
+      "loss": 0.0943,
+      "step": 25430
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.377845475703574e-06,
+      "loss": 0.0881,
+      "step": 25431
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.377200327496204e-06,
+      "loss": 0.1369,
+      "step": 25432
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3765552550142766e-06,
+      "loss": 0.1542,
+      "step": 25433
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3759102582642045e-06,
+      "loss": 0.1577,
+      "step": 25434
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3752653372523937e-06,
+      "loss": 0.0897,
+      "step": 25435
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.374620491985248e-06,
+      "loss": 0.098,
+      "step": 25436
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.373975722469174e-06,
+      "loss": 0.1768,
+      "step": 25437
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3733310287105816e-06,
+      "loss": 0.1075,
+      "step": 25438
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.372686410715873e-06,
+      "loss": 0.1075,
+      "step": 25439
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.372041868491447e-06,
+      "loss": 0.0874,
+      "step": 25440
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3713974020437136e-06,
+      "loss": 0.1005,
+      "step": 25441
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.37075301137907e-06,
+      "loss": 0.0982,
+      "step": 25442
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3701086965039187e-06,
+      "loss": 0.1137,
+      "step": 25443
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3694644574246672e-06,
+      "loss": 0.0547,
+      "step": 25444
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3688202941477033e-06,
+      "loss": 0.0997,
+      "step": 25445
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.368176206679432e-06,
+      "loss": 0.0948,
+      "step": 25446
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3675321950262553e-06,
+      "loss": 0.1042,
+      "step": 25447
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3668882591945673e-06,
+      "loss": 0.0865,
+      "step": 25448
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3662443991907625e-06,
+      "loss": 0.108,
+      "step": 25449
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.365600615021242e-06,
+      "loss": 0.095,
+      "step": 25450
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.364956906692396e-06,
+      "loss": 0.1075,
+      "step": 25451
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3643132742106233e-06,
+      "loss": 0.0562,
+      "step": 25452
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.363669717582321e-06,
+      "loss": 0.1108,
+      "step": 25453
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3630262368138724e-06,
+      "loss": 0.1203,
+      "step": 25454
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.362382831911675e-06,
+      "loss": 0.0729,
+      "step": 25455
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.361739502882123e-06,
+      "loss": 0.0941,
+      "step": 25456
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3610962497316035e-06,
+      "loss": 0.0967,
+      "step": 25457
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.36045307246651e-06,
+      "loss": 0.107,
+      "step": 25458
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3598099710932296e-06,
+      "loss": 0.123,
+      "step": 25459
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.35916694561815e-06,
+      "loss": 0.0713,
+      "step": 25460
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.358523996047658e-06,
+      "loss": 0.1014,
+      "step": 25461
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.357881122388148e-06,
+      "loss": 0.0798,
+      "step": 25462
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3572383246459996e-06,
+      "loss": 0.0945,
+      "step": 25463
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.356595602827598e-06,
+      "loss": 0.0879,
+      "step": 25464
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3559529569393337e-06,
+      "loss": 0.1165,
+      "step": 25465
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3553103869875836e-06,
+      "loss": 0.1412,
+      "step": 25466
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.354667892978736e-06,
+      "loss": 0.11,
+      "step": 25467
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.354025474919176e-06,
+      "loss": 0.1195,
+      "step": 25468
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3533831328152777e-06,
+      "loss": 0.1207,
+      "step": 25469
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3527408666734252e-06,
+      "loss": 0.1191,
+      "step": 25470
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3520986765000033e-06,
+      "loss": 0.1132,
+      "step": 25471
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.351456562301384e-06,
+      "loss": 0.1192,
+      "step": 25472
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.350814524083953e-06,
+      "loss": 0.0846,
+      "step": 25473
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.350172561854086e-06,
+      "loss": 0.0961,
+      "step": 25474
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.349530675618158e-06,
+      "loss": 0.1143,
+      "step": 25475
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3488888653825457e-06,
+      "loss": 0.1493,
+      "step": 25476
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3482471311536325e-06,
+      "loss": 0.0949,
+      "step": 25477
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.347605472937783e-06,
+      "loss": 0.135,
+      "step": 25478
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.346963890741374e-06,
+      "loss": 0.1458,
+      "step": 25479
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3463223845707848e-06,
+      "loss": 0.1101,
+      "step": 25480
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.345680954432381e-06,
+      "loss": 0.106,
+      "step": 25481
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3450396003325414e-06,
+      "loss": 0.1021,
+      "step": 25482
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3443983222776333e-06,
+      "loss": 0.0899,
+      "step": 25483
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.343757120274026e-06,
+      "loss": 0.0783,
+      "step": 25484
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.34311599432809e-06,
+      "loss": 0.1048,
+      "step": 25485
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.342474944446198e-06,
+      "loss": 0.0694,
+      "step": 25486
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3418339706347148e-06,
+      "loss": 0.1249,
+      "step": 25487
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3411930729000064e-06,
+      "loss": 0.0704,
+      "step": 25488
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3405522512484447e-06,
+      "loss": 0.1234,
+      "step": 25489
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3399115056863897e-06,
+      "loss": 0.1097,
+      "step": 25490
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.339270836220212e-06,
+      "loss": 0.0637,
+      "step": 25491
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3386302428562747e-06,
+      "loss": 0.1223,
+      "step": 25492
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3379897256009366e-06,
+      "loss": 0.1306,
+      "step": 25493
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.337349284460567e-06,
+      "loss": 0.1214,
+      "step": 25494
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3367089194415226e-06,
+      "loss": 0.1426,
+      "step": 25495
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.336068630550169e-06,
+      "loss": 0.1171,
+      "step": 25496
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3354284177928675e-06,
+      "loss": 0.1174,
+      "step": 25497
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.334788281175977e-06,
+      "loss": 0.0687,
+      "step": 25498
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.334148220705852e-06,
+      "loss": 0.1101,
+      "step": 25499
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3335082363888584e-06,
+      "loss": 0.1168,
+      "step": 25500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3328683282313467e-06,
+      "loss": 0.0956,
+      "step": 25501
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.33222849623968e-06,
+      "loss": 0.1222,
+      "step": 25502
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.331588740420212e-06,
+      "loss": 0.1041,
+      "step": 25503
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.330949060779295e-06,
+      "loss": 0.1062,
+      "step": 25504
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3303094573232865e-06,
+      "loss": 0.1386,
+      "step": 25505
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3296699300585434e-06,
+      "loss": 0.1355,
+      "step": 25506
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3290304789914155e-06,
+      "loss": 0.1195,
+      "step": 25507
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3283911041282526e-06,
+      "loss": 0.1372,
+      "step": 25508
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.327751805475411e-06,
+      "loss": 0.1188,
+      "step": 25509
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3271125830392373e-06,
+      "loss": 0.1049,
+      "step": 25510
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3264734368260865e-06,
+      "loss": 0.0719,
+      "step": 25511
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3258343668423067e-06,
+      "loss": 0.1034,
+      "step": 25512
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.325195373094241e-06,
+      "loss": 0.0934,
+      "step": 25513
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3245564555882405e-06,
+      "loss": 0.1136,
+      "step": 25514
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3239176143306576e-06,
+      "loss": 0.1209,
+      "step": 25515
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3232788493278323e-06,
+      "loss": 0.1238,
+      "step": 25516
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3226401605861104e-06,
+      "loss": 0.1142,
+      "step": 25517
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3220015481118406e-06,
+      "loss": 0.0993,
+      "step": 25518
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.321363011911362e-06,
+      "loss": 0.1052,
+      "step": 25519
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.320724551991019e-06,
+      "loss": 0.0968,
+      "step": 25520
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.320086168357162e-06,
+      "loss": 0.1174,
+      "step": 25521
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3194478610161208e-06,
+      "loss": 0.1165,
+      "step": 25522
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3188096299742414e-06,
+      "loss": 0.1295,
+      "step": 25523
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.318171475237867e-06,
+      "loss": 0.1065,
+      "step": 25524
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.317533396813332e-06,
+      "loss": 0.1101,
+      "step": 25525
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3168953947069805e-06,
+      "loss": 0.1029,
+      "step": 25526
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3162574689251472e-06,
+      "loss": 0.0936,
+      "step": 25527
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3156196194741674e-06,
+      "loss": 0.0718,
+      "step": 25528
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.314981846360379e-06,
+      "loss": 0.1229,
+      "step": 25529
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3143441495901253e-06,
+      "loss": 0.0979,
+      "step": 25530
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.313706529169728e-06,
+      "loss": 0.1054,
+      "step": 25531
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3130689851055276e-06,
+      "loss": 0.0899,
+      "step": 25532
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.31243151740386e-06,
+      "loss": 0.104,
+      "step": 25533
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3117941260710543e-06,
+      "loss": 0.1027,
+      "step": 25534
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.311156811113444e-06,
+      "loss": 0.1218,
+      "step": 25535
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.310519572537362e-06,
+      "loss": 0.0928,
+      "step": 25536
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.309882410349132e-06,
+      "loss": 0.1192,
+      "step": 25537
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.309245324555088e-06,
+      "loss": 0.1235,
+      "step": 25538
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3086083151615635e-06,
+      "loss": 0.0952,
+      "step": 25539
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3079713821748805e-06,
+      "loss": 0.0933,
+      "step": 25540
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3073345256013656e-06,
+      "loss": 0.116,
+      "step": 25541
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3066977454473503e-06,
+      "loss": 0.1095,
+      "step": 25542
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.306061041719154e-06,
+      "loss": 0.0753,
+      "step": 25543
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3054244144231076e-06,
+      "loss": 0.1009,
+      "step": 25544
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3047878635655373e-06,
+      "loss": 0.1002,
+      "step": 25545
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.304151389152758e-06,
+      "loss": 0.1015,
+      "step": 25546
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.303514991191096e-06,
+      "loss": 0.1113,
+      "step": 25547
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3028786696868777e-06,
+      "loss": 0.108,
+      "step": 25548
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.302242424646419e-06,
+      "loss": 0.142,
+      "step": 25549
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3016062560760444e-06,
+      "loss": 0.1131,
+      "step": 25550
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3009701639820724e-06,
+      "loss": 0.1279,
+      "step": 25551
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3003341483708188e-06,
+      "loss": 0.0793,
+      "step": 25552
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2996982092486033e-06,
+      "loss": 0.0866,
+      "step": 25553
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2990623466217477e-06,
+      "loss": 0.0935,
+      "step": 25554
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.298426560496566e-06,
+      "loss": 0.0944,
+      "step": 25555
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.29779085087937e-06,
+      "loss": 0.1374,
+      "step": 25556
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.297155217776482e-06,
+      "loss": 0.0861,
+      "step": 25557
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.29651966119421e-06,
+      "loss": 0.0999,
+      "step": 25558
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.295884181138873e-06,
+      "loss": 0.0994,
+      "step": 25559
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.295248777616782e-06,
+      "loss": 0.1066,
+      "step": 25560
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2946134506342455e-06,
+      "loss": 0.1222,
+      "step": 25561
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2939782001975773e-06,
+      "loss": 0.0836,
+      "step": 25562
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2933430263130927e-06,
+      "loss": 0.0762,
+      "step": 25563
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2927079289870967e-06,
+      "loss": 0.1201,
+      "step": 25564
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2920729082258974e-06,
+      "loss": 0.1169,
+      "step": 25565
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2914379640358077e-06,
+      "loss": 0.1007,
+      "step": 25566
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.290803096423129e-06,
+      "loss": 0.0855,
+      "step": 25567
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2901683053941746e-06,
+      "loss": 0.1101,
+      "step": 25568
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2895335909552485e-06,
+      "loss": 0.1069,
+      "step": 25569
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2888989531126517e-06,
+      "loss": 0.1124,
+      "step": 25570
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.288264391872692e-06,
+      "loss": 0.122,
+      "step": 25571
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2876299072416764e-06,
+      "loss": 0.0843,
+      "step": 25572
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2869954992259013e-06,
+      "loss": 0.0846,
+      "step": 25573
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2863611678316755e-06,
+      "loss": 0.1133,
+      "step": 25574
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2857269130652983e-06,
+      "loss": 0.0919,
+      "step": 25575
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2850927349330666e-06,
+      "loss": 0.162,
+      "step": 25576
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2844586334412856e-06,
+      "loss": 0.0719,
+      "step": 25577
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.28382460859625e-06,
+      "loss": 0.114,
+      "step": 25578
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.283190660404263e-06,
+      "loss": 0.0685,
+      "step": 25579
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.282556788871617e-06,
+      "loss": 0.0923,
+      "step": 25580
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.281922994004615e-06,
+      "loss": 0.1108,
+      "step": 25581
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2812892758095473e-06,
+      "loss": 0.0825,
+      "step": 25582
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2806556342927145e-06,
+      "loss": 0.0989,
+      "step": 25583
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2800220694604082e-06,
+      "loss": 0.112,
+      "step": 25584
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.27938858131892e-06,
+      "loss": 0.0886,
+      "step": 25585
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2787551698745492e-06,
+      "loss": 0.1152,
+      "step": 25586
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.278121835133581e-06,
+      "loss": 0.105,
+      "step": 25587
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2774885771023137e-06,
+      "loss": 0.0715,
+      "step": 25588
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.276855395787033e-06,
+      "loss": 0.1044,
+      "step": 25589
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2762222911940336e-06,
+      "loss": 0.1727,
+      "step": 25590
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2755892633295996e-06,
+      "loss": 0.107,
+      "step": 25591
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2749563122000253e-06,
+      "loss": 0.1097,
+      "step": 25592
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2743234378115954e-06,
+      "loss": 0.1131,
+      "step": 25593
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2736906401705927e-06,
+      "loss": 0.102,
+      "step": 25594
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2730579192833124e-06,
+      "loss": 0.1304,
+      "step": 25595
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2724252751560314e-06,
+      "loss": 0.1539,
+      "step": 25596
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2717927077950386e-06,
+      "loss": 0.1271,
+      "step": 25597
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2711602172066218e-06,
+      "loss": 0.1054,
+      "step": 25598
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.270527803397058e-06,
+      "loss": 0.0864,
+      "step": 25599
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.26989546637263e-06,
+      "loss": 0.1117,
+      "step": 25600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2692632061396247e-06,
+      "loss": 0.0862,
+      "step": 25601
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2686310227043162e-06,
+      "loss": 0.117,
+      "step": 25602
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.267998916072991e-06,
+      "loss": 0.1378,
+      "step": 25603
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.267366886251925e-06,
+      "loss": 0.0943,
+      "step": 25604
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2667349332473944e-06,
+      "loss": 0.1268,
+      "step": 25605
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2661030570656804e-06,
+      "loss": 0.1146,
+      "step": 25606
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2654712577130632e-06,
+      "loss": 0.0943,
+      "step": 25607
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2648395351958152e-06,
+      "loss": 0.0794,
+      "step": 25608
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.264207889520209e-06,
+      "loss": 0.1048,
+      "step": 25609
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2635763206925264e-06,
+      "loss": 0.1471,
+      "step": 25610
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2629448287190348e-06,
+      "loss": 0.1089,
+      "step": 25611
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.262313413606011e-06,
+      "loss": 0.0859,
+      "step": 25612
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2616820753597325e-06,
+      "loss": 0.1452,
+      "step": 25613
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2610508139864597e-06,
+      "loss": 0.0939,
+      "step": 25614
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2604196294924696e-06,
+      "loss": 0.1093,
+      "step": 25615
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2597885218840353e-06,
+      "loss": 0.095,
+      "step": 25616
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.259157491167424e-06,
+      "loss": 0.0892,
+      "step": 25617
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2585265373489006e-06,
+      "loss": 0.1236,
+      "step": 25618
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.257895660434739e-06,
+      "loss": 0.1258,
+      "step": 25619
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2572648604312007e-06,
+      "loss": 0.1255,
+      "step": 25620
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2566341373445554e-06,
+      "loss": 0.1126,
+      "step": 25621
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2560034911810733e-06,
+      "loss": 0.0603,
+      "step": 25622
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.255372921947009e-06,
+      "loss": 0.093,
+      "step": 25623
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.254742429648632e-06,
+      "loss": 0.1241,
+      "step": 25624
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.254112014292209e-06,
+      "loss": 0.1104,
+      "step": 25625
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.253481675883995e-06,
+      "loss": 0.0905,
+      "step": 25626
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2528514144302603e-06,
+      "loss": 0.0978,
+      "step": 25627
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2522212299372615e-06,
+      "loss": 0.0811,
+      "step": 25628
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.251591122411255e-06,
+      "loss": 0.0825,
+      "step": 25629
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2509610918585057e-06,
+      "loss": 0.1251,
+      "step": 25630
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2503311382852735e-06,
+      "loss": 0.0803,
+      "step": 25631
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.249701261697814e-06,
+      "loss": 0.0605,
+      "step": 25632
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2490714621023824e-06,
+      "loss": 0.0864,
+      "step": 25633
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2484417395052393e-06,
+      "loss": 0.1666,
+      "step": 25634
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2478120939126357e-06,
+      "loss": 0.145,
+      "step": 25635
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.247182525330829e-06,
+      "loss": 0.0831,
+      "step": 25636
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.246553033766079e-06,
+      "loss": 0.0848,
+      "step": 25637
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2459236192246282e-06,
+      "loss": 0.1032,
+      "step": 25638
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2452942817127353e-06,
+      "loss": 0.0985,
+      "step": 25639
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2446650212366537e-06,
+      "loss": 0.0776,
+      "step": 25640
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2440358378026293e-06,
+      "loss": 0.1153,
+      "step": 25641
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2434067314169193e-06,
+      "loss": 0.0755,
+      "step": 25642
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.24277770208577e-06,
+      "loss": 0.1457,
+      "step": 25643
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.242148749815426e-06,
+      "loss": 0.1077,
+      "step": 25644
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2415198746121392e-06,
+      "loss": 0.0864,
+      "step": 25645
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2408910764821623e-06,
+      "loss": 0.0829,
+      "step": 25646
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2402623554317317e-06,
+      "loss": 0.0772,
+      "step": 25647
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2396337114670973e-06,
+      "loss": 0.1212,
+      "step": 25648
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.239005144594507e-06,
+      "loss": 0.1072,
+      "step": 25649
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2383766548202e-06,
+      "loss": 0.0931,
+      "step": 25650
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.237748242150426e-06,
+      "loss": 0.0678,
+      "step": 25651
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2371199065914227e-06,
+      "loss": 0.1233,
+      "step": 25652
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2364916481494326e-06,
+      "loss": 0.1028,
+      "step": 25653
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.235863466830699e-06,
+      "loss": 0.0914,
+      "step": 25654
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.235235362641458e-06,
+      "loss": 0.0982,
+      "step": 25655
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2346073355879563e-06,
+      "loss": 0.1301,
+      "step": 25656
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2339793856764246e-06,
+      "loss": 0.1263,
+      "step": 25657
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.233351512913109e-06,
+      "loss": 0.106,
+      "step": 25658
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.232723717304239e-06,
+      "loss": 0.1134,
+      "step": 25659
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2320959988560596e-06,
+      "loss": 0.104,
+      "step": 25660
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.231468357574801e-06,
+      "loss": 0.1088,
+      "step": 25661
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2308407934666965e-06,
+      "loss": 0.1219,
+      "step": 25662
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2302133065379862e-06,
+      "loss": 0.1349,
+      "step": 25663
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.229585896794898e-06,
+      "loss": 0.1212,
+      "step": 25664
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2289585642436675e-06,
+      "loss": 0.0997,
+      "step": 25665
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2283313088905276e-06,
+      "loss": 0.0938,
+      "step": 25666
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.22770413074171e-06,
+      "loss": 0.0827,
+      "step": 25667
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2270770298034406e-06,
+      "loss": 0.1088,
+      "step": 25668
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2264500060819537e-06,
+      "loss": 0.0948,
+      "step": 25669
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.225823059583474e-06,
+      "loss": 0.1006,
+      "step": 25670
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.225196190314235e-06,
+      "loss": 0.1237,
+      "step": 25671
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2245693982804615e-06,
+      "loss": 0.1207,
+      "step": 25672
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.223942683488377e-06,
+      "loss": 0.0835,
+      "step": 25673
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2233160459442093e-06,
+      "loss": 0.1098,
+      "step": 25674
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.222689485654187e-06,
+      "loss": 0.0711,
+      "step": 25675
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2220630026245315e-06,
+      "loss": 0.1075,
+      "step": 25676
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.221436596861465e-06,
+      "loss": 0.0967,
+      "step": 25677
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.220810268371213e-06,
+      "loss": 0.1464,
+      "step": 25678
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.220184017159993e-06,
+      "loss": 0.1144,
+      "step": 25679
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2195578432340335e-06,
+      "loss": 0.0835,
+      "step": 25680
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.218931746599551e-06,
+      "loss": 0.0975,
+      "step": 25681
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2183057272627608e-06,
+      "loss": 0.1244,
+      "step": 25682
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.217679785229887e-06,
+      "loss": 0.1085,
+      "step": 25683
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2170539205071496e-06,
+      "loss": 0.0914,
+      "step": 25684
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.216428133100763e-06,
+      "loss": 0.1237,
+      "step": 25685
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2158024230169415e-06,
+      "loss": 0.0878,
+      "step": 25686
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.215176790261906e-06,
+      "loss": 0.1059,
+      "step": 25687
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2145512348418663e-06,
+      "loss": 0.0627,
+      "step": 25688
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2139257567630402e-06,
+      "loss": 0.0806,
+      "step": 25689
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.213300356031645e-06,
+      "loss": 0.0926,
+      "step": 25690
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2126750326538837e-06,
+      "loss": 0.1117,
+      "step": 25691
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2120497866359736e-06,
+      "loss": 0.1174,
+      "step": 25692
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2114246179841294e-06,
+      "loss": 0.1066,
+      "step": 25693
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2107995267045546e-06,
+      "loss": 0.1178,
+      "step": 25694
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2101745128034657e-06,
+      "loss": 0.106,
+      "step": 25695
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.209549576287068e-06,
+      "loss": 0.1218,
+      "step": 25696
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.208924717161567e-06,
+      "loss": 0.1253,
+      "step": 25697
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2082999354331735e-06,
+      "loss": 0.1193,
+      "step": 25698
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2076752311080997e-06,
+      "loss": 0.1343,
+      "step": 25699
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2070506041925386e-06,
+      "loss": 0.093,
+      "step": 25700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.206426054692703e-06,
+      "loss": 0.0797,
+      "step": 25701
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2058015826147995e-06,
+      "loss": 0.0861,
+      "step": 25702
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.205177187965025e-06,
+      "loss": 0.0917,
+      "step": 25703
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.204552870749589e-06,
+      "loss": 0.1367,
+      "step": 25704
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.203928630974691e-06,
+      "loss": 0.101,
+      "step": 25705
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.203304468646528e-06,
+      "loss": 0.0913,
+      "step": 25706
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.202680383771304e-06,
+      "loss": 0.0948,
+      "step": 25707
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2020563763552217e-06,
+      "loss": 0.1037,
+      "step": 25708
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.201432446404477e-06,
+      "loss": 0.0924,
+      "step": 25709
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2008085939252657e-06,
+      "loss": 0.1151,
+      "step": 25710
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2001848189237905e-06,
+      "loss": 0.0919,
+      "step": 25711
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1995611214062416e-06,
+      "loss": 0.092,
+      "step": 25712
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1989375013788196e-06,
+      "loss": 0.0827,
+      "step": 25713
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.198313958847723e-06,
+      "loss": 0.1298,
+      "step": 25714
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1976904938191366e-06,
+      "loss": 0.0923,
+      "step": 25715
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.197067106299259e-06,
+      "loss": 0.1064,
+      "step": 25716
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1964437962942854e-06,
+      "loss": 0.0862,
+      "step": 25717
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1958205638104025e-06,
+      "loss": 0.1155,
+      "step": 25718
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1951974088538063e-06,
+      "loss": 0.1182,
+      "step": 25719
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1945743314306865e-06,
+      "loss": 0.1115,
+      "step": 25720
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1939513315472284e-06,
+      "loss": 0.1059,
+      "step": 25721
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.193328409209624e-06,
+      "loss": 0.1516,
+      "step": 25722
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1927055644240646e-06,
+      "loss": 0.1123,
+      "step": 25723
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.192082797196734e-06,
+      "loss": 0.1129,
+      "step": 25724
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.191460107533817e-06,
+      "loss": 0.0869,
+      "step": 25725
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1908374954415045e-06,
+      "loss": 0.0889,
+      "step": 25726
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.190214960925976e-06,
+      "loss": 0.1603,
+      "step": 25727
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.189592503993422e-06,
+      "loss": 0.1276,
+      "step": 25728
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1889701246500217e-06,
+      "loss": 0.1269,
+      "step": 25729
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1883478229019574e-06,
+      "loss": 0.1095,
+      "step": 25730
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1877255987554112e-06,
+      "loss": 0.0844,
+      "step": 25731
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.187103452216569e-06,
+      "loss": 0.1429,
+      "step": 25732
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1864813832916077e-06,
+      "loss": 0.1176,
+      "step": 25733
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1858593919867056e-06,
+      "loss": 0.1171,
+      "step": 25734
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1852374783080443e-06,
+      "loss": 0.1135,
+      "step": 25735
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1846156422617993e-06,
+      "loss": 0.1027,
+      "step": 25736
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.183993883854152e-06,
+      "loss": 0.0931,
+      "step": 25737
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1833722030912765e-06,
+      "loss": 0.0805,
+      "step": 25738
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.182750599979345e-06,
+      "loss": 0.0837,
+      "step": 25739
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.182129074524537e-06,
+      "loss": 0.0902,
+      "step": 25740
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1815076267330292e-06,
+      "loss": 0.1079,
+      "step": 25741
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.180886256610988e-06,
+      "loss": 0.098,
+      "step": 25742
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1802649641645934e-06,
+      "loss": 0.1438,
+      "step": 25743
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1796437494000143e-06,
+      "loss": 0.0876,
+      "step": 25744
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.179022612323418e-06,
+      "loss": 0.1189,
+      "step": 25745
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.178401552940982e-06,
+      "loss": 0.1221,
+      "step": 25746
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1777805712588694e-06,
+      "loss": 0.0941,
+      "step": 25747
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.177159667283255e-06,
+      "loss": 0.1042,
+      "step": 25748
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1765388410203004e-06,
+      "loss": 0.091,
+      "step": 25749
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.17591809247618e-06,
+      "loss": 0.1435,
+      "step": 25750
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1752974216570545e-06,
+      "loss": 0.095,
+      "step": 25751
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1746768285690942e-06,
+      "loss": 0.0788,
+      "step": 25752
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1740563132184624e-06,
+      "loss": 0.0835,
+      "step": 25753
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1734358756113206e-06,
+      "loss": 0.1067,
+      "step": 25754
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1728155157538367e-06,
+      "loss": 0.1026,
+      "step": 25755
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.172195233652168e-06,
+      "loss": 0.1199,
+      "step": 25756
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1715750293124836e-06,
+      "loss": 0.0956,
+      "step": 25757
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1709549027409384e-06,
+      "loss": 0.122,
+      "step": 25758
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1703348539436962e-06,
+      "loss": 0.102,
+      "step": 25759
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1697148829269142e-06,
+      "loss": 0.0876,
+      "step": 25760
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.169094989696755e-06,
+      "loss": 0.1046,
+      "step": 25761
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1684751742593746e-06,
+      "loss": 0.1202,
+      "step": 25762
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.167855436620927e-06,
+      "loss": 0.1271,
+      "step": 25763
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.167235776787574e-06,
+      "loss": 0.0795,
+      "step": 25764
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1666161947654673e-06,
+      "loss": 0.1634,
+      "step": 25765
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.165996690560763e-06,
+      "loss": 0.1379,
+      "step": 25766
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.165377264179619e-06,
+      "loss": 0.1142,
+      "step": 25767
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.164757915628185e-06,
+      "loss": 0.0796,
+      "step": 25768
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.164138644912612e-06,
+      "loss": 0.1112,
+      "step": 25769
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.163519452039058e-06,
+      "loss": 0.0939,
+      "step": 25770
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.162900337013666e-06,
+      "loss": 0.1205,
+      "step": 25771
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1622812998425946e-06,
+      "loss": 0.0961,
+      "step": 25772
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.161662340531988e-06,
+      "loss": 0.097,
+      "step": 25773
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.161043459087995e-06,
+      "loss": 0.1244,
+      "step": 25774
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.160424655516764e-06,
+      "loss": 0.1656,
+      "step": 25775
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1598059298244467e-06,
+      "loss": 0.1209,
+      "step": 25776
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1591872820171857e-06,
+      "loss": 0.0991,
+      "step": 25777
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1585687121011247e-06,
+      "loss": 0.1356,
+      "step": 25778
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.157950220082413e-06,
+      "loss": 0.0899,
+      "step": 25779
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.15733180596719e-06,
+      "loss": 0.0807,
+      "step": 25780
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.156713469761602e-06,
+      "loss": 0.1328,
+      "step": 25781
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1560952114717983e-06,
+      "loss": 0.1433,
+      "step": 25782
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1554770311039065e-06,
+      "loss": 0.102,
+      "step": 25783
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1548589286640764e-06,
+      "loss": 0.1396,
+      "step": 25784
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1542409041584487e-06,
+      "loss": 0.1638,
+      "step": 25785
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1536229575931613e-06,
+      "loss": 0.1052,
+      "step": 25786
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1530050889743503e-06,
+      "loss": 0.1217,
+      "step": 25787
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1523872983081584e-06,
+      "loss": 0.1178,
+      "step": 25788
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.151769585600717e-06,
+      "loss": 0.132,
+      "step": 25789
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.151151950858167e-06,
+      "loss": 0.1168,
+      "step": 25790
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1505343940866485e-06,
+      "loss": 0.1002,
+      "step": 25791
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.149916915292285e-06,
+      "loss": 0.0903,
+      "step": 25792
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1492995144812156e-06,
+      "loss": 0.1123,
+      "step": 25793
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.148682191659578e-06,
+      "loss": 0.1155,
+      "step": 25794
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.148064946833498e-06,
+      "loss": 0.1253,
+      "step": 25795
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.147447780009114e-06,
+      "loss": 0.0903,
+      "step": 25796
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.146830691192553e-06,
+      "loss": 0.1225,
+      "step": 25797
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1462136803899426e-06,
+      "loss": 0.1067,
+      "step": 25798
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.145596747607416e-06,
+      "loss": 0.1132,
+      "step": 25799
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1449798928511035e-06,
+      "loss": 0.0886,
+      "step": 25800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1443631161271305e-06,
+      "loss": 0.0967,
+      "step": 25801
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.143746417441621e-06,
+      "loss": 0.1021,
+      "step": 25802
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1431297968007082e-06,
+      "loss": 0.1392,
+      "step": 25803
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.142513254210511e-06,
+      "loss": 0.1031,
+      "step": 25804
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.141896789677157e-06,
+      "loss": 0.0678,
+      "step": 25805
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.141280403206776e-06,
+      "loss": 0.1016,
+      "step": 25806
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.14066409480548e-06,
+      "loss": 0.1063,
+      "step": 25807
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1400478644793976e-06,
+      "loss": 0.0939,
+      "step": 25808
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.139431712234652e-06,
+      "loss": 0.0783,
+      "step": 25809
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1388156380773595e-06,
+      "loss": 0.0724,
+      "step": 25810
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.138199642013645e-06,
+      "loss": 0.0916,
+      "step": 25811
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1375837240496267e-06,
+      "loss": 0.0805,
+      "step": 25812
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.136967884191419e-06,
+      "loss": 0.0978,
+      "step": 25813
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1363521224451456e-06,
+      "loss": 0.1171,
+      "step": 25814
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.135736438816921e-06,
+      "loss": 0.116,
+      "step": 25815
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.135120833312858e-06,
+      "loss": 0.1269,
+      "step": 25816
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.134505305939075e-06,
+      "loss": 0.1016,
+      "step": 25817
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1338898567016898e-06,
+      "loss": 0.098,
+      "step": 25818
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.133274485606811e-06,
+      "loss": 0.0799,
+      "step": 25819
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1326591926605567e-06,
+      "loss": 0.0873,
+      "step": 25820
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.132043977869037e-06,
+      "loss": 0.094,
+      "step": 25821
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.13142884123836e-06,
+      "loss": 0.0873,
+      "step": 25822
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.130813782774643e-06,
+      "loss": 0.0821,
+      "step": 25823
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1301988024839904e-06,
+      "loss": 0.0777,
+      "step": 25824
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.129583900372516e-06,
+      "loss": 0.1496,
+      "step": 25825
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.128969076446323e-06,
+      "loss": 0.1003,
+      "step": 25826
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1283543307115263e-06,
+      "loss": 0.1202,
+      "step": 25827
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1277396631742253e-06,
+      "loss": 0.1232,
+      "step": 25828
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1271250738405313e-06,
+      "loss": 0.0848,
+      "step": 25829
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1265105627165495e-06,
+      "loss": 0.1235,
+      "step": 25830
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.125896129808379e-06,
+      "loss": 0.1233,
+      "step": 25831
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1252817751221323e-06,
+      "loss": 0.129,
+      "step": 25832
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1246674986639025e-06,
+      "loss": 0.081,
+      "step": 25833
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.124053300439799e-06,
+      "loss": 0.1292,
+      "step": 25834
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.123439180455924e-06,
+      "loss": 0.1133,
+      "step": 25835
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.122825138718375e-06,
+      "loss": 0.1186,
+      "step": 25836
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1222111752332498e-06,
+      "loss": 0.1076,
+      "step": 25837
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.121597290006653e-06,
+      "loss": 0.1031,
+      "step": 25838
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1209834830446785e-06,
+      "loss": 0.1267,
+      "step": 25839
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1203697543534273e-06,
+      "loss": 0.1039,
+      "step": 25840
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.119756103938996e-06,
+      "loss": 0.1022,
+      "step": 25841
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1191425318074752e-06,
+      "loss": 0.1246,
+      "step": 25842
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.118529037964965e-06,
+      "loss": 0.1131,
+      "step": 25843
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.117915622417561e-06,
+      "loss": 0.1075,
+      "step": 25844
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1173022851713554e-06,
+      "loss": 0.1049,
+      "step": 25845
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.116689026232438e-06,
+      "loss": 0.1063,
+      "step": 25846
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1160758456069075e-06,
+      "loss": 0.1221,
+      "step": 25847
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.115462743300848e-06,
+      "loss": 0.1034,
+      "step": 25848
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1148497193203564e-06,
+      "loss": 0.095,
+      "step": 25849
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1142367736715196e-06,
+      "loss": 0.0901,
+      "step": 25850
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.113623906360425e-06,
+      "loss": 0.1068,
+      "step": 25851
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1130111173931613e-06,
+      "loss": 0.0885,
+      "step": 25852
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1123984067758206e-06,
+      "loss": 0.1025,
+      "step": 25853
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1117857745144867e-06,
+      "loss": 0.1284,
+      "step": 25854
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1111732206152424e-06,
+      "loss": 0.1055,
+      "step": 25855
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.110560745084178e-06,
+      "loss": 0.1012,
+      "step": 25856
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1099483479273732e-06,
+      "loss": 0.11,
+      "step": 25857
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1093360291509133e-06,
+      "loss": 0.1061,
+      "step": 25858
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.108723788760888e-06,
+      "loss": 0.1295,
+      "step": 25859
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.108111626763367e-06,
+      "loss": 0.0951,
+      "step": 25860
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1074995431644386e-06,
+      "loss": 0.0987,
+      "step": 25861
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1068875379701836e-06,
+      "loss": 0.1072,
+      "step": 25862
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.106275611186679e-06,
+      "loss": 0.1044,
+      "step": 25863
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.105663762820007e-06,
+      "loss": 0.1127,
+      "step": 25864
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.105051992876245e-06,
+      "loss": 0.113,
+      "step": 25865
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1044403013614655e-06,
+      "loss": 0.0865,
+      "step": 25866
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1038286882817494e-06,
+      "loss": 0.1191,
+      "step": 25867
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.103217153643178e-06,
+      "loss": 0.1175,
+      "step": 25868
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.102605697451814e-06,
+      "loss": 0.0867,
+      "step": 25869
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.101994319713738e-06,
+      "loss": 0.0976,
+      "step": 25870
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.101383020435026e-06,
+      "loss": 0.0957,
+      "step": 25871
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1007717996217456e-06,
+      "loss": 0.0878,
+      "step": 25872
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.100160657279975e-06,
+      "loss": 0.1329,
+      "step": 25873
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0995495934157807e-06,
+      "loss": 0.0914,
+      "step": 25874
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0989386080352304e-06,
+      "loss": 0.117,
+      "step": 25875
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0983277011443983e-06,
+      "loss": 0.0945,
+      "step": 25876
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.097716872749355e-06,
+      "loss": 0.1273,
+      "step": 25877
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.097106122856164e-06,
+      "loss": 0.1084,
+      "step": 25878
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0964954514708926e-06,
+      "loss": 0.0882,
+      "step": 25879
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0958848585996117e-06,
+      "loss": 0.1052,
+      "step": 25880
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0952743442483816e-06,
+      "loss": 0.0796,
+      "step": 25881
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.094663908423268e-06,
+      "loss": 0.1169,
+      "step": 25882
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0940535511303437e-06,
+      "loss": 0.1213,
+      "step": 25883
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.093443272375659e-06,
+      "loss": 0.0771,
+      "step": 25884
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.092833072165282e-06,
+      "loss": 0.0722,
+      "step": 25885
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.092222950505278e-06,
+      "loss": 0.0781,
+      "step": 25886
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0916129074017023e-06,
+      "loss": 0.1068,
+      "step": 25887
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0910029428606205e-06,
+      "loss": 0.1206,
+      "step": 25888
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.090393056888089e-06,
+      "loss": 0.0881,
+      "step": 25889
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.089783249490165e-06,
+      "loss": 0.0911,
+      "step": 25890
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.089173520672907e-06,
+      "loss": 0.1095,
+      "step": 25891
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.088563870442376e-06,
+      "loss": 0.1181,
+      "step": 25892
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0879542988046264e-06,
+      "loss": 0.107,
+      "step": 25893
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0873448057657086e-06,
+      "loss": 0.0842,
+      "step": 25894
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.086735391331686e-06,
+      "loss": 0.0862,
+      "step": 25895
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.086126055508604e-06,
+      "loss": 0.1158,
+      "step": 25896
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.085516798302523e-06,
+      "loss": 0.1331,
+      "step": 25897
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.084907619719493e-06,
+      "loss": 0.1339,
+      "step": 25898
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0842985197655606e-06,
+      "loss": 0.0765,
+      "step": 25899
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0836894984467826e-06,
+      "loss": 0.0858,
+      "step": 25900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0830805557692104e-06,
+      "loss": 0.1008,
+      "step": 25901
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0824716917388887e-06,
+      "loss": 0.0893,
+      "step": 25902
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0818629063618654e-06,
+      "loss": 0.132,
+      "step": 25903
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.081254199644194e-06,
+      "loss": 0.1213,
+      "step": 25904
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.080645571591915e-06,
+      "loss": 0.1127,
+      "step": 25905
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.08003702221108e-06,
+      "loss": 0.0778,
+      "step": 25906
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0794285515077317e-06,
+      "loss": 0.1054,
+      "step": 25907
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.078820159487912e-06,
+      "loss": 0.1148,
+      "step": 25908
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0782118461576674e-06,
+      "loss": 0.1214,
+      "step": 25909
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0776036115230436e-06,
+      "loss": 0.1003,
+      "step": 25910
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0769954555900786e-06,
+      "loss": 0.1007,
+      "step": 25911
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.076387378364817e-06,
+      "loss": 0.1132,
+      "step": 25912
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.075779379853299e-06,
+      "loss": 0.1352,
+      "step": 25913
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.07517146006156e-06,
+      "loss": 0.0923,
+      "step": 25914
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0745636189956465e-06,
+      "loss": 0.0878,
+      "step": 25915
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.073955856661589e-06,
+      "loss": 0.1184,
+      "step": 25916
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0733481730654327e-06,
+      "loss": 0.1159,
+      "step": 25917
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0727405682132086e-06,
+      "loss": 0.0944,
+      "step": 25918
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0721330421109565e-06,
+      "loss": 0.1078,
+      "step": 25919
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.071525594764707e-06,
+      "loss": 0.0993,
+      "step": 25920
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0709182261805004e-06,
+      "loss": 0.0929,
+      "step": 25921
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.070310936364368e-06,
+      "loss": 0.0946,
+      "step": 25922
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0697037253223385e-06,
+      "loss": 0.1515,
+      "step": 25923
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.069096593060451e-06,
+      "loss": 0.1073,
+      "step": 25924
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0684895395847306e-06,
+      "loss": 0.1279,
+      "step": 25925
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0678825649012126e-06,
+      "loss": 0.1004,
+      "step": 25926
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0672756690159223e-06,
+      "loss": 0.0781,
+      "step": 25927
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0666688519348934e-06,
+      "loss": 0.1162,
+      "step": 25928
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.066062113664149e-06,
+      "loss": 0.1052,
+      "step": 25929
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0654554542097228e-06,
+      "loss": 0.0999,
+      "step": 25930
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0648488735776374e-06,
+      "loss": 0.089,
+      "step": 25931
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.064242371773916e-06,
+      "loss": 0.1464,
+      "step": 25932
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0636359488045885e-06,
+      "loss": 0.144,
+      "step": 25933
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.063029604675676e-06,
+      "loss": 0.1135,
+      "step": 25934
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0624233393932024e-06,
+      "loss": 0.0793,
+      "step": 25935
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0618171529631936e-06,
+      "loss": 0.0803,
+      "step": 25936
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.06121104539167e-06,
+      "loss": 0.1183,
+      "step": 25937
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0606050166846493e-06,
+      "loss": 0.1043,
+      "step": 25938
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.059999066848156e-06,
+      "loss": 0.0916,
+      "step": 25939
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0593931958882065e-06,
+      "loss": 0.1039,
+      "step": 25940
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0587874038108234e-06,
+      "loss": 0.1268,
+      "step": 25941
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.058181690622022e-06,
+      "loss": 0.1576,
+      "step": 25942
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.057576056327817e-06,
+      "loss": 0.1042,
+      "step": 25943
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.056970500934228e-06,
+      "loss": 0.1129,
+      "step": 25944
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0563650244472733e-06,
+      "loss": 0.1141,
+      "step": 25945
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.055759626872964e-06,
+      "loss": 0.0811,
+      "step": 25946
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.055154308217313e-06,
+      "loss": 0.1077,
+      "step": 25947
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.054549068486338e-06,
+      "loss": 0.1099,
+      "step": 25948
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0539439076860455e-06,
+      "loss": 0.0988,
+      "step": 25949
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0533388258224516e-06,
+      "loss": 0.1132,
+      "step": 25950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0527338229015725e-06,
+      "loss": 0.1043,
+      "step": 25951
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0521288989294054e-06,
+      "loss": 0.074,
+      "step": 25952
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.051524053911966e-06,
+      "loss": 0.1339,
+      "step": 25953
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0509192878552677e-06,
+      "loss": 0.115,
+      "step": 25954
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.050314600765313e-06,
+      "loss": 0.113,
+      "step": 25955
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0497099926481067e-06,
+      "loss": 0.119,
+      "step": 25956
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.049105463509662e-06,
+      "loss": 0.0739,
+      "step": 25957
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0485010133559767e-06,
+      "loss": 0.1076,
+      "step": 25958
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.04789664219306e-06,
+      "loss": 0.0795,
+      "step": 25959
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0472923500269203e-06,
+      "loss": 0.0909,
+      "step": 25960
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0466881368635492e-06,
+      "loss": 0.102,
+      "step": 25961
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0460840027089555e-06,
+      "loss": 0.1262,
+      "step": 25962
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0454799475691434e-06,
+      "loss": 0.1294,
+      "step": 25963
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0448759714501077e-06,
+      "loss": 0.1081,
+      "step": 25964
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.044272074357854e-06,
+      "loss": 0.1024,
+      "step": 25965
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.043668256298379e-06,
+      "loss": 0.137,
+      "step": 25966
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0430645172776776e-06,
+      "loss": 0.12,
+      "step": 25967
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.04246085730175e-06,
+      "loss": 0.1053,
+      "step": 25968
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0418572763765975e-06,
+      "loss": 0.0949,
+      "step": 25969
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.041253774508212e-06,
+      "loss": 0.0894,
+      "step": 25970
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0406503517025857e-06,
+      "loss": 0.1176,
+      "step": 25971
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0400470079657187e-06,
+      "loss": 0.1251,
+      "step": 25972
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0394437433035998e-06,
+      "loss": 0.1117,
+      "step": 25973
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.038840557722227e-06,
+      "loss": 0.1079,
+      "step": 25974
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.038237451227589e-06,
+      "loss": 0.092,
+      "step": 25975
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.037634423825676e-06,
+      "loss": 0.0957,
+      "step": 25976
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0370314755224805e-06,
+      "loss": 0.0575,
+      "step": 25977
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0364286063239947e-06,
+      "loss": 0.0846,
+      "step": 25978
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.035825816236202e-06,
+      "loss": 0.0785,
+      "step": 25979
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.035223105265096e-06,
+      "loss": 0.1169,
+      "step": 25980
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.034620473416662e-06,
+      "loss": 0.0987,
+      "step": 25981
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0340179206968825e-06,
+      "loss": 0.1106,
+      "step": 25982
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0334154471117496e-06,
+      "loss": 0.1104,
+      "step": 25983
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.032813052667246e-06,
+      "loss": 0.1168,
+      "step": 25984
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0322107373693535e-06,
+      "loss": 0.0987,
+      "step": 25985
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.031608501224057e-06,
+      "loss": 0.1459,
+      "step": 25986
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0310063442373427e-06,
+      "loss": 0.0932,
+      "step": 25987
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.030404266415187e-06,
+      "loss": 0.0999,
+      "step": 25988
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0298022677635755e-06,
+      "loss": 0.0926,
+      "step": 25989
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.029200348288487e-06,
+      "loss": 0.0995,
+      "step": 25990
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.028598507995897e-06,
+      "loss": 0.0867,
+      "step": 25991
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0279967468917916e-06,
+      "loss": 0.1041,
+      "step": 25992
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0273950649821416e-06,
+      "loss": 0.113,
+      "step": 25993
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.02679346227293e-06,
+      "loss": 0.0914,
+      "step": 25994
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.026191938770128e-06,
+      "loss": 0.0958,
+      "step": 25995
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.025590494479717e-06,
+      "loss": 0.0986,
+      "step": 25996
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.024989129407665e-06,
+      "loss": 0.0493,
+      "step": 25997
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.024387843559953e-06,
+      "loss": 0.0317,
+      "step": 25998
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0237866369425505e-06,
+      "loss": 0.0737,
+      "step": 25999
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0231855095614285e-06,
+      "loss": 0.0532,
+      "step": 26000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0225844614225608e-06,
+      "loss": 0.0354,
+      "step": 26001
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.021983492531917e-06,
+      "loss": 0.0391,
+      "step": 26002
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0213826028954676e-06,
+      "loss": 0.0683,
+      "step": 26003
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0207817925191853e-06,
+      "loss": 0.0511,
+      "step": 26004
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0201810614090343e-06,
+      "loss": 0.0591,
+      "step": 26005
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0195804095709825e-06,
+      "loss": 0.0637,
+      "step": 26006
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0189798370109992e-06,
+      "loss": 0.0572,
+      "step": 26007
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0183793437350473e-06,
+      "loss": 0.0484,
+      "step": 26008
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.017778929749097e-06,
+      "loss": 0.0509,
+      "step": 26009
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0171785950591092e-06,
+      "loss": 0.0654,
+      "step": 26010
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.016578339671046e-06,
+      "loss": 0.0465,
+      "step": 26011
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0159781635908728e-06,
+      "loss": 0.0542,
+      "step": 26012
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0153780668245536e-06,
+      "loss": 0.045,
+      "step": 26013
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.014778049378049e-06,
+      "loss": 0.061,
+      "step": 26014
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0141781112573155e-06,
+      "loss": 0.0496,
+      "step": 26015
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.013578252468319e-06,
+      "loss": 0.055,
+      "step": 26016
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0129784730170133e-06,
+      "loss": 0.0374,
+      "step": 26017
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0123787729093615e-06,
+      "loss": 0.0577,
+      "step": 26018
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0117791521513175e-06,
+      "loss": 0.0445,
+      "step": 26019
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0111796107488367e-06,
+      "loss": 0.0539,
+      "step": 26020
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.010580148707878e-06,
+      "loss": 0.0314,
+      "step": 26021
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0099807660343973e-06,
+      "loss": 0.0302,
+      "step": 26022
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0093814627343476e-06,
+      "loss": 0.0567,
+      "step": 26023
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.00878223881368e-06,
+      "loss": 0.0419,
+      "step": 26024
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.008183094278352e-06,
+      "loss": 0.0671,
+      "step": 26025
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.00758402913431e-06,
+      "loss": 0.0692,
+      "step": 26026
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0069850433875083e-06,
+      "loss": 0.0506,
+      "step": 26027
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.006386137043902e-06,
+      "loss": 0.0533,
+      "step": 26028
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0057873101094306e-06,
+      "loss": 0.0621,
+      "step": 26029
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.005188562590048e-06,
+      "loss": 0.0439,
+      "step": 26030
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0045898944917054e-06,
+      "loss": 0.0358,
+      "step": 26031
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0039913058203433e-06,
+      "loss": 0.0401,
+      "step": 26032
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0033927965819145e-06,
+      "loss": 0.0604,
+      "step": 26033
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.002794366782362e-06,
+      "loss": 0.0239,
+      "step": 26034
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.002196016427629e-06,
+      "loss": 0.056,
+      "step": 26035
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.001597745523659e-06,
+      "loss": 0.0299,
+      "step": 26036
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.000999554076404e-06,
+      "loss": 0.0448,
+      "step": 26037
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0004014420917937e-06,
+      "loss": 0.0514,
+      "step": 26038
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.999803409575777e-06,
+      "loss": 0.042,
+      "step": 26039
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9992054565342954e-06,
+      "loss": 0.0446,
+      "step": 26040
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9986075829732845e-06,
+      "loss": 0.0485,
+      "step": 26041
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9980097888986894e-06,
+      "loss": 0.0452,
+      "step": 26042
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.997412074316445e-06,
+      "loss": 0.0391,
+      "step": 26043
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9968144392324873e-06,
+      "loss": 0.0576,
+      "step": 26044
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.996216883652756e-06,
+      "loss": 0.0418,
+      "step": 26045
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.995619407583189e-06,
+      "loss": 0.0557,
+      "step": 26046
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.99502201102972e-06,
+      "loss": 0.0442,
+      "step": 26047
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9944246939982804e-06,
+      "loss": 0.0627,
+      "step": 26048
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993827456494809e-06,
+      "loss": 0.0528,
+      "step": 26049
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993230298525234e-06,
+      "loss": 0.0431,
+      "step": 26050
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9926332200954903e-06,
+      "loss": 0.0377,
+      "step": 26051
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.992036221211514e-06,
+      "loss": 0.0377,
+      "step": 26052
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9914393018792255e-06,
+      "loss": 0.0572,
+      "step": 26053
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9908424621045596e-06,
+      "loss": 0.0404,
+      "step": 26054
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9902457018934496e-06,
+      "loss": 0.0356,
+      "step": 26055
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.989649021251817e-06,
+      "loss": 0.0443,
+      "step": 26056
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.989052420185594e-06,
+      "loss": 0.0445,
+      "step": 26057
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9884558987007063e-06,
+      "loss": 0.052,
+      "step": 26058
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.987859456803076e-06,
+      "loss": 0.0595,
+      "step": 26059
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9872630944986303e-06,
+      "loss": 0.0447,
+      "step": 26060
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9866668117932984e-06,
+      "loss": 0.0406,
+      "step": 26061
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.986070608692999e-06,
+      "loss": 0.0481,
+      "step": 26062
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9854744852036534e-06,
+      "loss": 0.0391,
+      "step": 26063
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9848784413311885e-06,
+      "loss": 0.0626,
+      "step": 26064
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9842824770815194e-06,
+      "loss": 0.0494,
+      "step": 26065
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.983686592460572e-06,
+      "loss": 0.0488,
+      "step": 26066
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9830907874742646e-06,
+      "loss": 0.0626,
+      "step": 26067
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.982495062128511e-06,
+      "loss": 0.0444,
+      "step": 26068
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.981899416429235e-06,
+      "loss": 0.0546,
+      "step": 26069
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9813038503823534e-06,
+      "loss": 0.0441,
+      "step": 26070
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.980708363993781e-06,
+      "loss": 0.0451,
+      "step": 26071
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9801129572694312e-06,
+      "loss": 0.0415,
+      "step": 26072
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.979517630215223e-06,
+      "loss": 0.0361,
+      "step": 26073
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9789223828370662e-06,
+      "loss": 0.0531,
+      "step": 26074
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9783272151408805e-06,
+      "loss": 0.0706,
+      "step": 26075
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9777321271325724e-06,
+      "loss": 0.0499,
+      "step": 26076
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9771371188180533e-06,
+      "loss": 0.0381,
+      "step": 26077
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9765421902032367e-06,
+      "loss": 0.0425,
+      "step": 26078
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9759473412940345e-06,
+      "loss": 0.0393,
+      "step": 26079
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9753525720963505e-06,
+      "loss": 0.0404,
+      "step": 26080
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9747578826160983e-06,
+      "loss": 0.0425,
+      "step": 26081
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.974163272859185e-06,
+      "loss": 0.044,
+      "step": 26082
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.973568742831512e-06,
+      "loss": 0.058,
+      "step": 26083
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.972974292538993e-06,
+      "loss": 0.0516,
+      "step": 26084
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.972379921987527e-06,
+      "loss": 0.0603,
+      "step": 26085
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9717856311830243e-06,
+      "loss": 0.0612,
+      "step": 26086
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.971191420131382e-06,
+      "loss": 0.0395,
+      "step": 26087
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9705972888385094e-06,
+      "loss": 0.0555,
+      "step": 26088
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.970003237310304e-06,
+      "loss": 0.0451,
+      "step": 26089
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.969409265552671e-06,
+      "loss": 0.0455,
+      "step": 26090
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9688153735715087e-06,
+      "loss": 0.054,
+      "step": 26091
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.968221561372715e-06,
+      "loss": 0.04,
+      "step": 26092
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.967627828962194e-06,
+      "loss": 0.0542,
+      "step": 26093
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.967034176345838e-06,
+      "loss": 0.0574,
+      "step": 26094
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9664406035295493e-06,
+      "loss": 0.0475,
+      "step": 26095
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.96584711051922e-06,
+      "loss": 0.0414,
+      "step": 26096
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9652536973207516e-06,
+      "loss": 0.0568,
+      "step": 26097
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.964660363940033e-06,
+      "loss": 0.0542,
+      "step": 26098
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9640671103829634e-06,
+      "loss": 0.0287,
+      "step": 26099
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9634739366554343e-06,
+      "loss": 0.0374,
+      "step": 26100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.962880842763336e-06,
+      "loss": 0.0403,
+      "step": 26101
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9622878287125658e-06,
+      "loss": 0.0431,
+      "step": 26102
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9616948945090076e-06,
+      "loss": 0.0439,
+      "step": 26103
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9611020401585556e-06,
+      "loss": 0.0405,
+      "step": 26104
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9605092656671055e-06,
+      "loss": 0.046,
+      "step": 26105
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9599165710405332e-06,
+      "loss": 0.0333,
+      "step": 26106
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.959323956284733e-06,
+      "loss": 0.0528,
+      "step": 26107
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9587314214055953e-06,
+      "loss": 0.0383,
+      "step": 26108
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.958138966409e-06,
+      "loss": 0.0569,
+      "step": 26109
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9575465913008386e-06,
+      "loss": 0.0449,
+      "step": 26110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9569542960869927e-06,
+      "loss": 0.0385,
+      "step": 26111
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.956362080773344e-06,
+      "loss": 0.0398,
+      "step": 26112
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9557699453657773e-06,
+      "loss": 0.0283,
+      "step": 26113
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.955177889870179e-06,
+      "loss": 0.042,
+      "step": 26114
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.954585914292426e-06,
+      "loss": 0.0384,
+      "step": 26115
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.953994018638399e-06,
+      "loss": 0.0441,
+      "step": 26116
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.953402202913981e-06,
+      "loss": 0.0356,
+      "step": 26117
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9528104671250468e-06,
+      "loss": 0.0429,
+      "step": 26118
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9522188112774764e-06,
+      "loss": 0.0935,
+      "step": 26119
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.951627235377155e-06,
+      "loss": 0.0553,
+      "step": 26120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.951035739429946e-06,
+      "loss": 0.0571,
+      "step": 26121
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9504443234417327e-06,
+      "loss": 0.0458,
+      "step": 26122
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9498529874183923e-06,
+      "loss": 0.0566,
+      "step": 26123
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9492617313657957e-06,
+      "loss": 0.0317,
+      "step": 26124
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9486705552898145e-06,
+      "loss": 0.035,
+      "step": 26125
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.948079459196327e-06,
+      "loss": 0.0456,
+      "step": 26126
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9474884430912e-06,
+      "loss": 0.0448,
+      "step": 26127
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9468975069803064e-06,
+      "loss": 0.0449,
+      "step": 26128
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9463066508695226e-06,
+      "loss": 0.0514,
+      "step": 26129
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9457158747647086e-06,
+      "loss": 0.041,
+      "step": 26130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9451251786717364e-06,
+      "loss": 0.0475,
+      "step": 26131
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.944534562596477e-06,
+      "loss": 0.0354,
+      "step": 26132
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.943944026544795e-06,
+      "loss": 0.0525,
+      "step": 26133
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9433535705225594e-06,
+      "loss": 0.0517,
+      "step": 26134
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.942763194535633e-06,
+      "loss": 0.0522,
+      "step": 26135
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.942172898589879e-06,
+      "loss": 0.0538,
+      "step": 26136
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9415826826911656e-06,
+      "loss": 0.0366,
+      "step": 26137
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9409925468453562e-06,
+      "loss": 0.0501,
+      "step": 26138
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.940402491058311e-06,
+      "loss": 0.0442,
+      "step": 26139
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9398125153358903e-06,
+      "loss": 0.0456,
+      "step": 26140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.939222619683959e-06,
+      "loss": 0.0478,
+      "step": 26141
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9386328041083723e-06,
+      "loss": 0.0568,
+      "step": 26142
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9380430686149953e-06,
+      "loss": 0.0497,
+      "step": 26143
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.937453413209682e-06,
+      "loss": 0.0347,
+      "step": 26144
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9368638378982908e-06,
+      "loss": 0.0397,
+      "step": 26145
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.936274342686678e-06,
+      "loss": 0.0391,
+      "step": 26146
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.935684927580704e-06,
+      "loss": 0.0389,
+      "step": 26147
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.935095592586218e-06,
+      "loss": 0.0411,
+      "step": 26148
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.93450633770908e-06,
+      "loss": 0.0423,
+      "step": 26149
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9339171629551423e-06,
+      "loss": 0.0474,
+      "step": 26150
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9333280683302534e-06,
+      "loss": 0.0526,
+      "step": 26151
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9327390538402713e-06,
+      "loss": 0.0497,
+      "step": 26152
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.932150119491044e-06,
+      "loss": 0.0332,
+      "step": 26153
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.931561265288421e-06,
+      "loss": 0.05,
+      "step": 26154
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.930972491238253e-06,
+      "loss": 0.0385,
+      "step": 26155
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9303837973463934e-06,
+      "loss": 0.0465,
+      "step": 26156
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9297951836186836e-06,
+      "loss": 0.0472,
+      "step": 26157
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9292066500609765e-06,
+      "loss": 0.0506,
+      "step": 26158
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.928618196679115e-06,
+      "loss": 0.0403,
+      "step": 26159
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9280298234789453e-06,
+      "loss": 0.0511,
+      "step": 26160
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.927441530466314e-06,
+      "loss": 0.081,
+      "step": 26161
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9268533176470617e-06,
+      "loss": 0.0488,
+      "step": 26162
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9262651850270376e-06,
+      "loss": 0.0444,
+      "step": 26163
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9256771326120784e-06,
+      "loss": 0.047,
+      "step": 26164
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9250891604080302e-06,
+      "loss": 0.0276,
+      "step": 26165
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9245012684207296e-06,
+      "loss": 0.0397,
+      "step": 26166
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9239134566560224e-06,
+      "loss": 0.0239,
+      "step": 26167
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9233257251197445e-06,
+      "loss": 0.0538,
+      "step": 26168
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9227380738177326e-06,
+      "loss": 0.0444,
+      "step": 26169
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9221505027558285e-06,
+      "loss": 0.0561,
+      "step": 26170
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9215630119398653e-06,
+      "loss": 0.0398,
+      "step": 26171
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.920975601375681e-06,
+      "loss": 0.0501,
+      "step": 26172
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9203882710691134e-06,
+      "loss": 0.0628,
+      "step": 26173
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9198010210259955e-06,
+      "loss": 0.067,
+      "step": 26174
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9192138512521585e-06,
+      "loss": 0.0623,
+      "step": 26175
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9186267617534393e-06,
+      "loss": 0.0373,
+      "step": 26176
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9180397525356678e-06,
+      "loss": 0.0341,
+      "step": 26177
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.917452823604673e-06,
+      "loss": 0.0598,
+      "step": 26178
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.916865974966292e-06,
+      "loss": 0.0341,
+      "step": 26179
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.916279206626347e-06,
+      "loss": 0.0472,
+      "step": 26180
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9156925185906715e-06,
+      "loss": 0.0565,
+      "step": 26181
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.915105910865096e-06,
+      "loss": 0.0612,
+      "step": 26182
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9145193834554443e-06,
+      "loss": 0.0469,
+      "step": 26183
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9139329363675408e-06,
+      "loss": 0.0441,
+      "step": 26184
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9133465696072174e-06,
+      "loss": 0.0401,
+      "step": 26185
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.912760283180293e-06,
+      "loss": 0.046,
+      "step": 26186
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.912174077092598e-06,
+      "loss": 0.0501,
+      "step": 26187
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9115879513499513e-06,
+      "loss": 0.055,
+      "step": 26188
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9110019059581743e-06,
+      "loss": 0.0613,
+      "step": 26189
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9104159409230916e-06,
+      "loss": 0.045,
+      "step": 26190
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.0342,
+      "step": 26191
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9092442519462963e-06,
+      "loss": 0.0311,
+      "step": 26192
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9086585280162185e-06,
+      "loss": 0.0317,
+      "step": 26193
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9080728844661158e-06,
+      "loss": 0.04,
+      "step": 26194
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.907487321301802e-06,
+      "loss": 0.0336,
+      "step": 26195
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.906901838529096e-06,
+      "loss": 0.0458,
+      "step": 26196
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.90631643615382e-06,
+      "loss": 0.0421,
+      "step": 26197
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.905731114181778e-06,
+      "loss": 0.0379,
+      "step": 26198
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.905145872618791e-06,
+      "loss": 0.0353,
+      "step": 26199
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9045607114706744e-06,
+      "loss": 0.0509,
+      "step": 26200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9039756307432367e-06,
+      "loss": 0.0534,
+      "step": 26201
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9033906304422945e-06,
+      "loss": 0.0329,
+      "step": 26202
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9028057105736575e-06,
+      "loss": 0.0421,
+      "step": 26203
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9022208711431333e-06,
+      "loss": 0.0512,
+      "step": 26204
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.901636112156534e-06,
+      "loss": 0.051,
+      "step": 26205
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9010514336196751e-06,
+      "loss": 0.0376,
+      "step": 26206
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9004668355383537e-06,
+      "loss": 0.0378,
+      "step": 26207
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8998823179183823e-06,
+      "loss": 0.0563,
+      "step": 26208
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8992978807655692e-06,
+      "loss": 0.0312,
+      "step": 26209
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8987135240857168e-06,
+      "loss": 0.0278,
+      "step": 26210
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8981292478846346e-06,
+      "loss": 0.0337,
+      "step": 26211
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8975450521681248e-06,
+      "loss": 0.052,
+      "step": 26212
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8969609369419872e-06,
+      "loss": 0.0462,
+      "step": 26213
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.896376902212027e-06,
+      "loss": 0.0411,
+      "step": 26214
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8957929479840497e-06,
+      "loss": 0.0406,
+      "step": 26215
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8952090742638528e-06,
+      "loss": 0.0344,
+      "step": 26216
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8946252810572352e-06,
+      "loss": 0.0392,
+      "step": 26217
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.894041568370001e-06,
+      "loss": 0.0552,
+      "step": 26218
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.893457936207942e-06,
+      "loss": 0.0504,
+      "step": 26219
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8928743845768605e-06,
+      "loss": 0.0611,
+      "step": 26220
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8922909134825585e-06,
+      "loss": 0.0431,
+      "step": 26221
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8917075229308212e-06,
+      "loss": 0.0739,
+      "step": 26222
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8911242129274498e-06,
+      "loss": 0.0453,
+      "step": 26223
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8905409834782418e-06,
+      "loss": 0.0523,
+      "step": 26224
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8899578345889846e-06,
+      "loss": 0.0422,
+      "step": 26225
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8893747662654781e-06,
+      "loss": 0.0466,
+      "step": 26226
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8887917785135102e-06,
+      "loss": 0.0592,
+      "step": 26227
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8882088713388713e-06,
+      "loss": 0.0359,
+      "step": 26228
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8876260447473539e-06,
+      "loss": 0.0436,
+      "step": 26229
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8870432987447495e-06,
+      "loss": 0.0569,
+      "step": 26230
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8864606333368473e-06,
+      "loss": 0.0552,
+      "step": 26231
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8858780485294303e-06,
+      "loss": 0.0411,
+      "step": 26232
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8852955443282917e-06,
+      "loss": 0.0478,
+      "step": 26233
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8847131207392144e-06,
+      "loss": 0.0383,
+      "step": 26234
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8841307777679874e-06,
+      "loss": 0.0676,
+      "step": 26235
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8835485154203936e-06,
+      "loss": 0.0403,
+      "step": 26236
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8829663337022163e-06,
+      "loss": 0.0315,
+      "step": 26237
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.88238423261924e-06,
+      "loss": 0.0568,
+      "step": 26238
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8818022121772494e-06,
+      "loss": 0.0432,
+      "step": 26239
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8812202723820262e-06,
+      "loss": 0.0551,
+      "step": 26240
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8806384132393463e-06,
+      "loss": 0.0394,
+      "step": 26241
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8800566347549954e-06,
+      "loss": 0.0354,
+      "step": 26242
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8794749369347499e-06,
+      "loss": 0.0291,
+      "step": 26243
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8788933197843906e-06,
+      "loss": 0.0313,
+      "step": 26244
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8783117833096953e-06,
+      "loss": 0.0441,
+      "step": 26245
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.877730327516437e-06,
+      "loss": 0.0277,
+      "step": 26246
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8771489524103948e-06,
+      "loss": 0.0437,
+      "step": 26247
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8765676579973468e-06,
+      "loss": 0.0525,
+      "step": 26248
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8759864442830622e-06,
+      "loss": 0.0535,
+      "step": 26249
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8754053112733206e-06,
+      "loss": 0.0287,
+      "step": 26250
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.874824258973893e-06,
+      "loss": 0.0471,
+      "step": 26251
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.874243287390547e-06,
+      "loss": 0.0403,
+      "step": 26252
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.87366239652906e-06,
+      "loss": 0.0575,
+      "step": 26253
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.873081586395199e-06,
+      "loss": 0.0306,
+      "step": 26254
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8725008569947366e-06,
+      "loss": 0.0421,
+      "step": 26255
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8719202083334375e-06,
+      "loss": 0.0607,
+      "step": 26256
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.871339640417077e-06,
+      "loss": 0.039,
+      "step": 26257
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8707591532514147e-06,
+      "loss": 0.0436,
+      "step": 26258
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8701787468422227e-06,
+      "loss": 0.0487,
+      "step": 26259
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8695984211952656e-06,
+      "loss": 0.0475,
+      "step": 26260
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.869018176316305e-06,
+      "loss": 0.0547,
+      "step": 26261
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.86843801221111e-06,
+      "loss": 0.0432,
+      "step": 26262
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8678579288854393e-06,
+      "loss": 0.0273,
+      "step": 26263
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8672779263450601e-06,
+      "loss": 0.0471,
+      "step": 26264
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.866698004595733e-06,
+      "loss": 0.052,
+      "step": 26265
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8661181636432147e-06,
+      "loss": 0.0481,
+      "step": 26266
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8655384034932688e-06,
+      "loss": 0.0332,
+      "step": 26267
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8649587241516576e-06,
+      "loss": 0.0498,
+      "step": 26268
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.864379125624136e-06,
+      "loss": 0.0358,
+      "step": 26269
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.86379960791646e-06,
+      "loss": 0.0395,
+      "step": 26270
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.863220171034391e-06,
+      "loss": 0.0371,
+      "step": 26271
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8626408149836816e-06,
+      "loss": 0.0428,
+      "step": 26272
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8620615397700881e-06,
+      "loss": 0.0352,
+      "step": 26273
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8614823453993713e-06,
+      "loss": 0.0427,
+      "step": 26274
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8609032318772735e-06,
+      "loss": 0.0409,
+      "step": 26275
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8603241992095545e-06,
+      "loss": 0.0322,
+      "step": 26276
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8597452474019672e-06,
+      "loss": 0.0527,
+      "step": 26277
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8591663764602586e-06,
+      "loss": 0.0784,
+      "step": 26278
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8585875863901848e-06,
+      "loss": 0.0381,
+      "step": 26279
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8580088771974925e-06,
+      "loss": 0.0585,
+      "step": 26280
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.857430248887927e-06,
+      "loss": 0.0576,
+      "step": 26281
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8568517014672416e-06,
+      "loss": 0.0435,
+      "step": 26282
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.856273234941185e-06,
+      "loss": 0.0414,
+      "step": 26283
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8556948493154992e-06,
+      "loss": 0.0338,
+      "step": 26284
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8551165445959308e-06,
+      "loss": 0.0684,
+      "step": 26285
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8545383207882273e-06,
+      "loss": 0.0492,
+      "step": 26286
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8539601778981287e-06,
+      "loss": 0.0451,
+      "step": 26287
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8533821159313802e-06,
+      "loss": 0.0414,
+      "step": 26288
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.852804134893731e-06,
+      "loss": 0.0909,
+      "step": 26289
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8522262347909104e-06,
+      "loss": 0.0654,
+      "step": 26290
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8516484156286663e-06,
+      "loss": 0.0478,
+      "step": 26291
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8510706774127397e-06,
+      "loss": 0.0535,
+      "step": 26292
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8504930201488692e-06,
+      "loss": 0.0519,
+      "step": 26293
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8499154438427892e-06,
+      "loss": 0.0585,
+      "step": 26294
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8493379485002427e-06,
+      "loss": 0.0287,
+      "step": 26295
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8487605341269622e-06,
+      "loss": 0.043,
+      "step": 26296
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.848183200728686e-06,
+      "loss": 0.0295,
+      "step": 26297
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.847605948311154e-06,
+      "loss": 0.054,
+      "step": 26298
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.847028776880091e-06,
+      "loss": 0.0304,
+      "step": 26299
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8464516864412363e-06,
+      "loss": 0.0383,
+      "step": 26300
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8458746770003243e-06,
+      "loss": 0.0562,
+      "step": 26301
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.845297748563082e-06,
+      "loss": 0.0473,
+      "step": 26302
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8447209011352462e-06,
+      "loss": 0.0483,
+      "step": 26303
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8441441347225442e-06,
+      "loss": 0.0494,
+      "step": 26304
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8435674493307043e-06,
+      "loss": 0.0533,
+      "step": 26305
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8429908449654566e-06,
+      "loss": 0.0397,
+      "step": 26306
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.842414321632532e-06,
+      "loss": 0.0283,
+      "step": 26307
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8418378793376557e-06,
+      "loss": 0.0574,
+      "step": 26308
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8412615180865511e-06,
+      "loss": 0.0408,
+      "step": 26309
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8406852378849492e-06,
+      "loss": 0.045,
+      "step": 26310
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.84010903873857e-06,
+      "loss": 0.0515,
+      "step": 26311
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8395329206531409e-06,
+      "loss": 0.04,
+      "step": 26312
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8389568836343852e-06,
+      "loss": 0.048,
+      "step": 26313
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8383809276880205e-06,
+      "loss": 0.049,
+      "step": 26314
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8378050528197722e-06,
+      "loss": 0.0488,
+      "step": 26315
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8372292590353625e-06,
+      "loss": 0.0552,
+      "step": 26316
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8366535463405076e-06,
+      "loss": 0.0454,
+      "step": 26317
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8360779147409313e-06,
+      "loss": 0.0499,
+      "step": 26318
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8355023642423486e-06,
+      "loss": 0.0417,
+      "step": 26319
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.834926894850476e-06,
+      "loss": 0.0483,
+      "step": 26320
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.834351506571035e-06,
+      "loss": 0.0436,
+      "step": 26321
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.833776199409737e-06,
+      "loss": 0.0362,
+      "step": 26322
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8332009733722978e-06,
+      "loss": 0.0453,
+      "step": 26323
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8326258284644317e-06,
+      "loss": 0.0442,
+      "step": 26324
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8320507646918552e-06,
+      "loss": 0.0604,
+      "step": 26325
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.831475782060277e-06,
+      "loss": 0.0487,
+      "step": 26326
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8309008805754135e-06,
+      "loss": 0.0295,
+      "step": 26327
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.830326060242974e-06,
+      "loss": 0.0449,
+      "step": 26328
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8297513210686647e-06,
+      "loss": 0.0507,
+      "step": 26329
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8291766630582008e-06,
+      "loss": 0.0413,
+      "step": 26330
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8286020862172871e-06,
+      "loss": 0.0287,
+      "step": 26331
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8280275905516353e-06,
+      "loss": 0.0352,
+      "step": 26332
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8274531760669478e-06,
+      "loss": 0.0452,
+      "step": 26333
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8268788427689354e-06,
+      "loss": 0.0428,
+      "step": 26334
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8263045906633004e-06,
+      "loss": 0.0291,
+      "step": 26335
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8257304197557502e-06,
+      "loss": 0.0404,
+      "step": 26336
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.825156330051987e-06,
+      "loss": 0.0376,
+      "step": 26337
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8245823215577108e-06,
+      "loss": 0.0746,
+      "step": 26338
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8240083942786302e-06,
+      "loss": 0.0523,
+      "step": 26339
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8234345482204407e-06,
+      "loss": 0.0431,
+      "step": 26340
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8228607833888456e-06,
+      "loss": 0.0485,
+      "step": 26341
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.822287099789548e-06,
+      "loss": 0.0632,
+      "step": 26342
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8217134974282425e-06,
+      "loss": 0.054,
+      "step": 26343
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8211399763106262e-06,
+      "loss": 0.0566,
+      "step": 26344
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8205665364424019e-06,
+      "loss": 0.0466,
+      "step": 26345
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8199931778292623e-06,
+      "loss": 0.0405,
+      "step": 26346
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.819419900476902e-06,
+      "loss": 0.0543,
+      "step": 26347
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8188467043910197e-06,
+      "loss": 0.0378,
+      "step": 26348
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8182735895773064e-06,
+      "loss": 0.0601,
+      "step": 26349
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8177005560414562e-06,
+      "loss": 0.0754,
+      "step": 26350
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.817127603789165e-06,
+      "loss": 0.031,
+      "step": 26351
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8165547328261212e-06,
+      "loss": 0.0635,
+      "step": 26352
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8159819431580139e-06,
+      "loss": 0.0491,
+      "step": 26353
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8154092347905395e-06,
+      "loss": 0.0484,
+      "step": 26354
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.81483660772938e-06,
+      "loss": 0.042,
+      "step": 26355
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8142640619802298e-06,
+      "loss": 0.0306,
+      "step": 26356
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8136915975487747e-06,
+      "loss": 0.0414,
+      "step": 26357
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8131192144406984e-06,
+      "loss": 0.0406,
+      "step": 26358
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8125469126616902e-06,
+      "loss": 0.0438,
+      "step": 26359
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8119746922174375e-06,
+      "loss": 0.0282,
+      "step": 26360
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8114025531136225e-06,
+      "loss": 0.0455,
+      "step": 26361
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8108304953559252e-06,
+      "loss": 0.0369,
+      "step": 26362
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8102585189500355e-06,
+      "loss": 0.0409,
+      "step": 26363
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8096866239016285e-06,
+      "loss": 0.0583,
+      "step": 26364
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.809114810216388e-06,
+      "loss": 0.0359,
+      "step": 26365
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8085430779000024e-06,
+      "loss": 0.0541,
+      "step": 26366
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8079714269581372e-06,
+      "loss": 0.0481,
+      "step": 26367
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.807399857396479e-06,
+      "loss": 0.0294,
+      "step": 26368
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8068283692207079e-06,
+      "loss": 0.0435,
+      "step": 26369
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8062569624364944e-06,
+      "loss": 0.0403,
+      "step": 26370
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8056856370495224e-06,
+      "loss": 0.0382,
+      "step": 26371
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8051143930654635e-06,
+      "loss": 0.038,
+      "step": 26372
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.80454323048999e-06,
+      "loss": 0.0347,
+      "step": 26373
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8039721493287787e-06,
+      "loss": 0.0513,
+      "step": 26374
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8034011495875082e-06,
+      "loss": 0.0625,
+      "step": 26375
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8028302312718393e-06,
+      "loss": 0.0651,
+      "step": 26376
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.802259394387451e-06,
+      "loss": 0.0485,
+      "step": 26377
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8016886389400146e-06,
+      "loss": 0.043,
+      "step": 26378
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.801117964935195e-06,
+      "loss": 0.0432,
+      "step": 26379
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.800547372378667e-06,
+      "loss": 0.0496,
+      "step": 26380
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7999768612760958e-06,
+      "loss": 0.0516,
+      "step": 26381
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7994064316331482e-06,
+      "loss": 0.0422,
+      "step": 26382
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7988360834554909e-06,
+      "loss": 0.0415,
+      "step": 26383
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7982658167487944e-06,
+      "loss": 0.0365,
+      "step": 26384
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.79769563151872e-06,
+      "loss": 0.0378,
+      "step": 26385
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7971255277709299e-06,
+      "loss": 0.0529,
+      "step": 26386
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7965555055110918e-06,
+      "loss": 0.0461,
+      "step": 26387
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7959855647448642e-06,
+      "loss": 0.0446,
+      "step": 26388
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.795415705477911e-06,
+      "loss": 0.0522,
+      "step": 26389
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7948459277158991e-06,
+      "loss": 0.0396,
+      "step": 26390
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.794276231464478e-06,
+      "loss": 0.0418,
+      "step": 26391
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7937066167293116e-06,
+      "loss": 0.0598,
+      "step": 26392
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7931370835160612e-06,
+      "loss": 0.0478,
+      "step": 26393
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.792567631830381e-06,
+      "loss": 0.0437,
+      "step": 26394
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7919982616779308e-06,
+      "loss": 0.0566,
+      "step": 26395
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7914289730643663e-06,
+      "loss": 0.0405,
+      "step": 26396
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7908597659953386e-06,
+      "loss": 0.041,
+      "step": 26397
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7902906404765064e-06,
+      "loss": 0.0511,
+      "step": 26398
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7897215965135251e-06,
+      "loss": 0.0297,
+      "step": 26399
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7891526341120448e-06,
+      "loss": 0.0418,
+      "step": 26400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.788583753277715e-06,
+      "loss": 0.0401,
+      "step": 26401
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7880149540161928e-06,
+      "loss": 0.0409,
+      "step": 26402
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7874462363331224e-06,
+      "loss": 0.0596,
+      "step": 26403
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7868776002341604e-06,
+      "loss": 0.0399,
+      "step": 26404
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.786309045724952e-06,
+      "loss": 0.0485,
+      "step": 26405
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7857405728111432e-06,
+      "loss": 0.0317,
+      "step": 26406
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7851721814983837e-06,
+      "loss": 0.0331,
+      "step": 26407
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.784603871792322e-06,
+      "loss": 0.0649,
+      "step": 26408
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7840356436986016e-06,
+      "loss": 0.048,
+      "step": 26409
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7834674972228638e-06,
+      "loss": 0.0387,
+      "step": 26410
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7828994323707594e-06,
+      "loss": 0.0373,
+      "step": 26411
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.782331449147925e-06,
+      "loss": 0.0492,
+      "step": 26412
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7817635475600093e-06,
+      "loss": 0.0496,
+      "step": 26413
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7811957276126511e-06,
+      "loss": 0.0476,
+      "step": 26414
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7806279893114874e-06,
+      "loss": 0.0389,
+      "step": 26415
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7800603326621623e-06,
+      "loss": 0.0348,
+      "step": 26416
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7794927576703158e-06,
+      "loss": 0.031,
+      "step": 26417
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7789252643415821e-06,
+      "loss": 0.048,
+      "step": 26418
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7783578526816049e-06,
+      "loss": 0.0364,
+      "step": 26419
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7777905226960168e-06,
+      "loss": 0.0443,
+      "step": 26420
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7772232743904515e-06,
+      "loss": 0.0358,
+      "step": 26421
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.776656107770549e-06,
+      "loss": 0.0528,
+      "step": 26422
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7760890228419391e-06,
+      "loss": 0.038,
+      "step": 26423
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7755220196102607e-06,
+      "loss": 0.0472,
+      "step": 26424
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7749550980811425e-06,
+      "loss": 0.0624,
+      "step": 26425
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7743882582602145e-06,
+      "loss": 0.0303,
+      "step": 26426
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.773821500153111e-06,
+      "loss": 0.0552,
+      "step": 26427
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7732548237654634e-06,
+      "loss": 0.0401,
+      "step": 26428
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.772688229102899e-06,
+      "loss": 0.0319,
+      "step": 26429
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7721217161710447e-06,
+      "loss": 0.0439,
+      "step": 26430
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7715552849755324e-06,
+      "loss": 0.0606,
+      "step": 26431
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7709889355219845e-06,
+      "loss": 0.0333,
+      "step": 26432
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.770422667816032e-06,
+      "loss": 0.0557,
+      "step": 26433
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7698564818632979e-06,
+      "loss": 0.0448,
+      "step": 26434
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7692903776694037e-06,
+      "loss": 0.0419,
+      "step": 26435
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.768724355239977e-06,
+      "loss": 0.0403,
+      "step": 26436
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7681584145806407e-06,
+      "loss": 0.0546,
+      "step": 26437
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7675925556970175e-06,
+      "loss": 0.0248,
+      "step": 26438
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7670267785947238e-06,
+      "loss": 0.0508,
+      "step": 26439
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.766461083279386e-06,
+      "loss": 0.052,
+      "step": 26440
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7658954697566178e-06,
+      "loss": 0.0319,
+      "step": 26441
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.765329938032042e-06,
+      "loss": 0.051,
+      "step": 26442
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7647644881112825e-06,
+      "loss": 0.0502,
+      "step": 26443
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.764199119999943e-06,
+      "loss": 0.0446,
+      "step": 26444
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7636338337036474e-06,
+      "loss": 0.047,
+      "step": 26445
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7630686292280142e-06,
+      "loss": 0.0409,
+      "step": 26446
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7625035065786521e-06,
+      "loss": 0.0427,
+      "step": 26447
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7619384657611793e-06,
+      "loss": 0.039,
+      "step": 26448
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7613735067812088e-06,
+      "loss": 0.0358,
+      "step": 26449
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7608086296443483e-06,
+      "loss": 0.0541,
+      "step": 26450
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7602438343562122e-06,
+      "loss": 0.0354,
+      "step": 26451
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7596791209224152e-06,
+      "loss": 0.0307,
+      "step": 26452
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7591144893485633e-06,
+      "loss": 0.0636,
+      "step": 26453
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7585499396402629e-06,
+      "loss": 0.0414,
+      "step": 26454
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7579854718031285e-06,
+      "loss": 0.0495,
+      "step": 26455
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.757421085842761e-06,
+      "loss": 0.054,
+      "step": 26456
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.756856781764772e-06,
+      "loss": 0.0547,
+      "step": 26457
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7562925595747693e-06,
+      "loss": 0.053,
+      "step": 26458
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7557284192783497e-06,
+      "loss": 0.0317,
+      "step": 26459
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.755164360881122e-06,
+      "loss": 0.07,
+      "step": 26460
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7546003843886926e-06,
+      "loss": 0.0296,
+      "step": 26461
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.754036489806662e-06,
+      "loss": 0.0405,
+      "step": 26462
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7534726771406275e-06,
+      "loss": 0.0461,
+      "step": 26463
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7529089463961957e-06,
+      "loss": 0.0627,
+      "step": 26464
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7523452975789634e-06,
+      "loss": 0.0544,
+      "step": 26465
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7517817306945307e-06,
+      "loss": 0.0315,
+      "step": 26466
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7512182457485028e-06,
+      "loss": 0.0457,
+      "step": 26467
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7506548427464653e-06,
+      "loss": 0.0484,
+      "step": 26468
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7500915216940207e-06,
+      "loss": 0.0338,
+      "step": 26469
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7495282825967686e-06,
+      "loss": 0.0493,
+      "step": 26470
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.748965125460298e-06,
+      "loss": 0.0425,
+      "step": 26471
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7484020502902099e-06,
+      "loss": 0.0394,
+      "step": 26472
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7478390570920934e-06,
+      "loss": 0.0548,
+      "step": 26473
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7472761458715403e-06,
+      "loss": 0.0432,
+      "step": 26474
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7467133166341454e-06,
+      "loss": 0.0447,
+      "step": 26475
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7461505693855008e-06,
+      "loss": 0.0362,
+      "step": 26476
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7455879041311963e-06,
+      "loss": 0.058,
+      "step": 26477
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7450253208768176e-06,
+      "loss": 0.0476,
+      "step": 26478
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7444628196279577e-06,
+      "loss": 0.0377,
+      "step": 26479
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7439004003902015e-06,
+      "loss": 0.0543,
+      "step": 26480
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.74333806316914e-06,
+      "loss": 0.0535,
+      "step": 26481
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7427758079703572e-06,
+      "loss": 0.0375,
+      "step": 26482
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7422136347994368e-06,
+      "loss": 0.0567,
+      "step": 26483
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7416515436619642e-06,
+      "loss": 0.0411,
+      "step": 26484
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.741089534563527e-06,
+      "loss": 0.039,
+      "step": 26485
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7405276075097033e-06,
+      "loss": 0.07,
+      "step": 26486
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7399657625060795e-06,
+      "loss": 0.0406,
+      "step": 26487
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7394039995582356e-06,
+      "loss": 0.039,
+      "step": 26488
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7388423186717496e-06,
+      "loss": 0.0535,
+      "step": 26489
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7382807198522056e-06,
+      "loss": 0.0444,
+      "step": 26490
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7377192031051803e-06,
+      "loss": 0.0722,
+      "step": 26491
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7371577684362484e-06,
+      "loss": 0.0413,
+      "step": 26492
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7365964158509918e-06,
+      "loss": 0.0447,
+      "step": 26493
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7360351453549884e-06,
+      "loss": 0.0605,
+      "step": 26494
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7354739569538081e-06,
+      "loss": 0.0392,
+      "step": 26495
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7349128506530323e-06,
+      "loss": 0.0319,
+      "step": 26496
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7343518264582315e-06,
+      "loss": 0.0427,
+      "step": 26497
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7337908843749772e-06,
+      "loss": 0.0306,
+      "step": 26498
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7332300244088463e-06,
+      "loss": 0.0401,
+      "step": 26499
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7326692465654061e-06,
+      "loss": 0.0526,
+      "step": 26500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7321085508502311e-06,
+      "loss": 0.0568,
+      "step": 26501
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.731547937268887e-06,
+      "loss": 0.0431,
+      "step": 26502
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7309874058269493e-06,
+      "loss": 0.0528,
+      "step": 26503
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7304269565299791e-06,
+      "loss": 0.0437,
+      "step": 26504
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7298665893835498e-06,
+      "loss": 0.0473,
+      "step": 26505
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7293063043932267e-06,
+      "loss": 0.0507,
+      "step": 26506
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7287461015645713e-06,
+      "loss": 0.028,
+      "step": 26507
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7281859809031554e-06,
+      "loss": 0.0579,
+      "step": 26508
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.727625942414537e-06,
+      "loss": 0.0463,
+      "step": 26509
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7270659861042827e-06,
+      "loss": 0.0489,
+      "step": 26510
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7265061119779581e-06,
+      "loss": 0.0561,
+      "step": 26511
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.725946320041122e-06,
+      "loss": 0.0535,
+      "step": 26512
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7253866102993333e-06,
+      "loss": 0.0275,
+      "step": 26513
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7248269827581555e-06,
+      "loss": 0.073,
+      "step": 26514
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7242674374231483e-06,
+      "loss": 0.0362,
+      "step": 26515
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7237079742998652e-06,
+      "loss": 0.0534,
+      "step": 26516
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7231485933938708e-06,
+      "loss": 0.042,
+      "step": 26517
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7225892947107159e-06,
+      "loss": 0.0623,
+      "step": 26518
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7220300782559596e-06,
+      "loss": 0.0446,
+      "step": 26519
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7214709440351595e-06,
+      "loss": 0.0489,
+      "step": 26520
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.720911892053867e-06,
+      "loss": 0.0557,
+      "step": 26521
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7203529223176352e-06,
+      "loss": 0.0439,
+      "step": 26522
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7197940348320197e-06,
+      "loss": 0.0403,
+      "step": 26523
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7192352296025682e-06,
+      "loss": 0.0441,
+      "step": 26524
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7186765066348388e-06,
+      "loss": 0.0525,
+      "step": 26525
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7181178659343767e-06,
+      "loss": 0.0305,
+      "step": 26526
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7175593075067298e-06,
+      "loss": 0.0521,
+      "step": 26527
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7170008313574494e-06,
+      "loss": 0.0424,
+      "step": 26528
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7164424374920875e-06,
+      "loss": 0.047,
+      "step": 26529
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7158841259161863e-06,
+      "loss": 0.054,
+      "step": 26530
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7153258966352915e-06,
+      "loss": 0.0471,
+      "step": 26531
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.714767749654952e-06,
+      "loss": 0.031,
+      "step": 26532
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7142096849807087e-06,
+      "loss": 0.0715,
+      "step": 26533
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7136517026181076e-06,
+      "loss": 0.0417,
+      "step": 26534
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7130938025726963e-06,
+      "loss": 0.0282,
+      "step": 26535
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7125359848500077e-06,
+      "loss": 0.0395,
+      "step": 26536
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.711978249455587e-06,
+      "loss": 0.037,
+      "step": 26537
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7114205963949793e-06,
+      "loss": 0.0359,
+      "step": 26538
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.710863025673718e-06,
+      "loss": 0.039,
+      "step": 26539
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7103055372973465e-06,
+      "loss": 0.053,
+      "step": 26540
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7097481312714026e-06,
+      "loss": 0.0666,
+      "step": 26541
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7091908076014185e-06,
+      "loss": 0.0435,
+      "step": 26542
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7086335662929354e-06,
+      "loss": 0.0509,
+      "step": 26543
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.708076407351492e-06,
+      "loss": 0.0354,
+      "step": 26544
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7075193307826144e-06,
+      "loss": 0.0707,
+      "step": 26545
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.706962336591842e-06,
+      "loss": 0.0506,
+      "step": 26546
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7064054247847096e-06,
+      "loss": 0.0354,
+      "step": 26547
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7058485953667448e-06,
+      "loss": 0.0377,
+      "step": 26548
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7052918483434844e-06,
+      "loss": 0.0642,
+      "step": 26549
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7047351837204562e-06,
+      "loss": 0.0435,
+      "step": 26550
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7041786015031891e-06,
+      "loss": 0.0433,
+      "step": 26551
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7036221016972122e-06,
+      "loss": 0.0212,
+      "step": 26552
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7030656843080583e-06,
+      "loss": 0.043,
+      "step": 26553
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7025093493412526e-06,
+      "loss": 0.0576,
+      "step": 26554
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.701953096802318e-06,
+      "loss": 0.0393,
+      "step": 26555
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7013969266967856e-06,
+      "loss": 0.0514,
+      "step": 26556
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7008408390301757e-06,
+      "loss": 0.0425,
+      "step": 26557
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7002848338080148e-06,
+      "loss": 0.0371,
+      "step": 26558
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6997289110358318e-06,
+      "loss": 0.0375,
+      "step": 26559
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.699173070719139e-06,
+      "loss": 0.0488,
+      "step": 26560
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.698617312863463e-06,
+      "loss": 0.0356,
+      "step": 26561
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6980616374743276e-06,
+      "loss": 0.078,
+      "step": 26562
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6975060445572477e-06,
+      "loss": 0.0294,
+      "step": 26563
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6969505341177473e-06,
+      "loss": 0.0225,
+      "step": 26564
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6963951061613426e-06,
+      "loss": 0.0474,
+      "step": 26565
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6958397606935495e-06,
+      "loss": 0.035,
+      "step": 26566
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6952844977198868e-06,
+      "loss": 0.0583,
+      "step": 26567
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6947293172458723e-06,
+      "loss": 0.038,
+      "step": 26568
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6941742192770194e-06,
+      "loss": 0.0593,
+      "step": 26569
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6936192038188393e-06,
+      "loss": 0.0387,
+      "step": 26570
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6930642708768518e-06,
+      "loss": 0.0358,
+      "step": 26571
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.692509420456564e-06,
+      "loss": 0.0386,
+      "step": 26572
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6919546525634934e-06,
+      "loss": 0.0414,
+      "step": 26573
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6913999672031467e-06,
+      "loss": 0.0386,
+      "step": 26574
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6908453643810342e-06,
+      "loss": 0.0454,
+      "step": 26575
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6902908441026667e-06,
+      "loss": 0.0298,
+      "step": 26576
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6897364063735545e-06,
+      "loss": 0.03,
+      "step": 26577
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.689182051199204e-06,
+      "loss": 0.0563,
+      "step": 26578
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6886277785851202e-06,
+      "loss": 0.0691,
+      "step": 26579
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6880735885368127e-06,
+      "loss": 0.0474,
+      "step": 26580
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6875194810597827e-06,
+      "loss": 0.0383,
+      "step": 26581
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6869654561595395e-06,
+      "loss": 0.0345,
+      "step": 26582
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6864115138415848e-06,
+      "loss": 0.0272,
+      "step": 26583
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.685857654111419e-06,
+      "loss": 0.0357,
+      "step": 26584
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.0517,
+      "step": 26585
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6847501824364721e-06,
+      "loss": 0.0694,
+      "step": 26586
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6841965705026886e-06,
+      "loss": 0.0584,
+      "step": 26587
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.683643041178703e-06,
+      "loss": 0.0558,
+      "step": 26588
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6830895944700099e-06,
+      "loss": 0.0392,
+      "step": 26589
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6825362303821058e-06,
+      "loss": 0.0577,
+      "step": 26590
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.681982948920493e-06,
+      "loss": 0.0476,
+      "step": 26591
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6814297500906628e-06,
+      "loss": 0.0524,
+      "step": 26592
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6808766338981153e-06,
+      "loss": 0.0334,
+      "step": 26593
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6803236003483437e-06,
+      "loss": 0.0545,
+      "step": 26594
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.679770649446837e-06,
+      "loss": 0.0569,
+      "step": 26595
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.679217781199093e-06,
+      "loss": 0.0576,
+      "step": 26596
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.678664995610606e-06,
+      "loss": 0.0347,
+      "step": 26597
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6781122926868643e-06,
+      "loss": 0.042,
+      "step": 26598
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6775596724333565e-06,
+      "loss": 0.0498,
+      "step": 26599
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6770071348555772e-06,
+      "loss": 0.045,
+      "step": 26600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.676454679959011e-06,
+      "loss": 0.0489,
+      "step": 26601
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6759023077491487e-06,
+      "loss": 0.0297,
+      "step": 26602
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6753500182314785e-06,
+      "loss": 0.0321,
+      "step": 26603
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6747978114114817e-06,
+      "loss": 0.04,
+      "step": 26604
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6742456872946478e-06,
+      "loss": 0.0207,
+      "step": 26605
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.673693645886464e-06,
+      "loss": 0.0417,
+      "step": 26606
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6731416871924112e-06,
+      "loss": 0.0528,
+      "step": 26607
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6725898112179707e-06,
+      "loss": 0.0622,
+      "step": 26608
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6720380179686292e-06,
+      "loss": 0.0438,
+      "step": 26609
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6714863074498645e-06,
+      "loss": 0.0386,
+      "step": 26610
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6709346796671588e-06,
+      "loss": 0.0425,
+      "step": 26611
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6703831346259969e-06,
+      "loss": 0.0417,
+      "step": 26612
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6698316723318487e-06,
+      "loss": 0.0481,
+      "step": 26613
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6692802927901963e-06,
+      "loss": 0.0529,
+      "step": 26614
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6687289960065213e-06,
+      "loss": 0.079,
+      "step": 26615
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6681777819862932e-06,
+      "loss": 0.0398,
+      "step": 26616
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6676266507349947e-06,
+      "loss": 0.0433,
+      "step": 26617
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.667075602258098e-06,
+      "loss": 0.0341,
+      "step": 26618
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6665246365610743e-06,
+      "loss": 0.0477,
+      "step": 26619
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6659737536493981e-06,
+      "loss": 0.0625,
+      "step": 26620
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6654229535285472e-06,
+      "loss": 0.0374,
+      "step": 26621
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6648722362039893e-06,
+      "loss": 0.034,
+      "step": 26622
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6643216016811926e-06,
+      "loss": 0.0637,
+      "step": 26623
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6637710499656313e-06,
+      "loss": 0.0458,
+      "step": 26624
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6632205810627722e-06,
+      "loss": 0.0672,
+      "step": 26625
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6626701949780832e-06,
+      "loss": 0.0554,
+      "step": 26626
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.66211989171704e-06,
+      "loss": 0.0546,
+      "step": 26627
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.661569671285097e-06,
+      "loss": 0.0598,
+      "step": 26628
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6610195336877254e-06,
+      "loss": 0.0407,
+      "step": 26629
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6604694789303933e-06,
+      "loss": 0.0518,
+      "step": 26630
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6599195070185624e-06,
+      "loss": 0.0549,
+      "step": 26631
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6593696179576946e-06,
+      "loss": 0.0417,
+      "step": 26632
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6588198117532561e-06,
+      "loss": 0.0452,
+      "step": 26633
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6582700884107038e-06,
+      "loss": 0.0516,
+      "step": 26634
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6577204479355026e-06,
+      "loss": 0.0361,
+      "step": 26635
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6571708903331163e-06,
+      "loss": 0.0389,
+      "step": 26636
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6566214156089943e-06,
+      "loss": 0.0376,
+      "step": 26637
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6560720237686002e-06,
+      "loss": 0.0268,
+      "step": 26638
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6555227148173947e-06,
+      "loss": 0.0412,
+      "step": 26639
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6549734887608304e-06,
+      "loss": 0.0392,
+      "step": 26640
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6544243456043664e-06,
+      "loss": 0.0404,
+      "step": 26641
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6538752853534567e-06,
+      "loss": 0.0673,
+      "step": 26642
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.653326308013553e-06,
+      "loss": 0.0326,
+      "step": 26643
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.652777413590111e-06,
+      "loss": 0.058,
+      "step": 26644
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.652228602088587e-06,
+      "loss": 0.0353,
+      "step": 26645
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.651679873514429e-06,
+      "loss": 0.0469,
+      "step": 26646
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6511312278730862e-06,
+      "loss": 0.0485,
+      "step": 26647
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6505826651700142e-06,
+      "loss": 0.0432,
+      "step": 26648
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6500341854106571e-06,
+      "loss": 0.046,
+      "step": 26649
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.649485788600469e-06,
+      "loss": 0.0445,
+      "step": 26650
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.648937474744895e-06,
+      "loss": 0.0398,
+      "step": 26651
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6483892438493799e-06,
+      "loss": 0.0475,
+      "step": 26652
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6478410959193713e-06,
+      "loss": 0.0364,
+      "step": 26653
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6472930309603175e-06,
+      "loss": 0.0344,
+      "step": 26654
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6467450489776581e-06,
+      "loss": 0.0495,
+      "step": 26655
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6461971499768426e-06,
+      "loss": 0.0648,
+      "step": 26656
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6456493339633105e-06,
+      "loss": 0.0323,
+      "step": 26657
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6451016009425014e-06,
+      "loss": 0.0677,
+      "step": 26658
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6445539509198616e-06,
+      "loss": 0.03,
+      "step": 26659
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.644006383900829e-06,
+      "loss": 0.0529,
+      "step": 26660
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6434588998908408e-06,
+      "loss": 0.0435,
+      "step": 26661
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6429114988953388e-06,
+      "loss": 0.0372,
+      "step": 26662
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6423641809197622e-06,
+      "loss": 0.0484,
+      "step": 26663
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6418169459695432e-06,
+      "loss": 0.037,
+      "step": 26664
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6412697940501244e-06,
+      "loss": 0.0414,
+      "step": 26665
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6407227251669366e-06,
+      "loss": 0.0551,
+      "step": 26666
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6401757393254148e-06,
+      "loss": 0.037,
+      "step": 26667
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6396288365309954e-06,
+      "loss": 0.0622,
+      "step": 26668
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6390820167891074e-06,
+      "loss": 0.0542,
+      "step": 26669
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6385352801051879e-06,
+      "loss": 0.038,
+      "step": 26670
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6379886264846624e-06,
+      "loss": 0.0339,
+      "step": 26671
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6374420559329674e-06,
+      "loss": 0.0562,
+      "step": 26672
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6368955684555277e-06,
+      "loss": 0.0297,
+      "step": 26673
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6363491640577756e-06,
+      "loss": 0.046,
+      "step": 26674
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.635802842745139e-06,
+      "loss": 0.0346,
+      "step": 26675
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.63525660452304e-06,
+      "loss": 0.0652,
+      "step": 26676
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6347104493969113e-06,
+      "loss": 0.0461,
+      "step": 26677
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6341643773721749e-06,
+      "loss": 0.0382,
+      "step": 26678
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6336183884542556e-06,
+      "loss": 0.0423,
+      "step": 26679
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.63307248264858e-06,
+      "loss": 0.035,
+      "step": 26680
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6325266599605704e-06,
+      "loss": 0.0459,
+      "step": 26681
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6319809203956449e-06,
+      "loss": 0.0453,
+      "step": 26682
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6314352639592312e-06,
+      "loss": 0.0385,
+      "step": 26683
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6308896906567462e-06,
+      "loss": 0.0572,
+      "step": 26684
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6303442004936077e-06,
+      "loss": 0.0418,
+      "step": 26685
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6297987934752401e-06,
+      "loss": 0.0371,
+      "step": 26686
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.629253469607055e-06,
+      "loss": 0.0461,
+      "step": 26687
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6287082288944744e-06,
+      "loss": 0.0416,
+      "step": 26688
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6281630713429153e-06,
+      "loss": 0.0306,
+      "step": 26689
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6276179969577922e-06,
+      "loss": 0.0382,
+      "step": 26690
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6270730057445162e-06,
+      "loss": 0.0344,
+      "step": 26691
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6265280977085075e-06,
+      "loss": 0.0403,
+      "step": 26692
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6259832728551738e-06,
+      "loss": 0.0628,
+      "step": 26693
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6254385311899324e-06,
+      "loss": 0.0425,
+      "step": 26694
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6248938727181917e-06,
+      "loss": 0.0232,
+      "step": 26695
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.62434929744536e-06,
+      "loss": 0.0453,
+      "step": 26696
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6238048053768507e-06,
+      "loss": 0.0385,
+      "step": 26697
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6232603965180749e-06,
+      "loss": 0.0433,
+      "step": 26698
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6227160708744372e-06,
+      "loss": 0.0306,
+      "step": 26699
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6221718284513432e-06,
+      "loss": 0.0358,
+      "step": 26700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6216276692542053e-06,
+      "loss": 0.035,
+      "step": 26701
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.621083593288424e-06,
+      "loss": 0.0554,
+      "step": 26702
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6205396005594054e-06,
+      "loss": 0.0442,
+      "step": 26703
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6199956910725588e-06,
+      "loss": 0.0303,
+      "step": 26704
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.619451864833279e-06,
+      "loss": 0.0365,
+      "step": 26705
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6189081218469715e-06,
+      "loss": 0.0528,
+      "step": 26706
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6183644621190419e-06,
+      "loss": 0.0613,
+      "step": 26707
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6178208856548848e-06,
+      "loss": 0.0542,
+      "step": 26708
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6172773924599062e-06,
+      "loss": 0.053,
+      "step": 26709
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6167339825395023e-06,
+      "loss": 0.0388,
+      "step": 26710
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.616190655899068e-06,
+      "loss": 0.0552,
+      "step": 26711
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6156474125440047e-06,
+      "loss": 0.0283,
+      "step": 26712
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6151042524797133e-06,
+      "loss": 0.0373,
+      "step": 26713
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6145611757115787e-06,
+      "loss": 0.0501,
+      "step": 26714
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.614018182245003e-06,
+      "loss": 0.0379,
+      "step": 26715
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6134752720853809e-06,
+      "loss": 0.0384,
+      "step": 26716
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6129324452381013e-06,
+      "loss": 0.0372,
+      "step": 26717
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.612389701708561e-06,
+      "loss": 0.0433,
+      "step": 26718
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6118470415021514e-06,
+      "loss": 0.0308,
+      "step": 26719
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6113044646242582e-06,
+      "loss": 0.0427,
+      "step": 26720
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6107619710802758e-06,
+      "loss": 0.0374,
+      "step": 26721
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6102195608755943e-06,
+      "loss": 0.0418,
+      "step": 26722
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.609677234015602e-06,
+      "loss": 0.0409,
+      "step": 26723
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6091349905056808e-06,
+      "loss": 0.0378,
+      "step": 26724
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6085928303512243e-06,
+      "loss": 0.0495,
+      "step": 26725
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6080507535576129e-06,
+      "loss": 0.0723,
+      "step": 26726
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6075087601302341e-06,
+      "loss": 0.0587,
+      "step": 26727
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6069668500744773e-06,
+      "loss": 0.0428,
+      "step": 26728
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6064250233957158e-06,
+      "loss": 0.042,
+      "step": 26729
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6058832800993363e-06,
+      "loss": 0.059,
+      "step": 26730
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6053416201907246e-06,
+      "loss": 0.038,
+      "step": 26731
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6048000436752554e-06,
+      "loss": 0.0306,
+      "step": 26732
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6042585505583142e-06,
+      "loss": 0.042,
+      "step": 26733
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6037171408452778e-06,
+      "loss": 0.0588,
+      "step": 26734
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6031758145415222e-06,
+      "loss": 0.0426,
+      "step": 26735
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6026345716524272e-06,
+      "loss": 0.0309,
+      "step": 26736
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.602093412183372e-06,
+      "loss": 0.0345,
+      "step": 26737
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.60155233613973e-06,
+      "loss": 0.0493,
+      "step": 26738
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6010113435268737e-06,
+      "loss": 0.0436,
+      "step": 26739
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6004704343501832e-06,
+      "loss": 0.0467,
+      "step": 26740
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.599929608615025e-06,
+      "loss": 0.0483,
+      "step": 26741
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5993888663267797e-06,
+      "loss": 0.0458,
+      "step": 26742
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.598848207490814e-06,
+      "loss": 0.0473,
+      "step": 26743
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.598307632112497e-06,
+      "loss": 0.0576,
+      "step": 26744
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5977671401972018e-06,
+      "loss": 0.0421,
+      "step": 26745
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5972267317503e-06,
+      "loss": 0.0289,
+      "step": 26746
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.596686406777157e-06,
+      "loss": 0.0529,
+      "step": 26747
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5961461652831389e-06,
+      "loss": 0.0314,
+      "step": 26748
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5956060072736167e-06,
+      "loss": 0.0323,
+      "step": 26749
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.595065932753952e-06,
+      "loss": 0.0501,
+      "step": 26750
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5945259417295144e-06,
+      "loss": 0.0505,
+      "step": 26751
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5939860342056656e-06,
+      "loss": 0.0341,
+      "step": 26752
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.593446210187768e-06,
+      "loss": 0.0432,
+      "step": 26753
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.592906469681187e-06,
+      "loss": 0.0592,
+      "step": 26754
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5923668126912806e-06,
+      "loss": 0.0515,
+      "step": 26755
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5918272392234123e-06,
+      "loss": 0.0442,
+      "step": 26756
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5912877492829448e-06,
+      "loss": 0.0412,
+      "step": 26757
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5907483428752356e-06,
+      "loss": 0.0328,
+      "step": 26758
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5902090200056397e-06,
+      "loss": 0.0336,
+      "step": 26759
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5896697806795191e-06,
+      "loss": 0.0417,
+      "step": 26760
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5891306249022276e-06,
+      "loss": 0.0847,
+      "step": 26761
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5885915526791252e-06,
+      "loss": 0.0402,
+      "step": 26762
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5880525640155631e-06,
+      "loss": 0.0384,
+      "step": 26763
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.587513658916896e-06,
+      "loss": 0.0484,
+      "step": 26764
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5869748373884786e-06,
+      "loss": 0.0573,
+      "step": 26765
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5864360994356654e-06,
+      "loss": 0.0353,
+      "step": 26766
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5858974450638064e-06,
+      "loss": 0.0471,
+      "step": 26767
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5853588742782512e-06,
+      "loss": 0.0344,
+      "step": 26768
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5848203870843526e-06,
+      "loss": 0.0502,
+      "step": 26769
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5842819834874569e-06,
+      "loss": 0.0439,
+      "step": 26770
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5837436634929171e-06,
+      "loss": 0.0522,
+      "step": 26771
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5832054271060782e-06,
+      "loss": 0.0354,
+      "step": 26772
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5826672743322847e-06,
+      "loss": 0.0412,
+      "step": 26773
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5821292051768844e-06,
+      "loss": 0.0359,
+      "step": 26774
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5815912196452265e-06,
+      "loss": 0.0467,
+      "step": 26775
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5810533177426524e-06,
+      "loss": 0.0349,
+      "step": 26776
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5805154994745031e-06,
+      "loss": 0.0422,
+      "step": 26777
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5799777648461257e-06,
+      "loss": 0.0566,
+      "step": 26778
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5794401138628579e-06,
+      "loss": 0.0664,
+      "step": 26779
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5789025465300423e-06,
+      "loss": 0.0383,
+      "step": 26780
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5783650628530256e-06,
+      "loss": 0.0417,
+      "step": 26781
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.577827662837136e-06,
+      "loss": 0.0404,
+      "step": 26782
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5772903464877176e-06,
+      "loss": 0.0427,
+      "step": 26783
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5767531138101099e-06,
+      "loss": 0.0201,
+      "step": 26784
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5762159648096453e-06,
+      "loss": 0.0496,
+      "step": 26785
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.575678899491666e-06,
+      "loss": 0.059,
+      "step": 26786
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5751419178615025e-06,
+      "loss": 0.053,
+      "step": 26787
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5746050199244888e-06,
+      "loss": 0.0511,
+      "step": 26788
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5740682056859602e-06,
+      "loss": 0.0552,
+      "step": 26789
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5735314751512508e-06,
+      "loss": 0.0334,
+      "step": 26790
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5729948283256913e-06,
+      "loss": 0.0545,
+      "step": 26791
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5724582652146103e-06,
+      "loss": 0.0528,
+      "step": 26792
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5719217858233438e-06,
+      "loss": 0.0393,
+      "step": 26793
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.571385390157214e-06,
+      "loss": 0.0531,
+      "step": 26794
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5708490782215535e-06,
+      "loss": 0.0336,
+      "step": 26795
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5703128500216947e-06,
+      "loss": 0.0399,
+      "step": 26796
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5697767055629553e-06,
+      "loss": 0.0297,
+      "step": 26797
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5692406448506648e-06,
+      "loss": 0.0383,
+      "step": 26798
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.568704667890153e-06,
+      "loss": 0.047,
+      "step": 26799
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.568168774686739e-06,
+      "loss": 0.0374,
+      "step": 26800
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5676329652457478e-06,
+      "loss": 0.0418,
+      "step": 26801
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5670972395725037e-06,
+      "loss": 0.0254,
+      "step": 26802
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5665615976723247e-06,
+      "loss": 0.0383,
+      "step": 26803
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5660260395505357e-06,
+      "loss": 0.042,
+      "step": 26804
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5654905652124596e-06,
+      "loss": 0.0388,
+      "step": 26805
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5649551746634083e-06,
+      "loss": 0.0457,
+      "step": 26806
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.564419867908703e-06,
+      "loss": 0.0337,
+      "step": 26807
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5638846449536659e-06,
+      "loss": 0.0408,
+      "step": 26808
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5633495058036086e-06,
+      "loss": 0.0428,
+      "step": 26809
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.562814450463852e-06,
+      "loss": 0.0337,
+      "step": 26810
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5622794789397078e-06,
+      "loss": 0.0558,
+      "step": 26811
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5617445912364904e-06,
+      "loss": 0.0794,
+      "step": 26812
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5612097873595134e-06,
+      "loss": 0.0393,
+      "step": 26813
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5606750673140936e-06,
+      "loss": 0.0485,
+      "step": 26814
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.560140431105539e-06,
+      "loss": 0.0589,
+      "step": 26815
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.55960587873916e-06,
+      "loss": 0.0332,
+      "step": 26816
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5590714102202708e-06,
+      "loss": 0.0448,
+      "step": 26817
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5585370255541765e-06,
+      "loss": 0.0508,
+      "step": 26818
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5580027247461893e-06,
+      "loss": 0.0497,
+      "step": 26819
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.557468507801616e-06,
+      "loss": 0.0466,
+      "step": 26820
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5569343747257614e-06,
+      "loss": 0.062,
+      "step": 26821
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5564003255239323e-06,
+      "loss": 0.0531,
+      "step": 26822
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.555866360201438e-06,
+      "loss": 0.0537,
+      "step": 26823
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5553324787635772e-06,
+      "loss": 0.0511,
+      "step": 26824
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5547986812156579e-06,
+      "loss": 0.0588,
+      "step": 26825
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5542649675629827e-06,
+      "loss": 0.0425,
+      "step": 26826
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5537313378108487e-06,
+      "loss": 0.0507,
+      "step": 26827
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5531977919645625e-06,
+      "loss": 0.0397,
+      "step": 26828
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5526643300294231e-06,
+      "loss": 0.0471,
+      "step": 26829
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5521309520107264e-06,
+      "loss": 0.0314,
+      "step": 26830
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5515976579137736e-06,
+      "loss": 0.063,
+      "step": 26831
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5510644477438651e-06,
+      "loss": 0.0359,
+      "step": 26832
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5505313215062922e-06,
+      "loss": 0.0453,
+      "step": 26833
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5499982792063573e-06,
+      "loss": 0.0426,
+      "step": 26834
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5494653208493527e-06,
+      "loss": 0.0635,
+      "step": 26835
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5489324464405696e-06,
+      "loss": 0.0498,
+      "step": 26836
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5483996559853077e-06,
+      "loss": 0.0421,
+      "step": 26837
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5478669494888544e-06,
+      "loss": 0.0256,
+      "step": 26838
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5473343269565067e-06,
+      "loss": 0.0412,
+      "step": 26839
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5468017883935504e-06,
+      "loss": 0.0481,
+      "step": 26840
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5462693338052803e-06,
+      "loss": 0.0364,
+      "step": 26841
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.545736963196982e-06,
+      "loss": 0.03,
+      "step": 26842
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5452046765739481e-06,
+      "loss": 0.0464,
+      "step": 26843
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5446724739414642e-06,
+      "loss": 0.0437,
+      "step": 26844
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5441403553048162e-06,
+      "loss": 0.0541,
+      "step": 26845
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.543608320669292e-06,
+      "loss": 0.0486,
+      "step": 26846
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5430763700401751e-06,
+      "loss": 0.036,
+      "step": 26847
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5425445034227504e-06,
+      "loss": 0.0472,
+      "step": 26848
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5420127208223057e-06,
+      "loss": 0.033,
+      "step": 26849
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.541481022244119e-06,
+      "loss": 0.0437,
+      "step": 26850
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5409494076934717e-06,
+      "loss": 0.0301,
+      "step": 26851
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5404178771756495e-06,
+      "loss": 0.0452,
+      "step": 26852
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5398864306959294e-06,
+      "loss": 0.0332,
+      "step": 26853
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5393550682595882e-06,
+      "loss": 0.0597,
+      "step": 26854
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5388237898719105e-06,
+      "loss": 0.0445,
+      "step": 26855
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5382925955381688e-06,
+      "loss": 0.0715,
+      "step": 26856
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5377614852636425e-06,
+      "loss": 0.0352,
+      "step": 26857
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5372304590536103e-06,
+      "loss": 0.0396,
+      "step": 26858
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5366995169133448e-06,
+      "loss": 0.048,
+      "step": 26859
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5361686588481172e-06,
+      "loss": 0.0549,
+      "step": 26860
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.535637884863206e-06,
+      "loss": 0.0514,
+      "step": 26861
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5351071949638807e-06,
+      "loss": 0.0743,
+      "step": 26862
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5345765891554166e-06,
+      "loss": 0.0514,
+      "step": 26863
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5340460674430835e-06,
+      "loss": 0.0663,
+      "step": 26864
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5335156298321485e-06,
+      "loss": 0.054,
+      "step": 26865
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.532985276327883e-06,
+      "loss": 0.0339,
+      "step": 26866
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5324550069355583e-06,
+      "loss": 0.0487,
+      "step": 26867
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5319248216604399e-06,
+      "loss": 0.0674,
+      "step": 26868
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5313947205077928e-06,
+      "loss": 0.0332,
+      "step": 26869
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5308647034828871e-06,
+      "loss": 0.0415,
+      "step": 26870
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5303347705909833e-06,
+      "loss": 0.053,
+      "step": 26871
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.529804921837349e-06,
+      "loss": 0.0415,
+      "step": 26872
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5292751572272512e-06,
+      "loss": 0.0357,
+      "step": 26873
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5287454767659438e-06,
+      "loss": 0.0503,
+      "step": 26874
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5282158804586934e-06,
+      "loss": 0.0336,
+      "step": 26875
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5276863683107634e-06,
+      "loss": 0.0329,
+      "step": 26876
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.52715694032741e-06,
+      "loss": 0.0303,
+      "step": 26877
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5266275965138967e-06,
+      "loss": 0.0481,
+      "step": 26878
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5260983368754788e-06,
+      "loss": 0.0463,
+      "step": 26879
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5255691614174129e-06,
+      "loss": 0.051,
+      "step": 26880
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5250400701449575e-06,
+      "loss": 0.0592,
+      "step": 26881
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5245110630633741e-06,
+      "loss": 0.0449,
+      "step": 26882
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5239821401779076e-06,
+      "loss": 0.0401,
+      "step": 26883
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5234533014938168e-06,
+      "loss": 0.0498,
+      "step": 26884
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.522924547016359e-06,
+      "loss": 0.0562,
+      "step": 26885
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5223958767507817e-06,
+      "loss": 0.0623,
+      "step": 26886
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5218672907023402e-06,
+      "loss": 0.0361,
+      "step": 26887
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5213387888762855e-06,
+      "loss": 0.0506,
+      "step": 26888
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5208103712778632e-06,
+      "loss": 0.0545,
+      "step": 26889
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.520282037912325e-06,
+      "loss": 0.0558,
+      "step": 26890
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5197537887849246e-06,
+      "loss": 0.0662,
+      "step": 26891
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.519225623900904e-06,
+      "loss": 0.0294,
+      "step": 26892
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5186975432655093e-06,
+      "loss": 0.0345,
+      "step": 26893
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5181695468839907e-06,
+      "loss": 0.0448,
+      "step": 26894
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5176416347615886e-06,
+      "loss": 0.0513,
+      "step": 26895
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5171138069035497e-06,
+      "loss": 0.0507,
+      "step": 26896
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5165860633151231e-06,
+      "loss": 0.0456,
+      "step": 26897
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5160584040015415e-06,
+      "loss": 0.0483,
+      "step": 26898
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5155308289680504e-06,
+      "loss": 0.0515,
+      "step": 26899
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5150033382198948e-06,
+      "loss": 0.0354,
+      "step": 26900
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5144759317623082e-06,
+      "loss": 0.0485,
+      "step": 26901
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5139486096005363e-06,
+      "loss": 0.0371,
+      "step": 26902
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.513421371739815e-06,
+      "loss": 0.0408,
+      "step": 26903
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.512894218185379e-06,
+      "loss": 0.0377,
+      "step": 26904
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5123671489424662e-06,
+      "loss": 0.0563,
+      "step": 26905
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5118401640163184e-06,
+      "loss": 0.0622,
+      "step": 26906
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5113132634121651e-06,
+      "loss": 0.0655,
+      "step": 26907
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5107864471352397e-06,
+      "loss": 0.0391,
+      "step": 26908
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5102597151907794e-06,
+      "loss": 0.0539,
+      "step": 26909
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5097330675840128e-06,
+      "loss": 0.0611,
+      "step": 26910
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5092065043201765e-06,
+      "loss": 0.0479,
+      "step": 26911
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5086800254044986e-06,
+      "loss": 0.0325,
+      "step": 26912
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5081536308422073e-06,
+      "loss": 0.0569,
+      "step": 26913
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5076273206385362e-06,
+      "loss": 0.0459,
+      "step": 26914
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5071010947987085e-06,
+      "loss": 0.0569,
+      "step": 26915
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5065749533279572e-06,
+      "loss": 0.0408,
+      "step": 26916
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5060488962315045e-06,
+      "loss": 0.0462,
+      "step": 26917
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5055229235145808e-06,
+      "loss": 0.0306,
+      "step": 26918
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5049970351824061e-06,
+      "loss": 0.037,
+      "step": 26919
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.50447123124021e-06,
+      "loss": 0.0834,
+      "step": 26920
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5039455116932123e-06,
+      "loss": 0.0341,
+      "step": 26921
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5034198765466346e-06,
+      "loss": 0.0337,
+      "step": 26922
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5028943258057017e-06,
+      "loss": 0.0401,
+      "step": 26923
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5023688594756315e-06,
+      "loss": 0.0323,
+      "step": 26924
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5018434775616453e-06,
+      "loss": 0.0526,
+      "step": 26925
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5013181800689646e-06,
+      "loss": 0.028,
+      "step": 26926
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5007929670028065e-06,
+      "loss": 0.049,
+      "step": 26927
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5002678383683844e-06,
+      "loss": 0.0464,
+      "step": 26928
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.499742794170921e-06,
+      "loss": 0.0447,
+      "step": 26929
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.499217834415627e-06,
+      "loss": 0.0536,
+      "step": 26930
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4986929591077215e-06,
+      "loss": 0.0678,
+      "step": 26931
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4981681682524174e-06,
+      "loss": 0.0435,
+      "step": 26932
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4976434618549263e-06,
+      "loss": 0.0292,
+      "step": 26933
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4971188399204606e-06,
+      "loss": 0.051,
+      "step": 26934
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4965943024542361e-06,
+      "loss": 0.0409,
+      "step": 26935
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4960698494614601e-06,
+      "loss": 0.0366,
+      "step": 26936
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4955454809473413e-06,
+      "loss": 0.069,
+      "step": 26937
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4950211969170924e-06,
+      "loss": 0.0345,
+      "step": 26938
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4944969973759182e-06,
+      "loss": 0.0444,
+      "step": 26939
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4939728823290311e-06,
+      "loss": 0.0346,
+      "step": 26940
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4934488517816338e-06,
+      "loss": 0.0464,
+      "step": 26941
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4929249057389295e-06,
+      "loss": 0.0373,
+      "step": 26942
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4924010442061277e-06,
+      "loss": 0.048,
+      "step": 26943
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.491877267188433e-06,
+      "loss": 0.042,
+      "step": 26944
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4913535746910467e-06,
+      "loss": 0.0394,
+      "step": 26945
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4908299667191683e-06,
+      "loss": 0.0419,
+      "step": 26946
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4903064432780057e-06,
+      "loss": 0.0512,
+      "step": 26947
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4897830043727535e-06,
+      "loss": 0.0597,
+      "step": 26948
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4892596500086144e-06,
+      "loss": 0.0622,
+      "step": 26949
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4887363801907917e-06,
+      "loss": 0.0537,
+      "step": 26950
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4882131949244761e-06,
+      "loss": 0.0731,
+      "step": 26951
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4876900942148665e-06,
+      "loss": 0.0438,
+      "step": 26952
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4871670780671643e-06,
+      "loss": 0.0623,
+      "step": 26953
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.486644146486559e-06,
+      "loss": 0.0454,
+      "step": 26954
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4861212994782514e-06,
+      "loss": 0.0471,
+      "step": 26955
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4855985370474312e-06,
+      "loss": 0.0471,
+      "step": 26956
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.485075859199292e-06,
+      "loss": 0.0401,
+      "step": 26957
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.484553265939026e-06,
+      "loss": 0.0276,
+      "step": 26958
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4840307572718294e-06,
+      "loss": 0.0522,
+      "step": 26959
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4835083332028876e-06,
+      "loss": 0.0402,
+      "step": 26960
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4829859937373902e-06,
+      "loss": 0.0337,
+      "step": 26961
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4824637388805297e-06,
+      "loss": 0.0412,
+      "step": 26962
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4819415686374905e-06,
+      "loss": 0.0485,
+      "step": 26963
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.481419483013462e-06,
+      "loss": 0.0382,
+      "step": 26964
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4808974820136345e-06,
+      "loss": 0.0471,
+      "step": 26965
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.480375565643185e-06,
+      "loss": 0.0453,
+      "step": 26966
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4798537339073027e-06,
+      "loss": 0.0332,
+      "step": 26967
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4793319868111733e-06,
+      "loss": 0.0381,
+      "step": 26968
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4788103243599794e-06,
+      "loss": 0.0404,
+      "step": 26969
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4782887465588992e-06,
+      "loss": 0.0418,
+      "step": 26970
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4777672534131183e-06,
+      "loss": 0.0407,
+      "step": 26971
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4772458449278137e-06,
+      "loss": 0.0492,
+      "step": 26972
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4767245211081683e-06,
+      "loss": 0.0422,
+      "step": 26973
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.476203281959363e-06,
+      "loss": 0.0659,
+      "step": 26974
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4756821274865696e-06,
+      "loss": 0.0536,
+      "step": 26975
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4751610576949672e-06,
+      "loss": 0.031,
+      "step": 26976
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4746400725897358e-06,
+      "loss": 0.0451,
+      "step": 26977
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4741191721760472e-06,
+      "loss": 0.0528,
+      "step": 26978
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4735983564590784e-06,
+      "loss": 0.059,
+      "step": 26979
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4730776254440028e-06,
+      "loss": 0.0387,
+      "step": 26980
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4725569791359894e-06,
+      "loss": 0.0531,
+      "step": 26981
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4720364175402147e-06,
+      "loss": 0.039,
+      "step": 26982
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4715159406618506e-06,
+      "loss": 0.0423,
+      "step": 26983
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4709955485060667e-06,
+      "loss": 0.024,
+      "step": 26984
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4704752410780287e-06,
+      "loss": 0.0367,
+      "step": 26985
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4699550183829114e-06,
+      "loss": 0.0304,
+      "step": 26986
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4694348804258773e-06,
+      "loss": 0.0539,
+      "step": 26987
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4689148272120978e-06,
+      "loss": 0.0604,
+      "step": 26988
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4683948587467379e-06,
+      "loss": 0.0738,
+      "step": 26989
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4678749750349597e-06,
+      "loss": 0.072,
+      "step": 26990
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4673551760819305e-06,
+      "loss": 0.04,
+      "step": 26991
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.466835461892817e-06,
+      "loss": 0.0434,
+      "step": 26992
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4663158324727766e-06,
+      "loss": 0.0399,
+      "step": 26993
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.465796287826976e-06,
+      "loss": 0.0421,
+      "step": 26994
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4652768279605734e-06,
+      "loss": 0.0388,
+      "step": 26995
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4647574528787289e-06,
+      "loss": 0.0549,
+      "step": 26996
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4642381625866042e-06,
+      "loss": 0.0455,
+      "step": 26997
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4637189570893585e-06,
+      "loss": 0.0385,
+      "step": 26998
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.463199836392144e-06,
+      "loss": 0.0393,
+      "step": 26999
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4626808005001226e-06,
+      "loss": 0.0495,
+      "step": 27000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.462161849418452e-06,
+      "loss": 0.0404,
+      "step": 27001
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4616429831522826e-06,
+      "loss": 0.0484,
+      "step": 27002
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4611242017067729e-06,
+      "loss": 0.046,
+      "step": 27003
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.460605505087075e-06,
+      "loss": 0.0377,
+      "step": 27004
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4600868932983402e-06,
+      "loss": 0.0417,
+      "step": 27005
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4595683663457228e-06,
+      "loss": 0.0454,
+      "step": 27006
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4590499242343714e-06,
+      "loss": 0.0549,
+      "step": 27007
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.45853156696944e-06,
+      "loss": 0.03,
+      "step": 27008
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4580132945560731e-06,
+      "loss": 0.0488,
+      "step": 27009
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4574951069994248e-06,
+      "loss": 0.0349,
+      "step": 27010
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.456977004304636e-06,
+      "loss": 0.0326,
+      "step": 27011
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4564589864768607e-06,
+      "loss": 0.0279,
+      "step": 27012
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4559410535212425e-06,
+      "loss": 0.0318,
+      "step": 27013
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4554232054429219e-06,
+      "loss": 0.0675,
+      "step": 27014
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.454905442247051e-06,
+      "loss": 0.0462,
+      "step": 27015
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.454387763938766e-06,
+      "loss": 0.0473,
+      "step": 27016
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4538701705232128e-06,
+      "loss": 0.0382,
+      "step": 27017
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.453352662005536e-06,
+      "loss": 0.0464,
+      "step": 27018
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.452835238390874e-06,
+      "loss": 0.0433,
+      "step": 27019
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4523178996843646e-06,
+      "loss": 0.036,
+      "step": 27020
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4518006458911516e-06,
+      "loss": 0.0511,
+      "step": 27021
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4512834770163709e-06,
+      "loss": 0.0399,
+      "step": 27022
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4507663930651583e-06,
+      "loss": 0.0388,
+      "step": 27023
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4502493940426544e-06,
+      "loss": 0.0315,
+      "step": 27024
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4497324799539914e-06,
+      "loss": 0.0546,
+      "step": 27025
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4492156508043065e-06,
+      "loss": 0.0461,
+      "step": 27026
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4486989065987357e-06,
+      "loss": 0.0321,
+      "step": 27027
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4481822473424101e-06,
+      "loss": 0.0643,
+      "step": 27028
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4476656730404605e-06,
+      "loss": 0.0445,
+      "step": 27029
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4471491836980234e-06,
+      "loss": 0.041,
+      "step": 27030
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.446632779320224e-06,
+      "loss": 0.0389,
+      "step": 27031
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4461164599121968e-06,
+      "loss": 0.0548,
+      "step": 27032
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4456002254790702e-06,
+      "loss": 0.0476,
+      "step": 27033
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4450840760259687e-06,
+      "loss": 0.0656,
+      "step": 27034
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.444568011558023e-06,
+      "loss": 0.0501,
+      "step": 27035
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.444052032080362e-06,
+      "loss": 0.0464,
+      "step": 27036
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4435361375981083e-06,
+      "loss": 0.0388,
+      "step": 27037
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4430203281163858e-06,
+      "loss": 0.0382,
+      "step": 27038
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4425046036403211e-06,
+      "loss": 0.0504,
+      "step": 27039
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.441988964175035e-06,
+      "loss": 0.0361,
+      "step": 27040
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4414734097256522e-06,
+      "loss": 0.0439,
+      "step": 27041
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4409579402972973e-06,
+      "loss": 0.0415,
+      "step": 27042
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4404425558950818e-06,
+      "loss": 0.0504,
+      "step": 27043
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4399272565241319e-06,
+      "loss": 0.0612,
+      "step": 27044
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4394120421895674e-06,
+      "loss": 0.0408,
+      "step": 27045
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4388969128965036e-06,
+      "loss": 0.035,
+      "step": 27046
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4383818686500607e-06,
+      "loss": 0.0332,
+      "step": 27047
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4378669094553532e-06,
+      "loss": 0.0388,
+      "step": 27048
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4373520353174942e-06,
+      "loss": 0.0527,
+      "step": 27049
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4368372462416026e-06,
+      "loss": 0.0416,
+      "step": 27050
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4363225422327942e-06,
+      "loss": 0.0571,
+      "step": 27051
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4358079232961752e-06,
+      "loss": 0.044,
+      "step": 27052
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4352933894368616e-06,
+      "loss": 0.047,
+      "step": 27053
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4347789406599666e-06,
+      "loss": 0.0537,
+      "step": 27054
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4342645769705977e-06,
+      "loss": 0.0438,
+      "step": 27055
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.433750298373867e-06,
+      "loss": 0.0378,
+      "step": 27056
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4332361048748834e-06,
+      "loss": 0.0482,
+      "step": 27057
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.432721996478752e-06,
+      "loss": 0.0319,
+      "step": 27058
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4322079731905803e-06,
+      "loss": 0.0381,
+      "step": 27059
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4316940350154796e-06,
+      "loss": 0.0508,
+      "step": 27060
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4311801819585513e-06,
+      "loss": 0.0266,
+      "step": 27061
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4306664140248994e-06,
+      "loss": 0.0424,
+      "step": 27062
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4301527312196306e-06,
+      "loss": 0.0506,
+      "step": 27063
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4296391335478431e-06,
+      "loss": 0.0358,
+      "step": 27064
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4291256210146442e-06,
+      "loss": 0.0361,
+      "step": 27065
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4286121936251362e-06,
+      "loss": 0.0525,
+      "step": 27066
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4280988513844129e-06,
+      "loss": 0.0527,
+      "step": 27067
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.427585594297577e-06,
+      "loss": 0.034,
+      "step": 27068
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4270724223697307e-06,
+      "loss": 0.0279,
+      "step": 27069
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.426559335605966e-06,
+      "loss": 0.0333,
+      "step": 27070
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4260463340113851e-06,
+      "loss": 0.054,
+      "step": 27071
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4255334175910818e-06,
+      "loss": 0.0355,
+      "step": 27072
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.42502058635015e-06,
+      "loss": 0.0433,
+      "step": 27073
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4245078402936874e-06,
+      "loss": 0.0343,
+      "step": 27074
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4239951794267848e-06,
+      "loss": 0.0309,
+      "step": 27075
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.423482603754538e-06,
+      "loss": 0.043,
+      "step": 27076
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4229701132820339e-06,
+      "loss": 0.0347,
+      "step": 27077
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4224577080143708e-06,
+      "loss": 0.0353,
+      "step": 27078
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4219453879566325e-06,
+      "loss": 0.0481,
+      "step": 27079
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4214331531139125e-06,
+      "loss": 0.0373,
+      "step": 27080
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.420921003491299e-06,
+      "loss": 0.0538,
+      "step": 27081
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4204089390938757e-06,
+      "loss": 0.0376,
+      "step": 27082
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4198969599267343e-06,
+      "loss": 0.0584,
+      "step": 27083
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4193850659949582e-06,
+      "loss": 0.048,
+      "step": 27084
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4188732573036346e-06,
+      "loss": 0.071,
+      "step": 27085
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4183615338578449e-06,
+      "loss": 0.0563,
+      "step": 27086
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4178498956626763e-06,
+      "loss": 0.03,
+      "step": 27087
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4173383427232068e-06,
+      "loss": 0.0544,
+      "step": 27088
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4168268750445236e-06,
+      "loss": 0.0483,
+      "step": 27089
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4163154926317047e-06,
+      "loss": 0.0298,
+      "step": 27090
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4158041954898283e-06,
+      "loss": 0.051,
+      "step": 27091
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4152929836239781e-06,
+      "loss": 0.0315,
+      "step": 27092
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.414781857039228e-06,
+      "loss": 0.0433,
+      "step": 27093
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4142708157406571e-06,
+      "loss": 0.0374,
+      "step": 27094
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4137598597333458e-06,
+      "loss": 0.0409,
+      "step": 27095
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4132489890223666e-06,
+      "loss": 0.0448,
+      "step": 27096
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4127382036127934e-06,
+      "loss": 0.0254,
+      "step": 27097
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4122275035097033e-06,
+      "loss": 0.0555,
+      "step": 27098
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4117168887181664e-06,
+      "loss": 0.0511,
+      "step": 27099
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.41120635924326e-06,
+      "loss": 0.0365,
+      "step": 27100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4106959150900513e-06,
+      "loss": 0.0579,
+      "step": 27101
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4101855562636112e-06,
+      "loss": 0.0468,
+      "step": 27102
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4096752827690108e-06,
+      "loss": 0.0417,
+      "step": 27103
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.409165094611321e-06,
+      "loss": 0.0584,
+      "step": 27104
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4086549917956094e-06,
+      "loss": 0.0367,
+      "step": 27105
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4081449743269393e-06,
+      "loss": 0.0519,
+      "step": 27106
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4076350422103834e-06,
+      "loss": 0.0438,
+      "step": 27107
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4071251954510023e-06,
+      "loss": 0.0367,
+      "step": 27108
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.406615434053864e-06,
+      "loss": 0.0658,
+      "step": 27109
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4061057580240323e-06,
+      "loss": 0.0314,
+      "step": 27110
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4055961673665674e-06,
+      "loss": 0.042,
+      "step": 27111
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4050866620865333e-06,
+      "loss": 0.0289,
+      "step": 27112
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4045772421889937e-06,
+      "loss": 0.0434,
+      "step": 27113
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4040679076790076e-06,
+      "loss": 0.0355,
+      "step": 27114
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4035586585616323e-06,
+      "loss": 0.0404,
+      "step": 27115
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4030494948419316e-06,
+      "loss": 0.0561,
+      "step": 27116
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.402540416524959e-06,
+      "loss": 0.0633,
+      "step": 27117
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4020314236157729e-06,
+      "loss": 0.0496,
+      "step": 27118
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4015225161194346e-06,
+      "loss": 0.0565,
+      "step": 27119
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4010136940409912e-06,
+      "loss": 0.0553,
+      "step": 27120
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4005049573855023e-06,
+      "loss": 0.0365,
+      "step": 27121
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3999963061580236e-06,
+      "loss": 0.0369,
+      "step": 27122
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.399487740363602e-06,
+      "loss": 0.0437,
+      "step": 27123
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.398979260007296e-06,
+      "loss": 0.0401,
+      "step": 27124
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3984708650941536e-06,
+      "loss": 0.0608,
+      "step": 27125
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3979625556292242e-06,
+      "loss": 0.0398,
+      "step": 27126
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3974543316175582e-06,
+      "loss": 0.0574,
+      "step": 27127
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3969461930642081e-06,
+      "loss": 0.0335,
+      "step": 27128
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3964381399742178e-06,
+      "loss": 0.0503,
+      "step": 27129
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.395930172352633e-06,
+      "loss": 0.0371,
+      "step": 27130
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3954222902045044e-06,
+      "loss": 0.0369,
+      "step": 27131
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3949144935348735e-06,
+      "loss": 0.046,
+      "step": 27132
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3944067823487861e-06,
+      "loss": 0.0529,
+      "step": 27133
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3938991566512894e-06,
+      "loss": 0.0525,
+      "step": 27134
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3933916164474193e-06,
+      "loss": 0.0356,
+      "step": 27135
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3928841617422218e-06,
+      "loss": 0.0294,
+      "step": 27136
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3923767925407394e-06,
+      "loss": 0.0531,
+      "step": 27137
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3918695088480095e-06,
+      "loss": 0.0529,
+      "step": 27138
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3913623106690711e-06,
+      "loss": 0.0548,
+      "step": 27139
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3908551980089656e-06,
+      "loss": 0.0478,
+      "step": 27140
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3903481708727274e-06,
+      "loss": 0.0541,
+      "step": 27141
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.389841229265395e-06,
+      "loss": 0.0314,
+      "step": 27142
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3893343731920095e-06,
+      "loss": 0.0536,
+      "step": 27143
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3888276026575965e-06,
+      "loss": 0.0385,
+      "step": 27144
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3883209176671942e-06,
+      "loss": 0.0686,
+      "step": 27145
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3878143182258386e-06,
+      "loss": 0.0435,
+      "step": 27146
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.387307804338559e-06,
+      "loss": 0.0384,
+      "step": 27147
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3868013760103905e-06,
+      "loss": 0.0586,
+      "step": 27148
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3862950332463621e-06,
+      "loss": 0.0572,
+      "step": 27149
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.385788776051501e-06,
+      "loss": 0.0282,
+      "step": 27150
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3852826044308398e-06,
+      "loss": 0.0282,
+      "step": 27151
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.384776518389408e-06,
+      "loss": 0.0428,
+      "step": 27152
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3842705179322312e-06,
+      "loss": 0.0477,
+      "step": 27153
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3837646030643347e-06,
+      "loss": 0.0554,
+      "step": 27154
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3832587737907476e-06,
+      "loss": 0.037,
+      "step": 27155
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3827530301164903e-06,
+      "loss": 0.031,
+      "step": 27156
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3822473720465923e-06,
+      "loss": 0.0432,
+      "step": 27157
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.381741799586074e-06,
+      "loss": 0.0547,
+      "step": 27158
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3812363127399564e-06,
+      "loss": 0.0669,
+      "step": 27159
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3807309115132617e-06,
+      "loss": 0.0414,
+      "step": 27160
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3802255959110133e-06,
+      "loss": 0.0307,
+      "step": 27161
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3797203659382296e-06,
+      "loss": 0.0459,
+      "step": 27162
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3792152215999277e-06,
+      "loss": 0.0521,
+      "step": 27163
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.378710162901129e-06,
+      "loss": 0.0352,
+      "step": 27164
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3782051898468463e-06,
+      "loss": 0.0467,
+      "step": 27165
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.377700302442101e-06,
+      "loss": 0.0526,
+      "step": 27166
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3771955006919068e-06,
+      "loss": 0.0616,
+      "step": 27167
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3766907846012756e-06,
+      "loss": 0.0429,
+      "step": 27168
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3761861541752242e-06,
+      "loss": 0.0667,
+      "step": 27169
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3756816094187675e-06,
+      "loss": 0.0274,
+      "step": 27170
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3751771503369127e-06,
+      "loss": 0.0311,
+      "step": 27171
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3746727769346757e-06,
+      "loss": 0.044,
+      "step": 27172
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.374168489217066e-06,
+      "loss": 0.0328,
+      "step": 27173
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3736642871890893e-06,
+      "loss": 0.0359,
+      "step": 27174
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3731601708557608e-06,
+      "loss": 0.0477,
+      "step": 27175
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3726561402220818e-06,
+      "loss": 0.0268,
+      "step": 27176
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.372152195293065e-06,
+      "loss": 0.0397,
+      "step": 27177
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3716483360737121e-06,
+      "loss": 0.0432,
+      "step": 27178
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3711445625690333e-06,
+      "loss": 0.0352,
+      "step": 27179
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3706408747840271e-06,
+      "loss": 0.0457,
+      "step": 27180
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3701372727237038e-06,
+      "loss": 0.0451,
+      "step": 27181
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.369633756393063e-06,
+      "loss": 0.0387,
+      "step": 27182
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3691303257971035e-06,
+      "loss": 0.0507,
+      "step": 27183
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3686269809408325e-06,
+      "loss": 0.0516,
+      "step": 27184
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3681237218292442e-06,
+      "loss": 0.0786,
+      "step": 27185
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3676205484673422e-06,
+      "loss": 0.0446,
+      "step": 27186
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3671174608601256e-06,
+      "loss": 0.0604,
+      "step": 27187
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3666144590125907e-06,
+      "loss": 0.0503,
+      "step": 27188
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3661115429297312e-06,
+      "loss": 0.0389,
+      "step": 27189
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3656087126165485e-06,
+      "loss": 0.0686,
+      "step": 27190
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3651059680780354e-06,
+      "loss": 0.0423,
+      "step": 27191
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3646033093191845e-06,
+      "loss": 0.0427,
+      "step": 27192
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.364100736344992e-06,
+      "loss": 0.0491,
+      "step": 27193
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3635982491604482e-06,
+      "loss": 0.0456,
+      "step": 27194
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3630958477705448e-06,
+      "loss": 0.0502,
+      "step": 27195
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3625935321802763e-06,
+      "loss": 0.0429,
+      "step": 27196
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3620913023946314e-06,
+      "loss": 0.0506,
+      "step": 27197
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3615891584185947e-06,
+      "loss": 0.0392,
+      "step": 27198
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3610871002571613e-06,
+      "loss": 0.0372,
+      "step": 27199
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3605851279153126e-06,
+      "loss": 0.0416,
+      "step": 27200
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3600832413980403e-06,
+      "loss": 0.0397,
+      "step": 27201
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.359581440710327e-06,
+      "loss": 0.0227,
+      "step": 27202
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3590797258571576e-06,
+      "loss": 0.0421,
+      "step": 27203
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3585780968435169e-06,
+      "loss": 0.0433,
+      "step": 27204
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.35807655367439e-06,
+      "loss": 0.0494,
+      "step": 27205
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3575750963547586e-06,
+      "loss": 0.0672,
+      "step": 27206
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3570737248895994e-06,
+      "loss": 0.0396,
+      "step": 27207
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3565724392838998e-06,
+      "loss": 0.0325,
+      "step": 27208
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3560712395426345e-06,
+      "loss": 0.0496,
+      "step": 27209
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3555701256707854e-06,
+      "loss": 0.0691,
+      "step": 27210
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3550690976733338e-06,
+      "loss": 0.0394,
+      "step": 27211
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3545681555552482e-06,
+      "loss": 0.0422,
+      "step": 27212
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3540672993215097e-06,
+      "loss": 0.056,
+      "step": 27213
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.353566528977096e-06,
+      "loss": 0.0665,
+      "step": 27214
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3530658445269784e-06,
+      "loss": 0.0418,
+      "step": 27215
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3525652459761328e-06,
+      "loss": 0.0655,
+      "step": 27216
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3520647333295333e-06,
+      "loss": 0.0357,
+      "step": 27217
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3515643065921468e-06,
+      "loss": 0.044,
+      "step": 27218
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3510639657689472e-06,
+      "loss": 0.0341,
+      "step": 27219
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3505637108649116e-06,
+      "loss": 0.0529,
+      "step": 27220
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3500635418849984e-06,
+      "loss": 0.0357,
+      "step": 27221
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3495634588341822e-06,
+      "loss": 0.0325,
+      "step": 27222
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3490634617174314e-06,
+      "loss": 0.0336,
+      "step": 27223
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.34856355053971e-06,
+      "loss": 0.071,
+      "step": 27224
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.348063725305988e-06,
+      "loss": 0.049,
+      "step": 27225
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3475639860212286e-06,
+      "loss": 0.047,
+      "step": 27226
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3470643326903943e-06,
+      "loss": 0.034,
+      "step": 27227
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.346564765318451e-06,
+      "loss": 0.0453,
+      "step": 27228
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3460652839103627e-06,
+      "loss": 0.048,
+      "step": 27229
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.345565888471091e-06,
+      "loss": 0.0446,
+      "step": 27230
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3450665790055917e-06,
+      "loss": 0.049,
+      "step": 27231
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.344567355518832e-06,
+      "loss": 0.045,
+      "step": 27232
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3440682180157661e-06,
+      "loss": 0.0345,
+      "step": 27233
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3435691665013562e-06,
+      "loss": 0.0588,
+      "step": 27234
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3430702009805586e-06,
+      "loss": 0.037,
+      "step": 27235
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.342571321458328e-06,
+      "loss": 0.0527,
+      "step": 27236
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3420725279396206e-06,
+      "loss": 0.0314,
+      "step": 27237
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.341573820429396e-06,
+      "loss": 0.0439,
+      "step": 27238
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3410751989326032e-06,
+      "loss": 0.0407,
+      "step": 27239
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3405766634541995e-06,
+      "loss": 0.0678,
+      "step": 27240
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3400782139991352e-06,
+      "loss": 0.047,
+      "step": 27241
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.33957985057236e-06,
+      "loss": 0.0323,
+      "step": 27242
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3390815731788298e-06,
+      "loss": 0.0324,
+      "step": 27243
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3385833818234883e-06,
+      "loss": 0.0345,
+      "step": 27244
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3380852765112916e-06,
+      "loss": 0.0342,
+      "step": 27245
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3375872572471805e-06,
+      "loss": 0.0289,
+      "step": 27246
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3370893240361093e-06,
+      "loss": 0.0521,
+      "step": 27247
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.336591476883018e-06,
+      "loss": 0.0537,
+      "step": 27248
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3360937157928588e-06,
+      "loss": 0.0445,
+      "step": 27249
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3355960407705726e-06,
+      "loss": 0.022,
+      "step": 27250
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3350984518211018e-06,
+      "loss": 0.0305,
+      "step": 27251
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3346009489493939e-06,
+      "loss": 0.042,
+      "step": 27252
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3341035321603867e-06,
+      "loss": 0.0462,
+      "step": 27253
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3336062014590255e-06,
+      "loss": 0.04,
+      "step": 27254
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3331089568502465e-06,
+      "loss": 0.063,
+      "step": 27255
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3326117983389953e-06,
+      "loss": 0.0544,
+      "step": 27256
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3321147259302047e-06,
+      "loss": 0.0496,
+      "step": 27257
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3316177396288167e-06,
+      "loss": 0.0178,
+      "step": 27258
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3311208394397678e-06,
+      "loss": 0.0382,
+      "step": 27259
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3306240253679914e-06,
+      "loss": 0.0316,
+      "step": 27260
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3301272974184264e-06,
+      "loss": 0.0349,
+      "step": 27261
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3296306555960036e-06,
+      "loss": 0.04,
+      "step": 27262
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.329134099905659e-06,
+      "loss": 0.0381,
+      "step": 27263
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3286376303523284e-06,
+      "loss": 0.0448,
+      "step": 27264
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3281412469409394e-06,
+      "loss": 0.0418,
+      "step": 27265
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3276449496764233e-06,
+      "loss": 0.0467,
+      "step": 27266
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3271487385637127e-06,
+      "loss": 0.0452,
+      "step": 27267
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.326652613607734e-06,
+      "loss": 0.025,
+      "step": 27268
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3261565748134198e-06,
+      "loss": 0.042,
+      "step": 27269
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.325660622185696e-06,
+      "loss": 0.0414,
+      "step": 27270
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3251647557294855e-06,
+      "loss": 0.038,
+      "step": 27271
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3246689754497188e-06,
+      "loss": 0.043,
+      "step": 27272
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3241732813513219e-06,
+      "loss": 0.0463,
+      "step": 27273
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3236776734392166e-06,
+      "loss": 0.0535,
+      "step": 27274
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.323182151718324e-06,
+      "loss": 0.0592,
+      "step": 27275
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.322686716193573e-06,
+      "loss": 0.0403,
+      "step": 27276
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3221913668698783e-06,
+      "loss": 0.0406,
+      "step": 27277
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3216961037521669e-06,
+      "loss": 0.0605,
+      "step": 27278
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3212009268453564e-06,
+      "loss": 0.0505,
+      "step": 27279
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3207058361543623e-06,
+      "loss": 0.0602,
+      "step": 27280
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3202108316841067e-06,
+      "loss": 0.0587,
+      "step": 27281
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3197159134395087e-06,
+      "loss": 0.0428,
+      "step": 27282
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3192210814254824e-06,
+      "loss": 0.0363,
+      "step": 27283
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3187263356469414e-06,
+      "loss": 0.0485,
+      "step": 27284
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3182316761088055e-06,
+      "loss": 0.0413,
+      "step": 27285
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3177371028159835e-06,
+      "loss": 0.065,
+      "step": 27286
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3172426157733908e-06,
+      "loss": 0.0621,
+      "step": 27287
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3167482149859445e-06,
+      "loss": 0.0495,
+      "step": 27288
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3162539004585473e-06,
+      "loss": 0.0522,
+      "step": 27289
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.315759672196113e-06,
+      "loss": 0.0434,
+      "step": 27290
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3152655302035556e-06,
+      "loss": 0.0698,
+      "step": 27291
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3147714744857776e-06,
+      "loss": 0.032,
+      "step": 27292
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3142775050476919e-06,
+      "loss": 0.0633,
+      "step": 27293
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3137836218942036e-06,
+      "loss": 0.0338,
+      "step": 27294
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3132898250302173e-06,
+      "loss": 0.0466,
+      "step": 27295
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3127961144606394e-06,
+      "loss": 0.0352,
+      "step": 27296
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3123024901903781e-06,
+      "loss": 0.0394,
+      "step": 27297
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3118089522243339e-06,
+      "loss": 0.0642,
+      "step": 27298
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3113155005674072e-06,
+      "loss": 0.0389,
+      "step": 27299
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3108221352245054e-06,
+      "loss": 0.0461,
+      "step": 27300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3103288562005245e-06,
+      "loss": 0.0502,
+      "step": 27301
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3098356635003674e-06,
+      "loss": 0.0454,
+      "step": 27302
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3093425571289375e-06,
+      "loss": 0.0577,
+      "step": 27303
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3088495370911247e-06,
+      "loss": 0.0479,
+      "step": 27304
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3083566033918316e-06,
+      "loss": 0.0571,
+      "step": 27305
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3078637560359564e-06,
+      "loss": 0.0457,
+      "step": 27306
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.307370995028393e-06,
+      "loss": 0.0683,
+      "step": 27307
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3068783203740344e-06,
+      "loss": 0.0403,
+      "step": 27308
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3063857320777807e-06,
+      "loss": 0.0314,
+      "step": 27309
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.305893230144518e-06,
+      "loss": 0.0431,
+      "step": 27310
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3054008145791442e-06,
+      "loss": 0.0322,
+      "step": 27311
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3049084853865534e-06,
+      "loss": 0.0453,
+      "step": 27312
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3044162425716278e-06,
+      "loss": 0.0504,
+      "step": 27313
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3039240861392622e-06,
+      "loss": 0.0573,
+      "step": 27314
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.303432016094348e-06,
+      "loss": 0.0331,
+      "step": 27315
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3029400324417696e-06,
+      "loss": 0.0396,
+      "step": 27316
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3024481351864172e-06,
+      "loss": 0.0377,
+      "step": 27317
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3019563243331757e-06,
+      "loss": 0.0408,
+      "step": 27318
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.301464599886929e-06,
+      "loss": 0.0321,
+      "step": 27319
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3009729618525647e-06,
+      "loss": 0.038,
+      "step": 27320
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3004814102349684e-06,
+      "loss": 0.0575,
+      "step": 27321
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2999899450390208e-06,
+      "loss": 0.0394,
+      "step": 27322
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2994985662696014e-06,
+      "loss": 0.0429,
+      "step": 27323
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2990072739315973e-06,
+      "loss": 0.072,
+      "step": 27324
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2985160680298835e-06,
+      "loss": 0.0437,
+      "step": 27325
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2980249485693452e-06,
+      "loss": 0.0488,
+      "step": 27326
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2975339155548582e-06,
+      "loss": 0.0325,
+      "step": 27327
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2970429689912977e-06,
+      "loss": 0.0354,
+      "step": 27328
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2965521088835453e-06,
+      "loss": 0.0422,
+      "step": 27329
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.296061335236477e-06,
+      "loss": 0.0461,
+      "step": 27330
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.295570648054967e-06,
+      "loss": 0.0486,
+      "step": 27331
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2950800473438874e-06,
+      "loss": 0.0665,
+      "step": 27332
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.294589533108116e-06,
+      "loss": 0.0457,
+      "step": 27333
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.294099105352522e-06,
+      "loss": 0.0389,
+      "step": 27334
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2936087640819805e-06,
+      "loss": 0.0392,
+      "step": 27335
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2931185093013621e-06,
+      "loss": 0.0427,
+      "step": 27336
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2926283410155339e-06,
+      "loss": 0.0457,
+      "step": 27337
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2921382592293675e-06,
+      "loss": 0.0435,
+      "step": 27338
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2916482639477324e-06,
+      "loss": 0.0295,
+      "step": 27339
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2911583551754947e-06,
+      "loss": 0.0405,
+      "step": 27340
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2906685329175228e-06,
+      "loss": 0.0367,
+      "step": 27341
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2901787971786816e-06,
+      "loss": 0.0332,
+      "step": 27342
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.289689147963834e-06,
+      "loss": 0.0363,
+      "step": 27343
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2891995852778494e-06,
+      "loss": 0.0387,
+      "step": 27344
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2887101091255849e-06,
+      "loss": 0.0523,
+      "step": 27345
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.288220719511909e-06,
+      "loss": 0.0505,
+      "step": 27346
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2877314164416788e-06,
+      "loss": 0.0284,
+      "step": 27347
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2872421999197582e-06,
+      "loss": 0.0293,
+      "step": 27348
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2867530699510033e-06,
+      "loss": 0.0477,
+      "step": 27349
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.286264026540278e-06,
+      "loss": 0.0687,
+      "step": 27350
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2857750696924375e-06,
+      "loss": 0.0404,
+      "step": 27351
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2852861994123388e-06,
+      "loss": 0.0455,
+      "step": 27352
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2847974157048405e-06,
+      "loss": 0.0459,
+      "step": 27353
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2843087185747949e-06,
+      "loss": 0.0604,
+      "step": 27354
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2838201080270585e-06,
+      "loss": 0.0409,
+      "step": 27355
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2833315840664883e-06,
+      "loss": 0.0595,
+      "step": 27356
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.282843146697934e-06,
+      "loss": 0.0668,
+      "step": 27357
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2823547959262473e-06,
+      "loss": 0.0429,
+      "step": 27358
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2818665317562818e-06,
+      "loss": 0.0486,
+      "step": 27359
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2813783541928859e-06,
+      "loss": 0.04,
+      "step": 27360
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2808902632409093e-06,
+      "loss": 0.0311,
+      "step": 27361
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2804022589052023e-06,
+      "loss": 0.0473,
+      "step": 27362
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.27991434119061e-06,
+      "loss": 0.0313,
+      "step": 27363
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2794265101019821e-06,
+      "loss": 0.0348,
+      "step": 27364
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2789387656441676e-06,
+      "loss": 0.035,
+      "step": 27365
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.278451107822004e-06,
+      "loss": 0.0355,
+      "step": 27366
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2779635366403397e-06,
+      "loss": 0.058,
+      "step": 27367
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2774760521040207e-06,
+      "loss": 0.024,
+      "step": 27368
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2769886542178843e-06,
+      "loss": 0.0351,
+      "step": 27369
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.276501342986779e-06,
+      "loss": 0.0473,
+      "step": 27370
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2760141184155418e-06,
+      "loss": 0.0388,
+      "step": 27371
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2755269805090108e-06,
+      "loss": 0.0475,
+      "step": 27372
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2750399292720284e-06,
+      "loss": 0.0486,
+      "step": 27373
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2745529647094347e-06,
+      "loss": 0.0291,
+      "step": 27374
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2740660868260634e-06,
+      "loss": 0.0472,
+      "step": 27375
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2735792956267512e-06,
+      "loss": 0.0401,
+      "step": 27376
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2730925911163383e-06,
+      "loss": 0.0456,
+      "step": 27377
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2726059732996543e-06,
+      "loss": 0.0498,
+      "step": 27378
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2721194421815353e-06,
+      "loss": 0.0553,
+      "step": 27379
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2716329977668195e-06,
+      "loss": 0.0277,
+      "step": 27380
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2711466400603301e-06,
+      "loss": 0.0296,
+      "step": 27381
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2706603690669028e-06,
+      "loss": 0.0332,
+      "step": 27382
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2701741847913717e-06,
+      "loss": 0.0494,
+      "step": 27383
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.269688087238562e-06,
+      "loss": 0.0475,
+      "step": 27384
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.269202076413305e-06,
+      "loss": 0.0397,
+      "step": 27385
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2687161523204293e-06,
+      "loss": 0.0394,
+      "step": 27386
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2682303149647579e-06,
+      "loss": 0.0549,
+      "step": 27387
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.26774456435112e-06,
+      "loss": 0.0469,
+      "step": 27388
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.267258900484346e-06,
+      "loss": 0.0534,
+      "step": 27389
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2667733233692502e-06,
+      "loss": 0.0321,
+      "step": 27390
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2662878330106632e-06,
+      "loss": 0.045,
+      "step": 27391
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2658024294134074e-06,
+      "loss": 0.0421,
+      "step": 27392
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2653171125823028e-06,
+      "loss": 0.054,
+      "step": 27393
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.264831882522174e-06,
+      "loss": 0.0622,
+      "step": 27394
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2643467392378383e-06,
+      "loss": 0.0364,
+      "step": 27395
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2638616827341133e-06,
+      "loss": 0.0198,
+      "step": 27396
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2633767130158215e-06,
+      "loss": 0.0506,
+      "step": 27397
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2628918300877802e-06,
+      "loss": 0.0696,
+      "step": 27398
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2624070339548056e-06,
+      "loss": 0.0403,
+      "step": 27399
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2619223246217116e-06,
+      "loss": 0.0279,
+      "step": 27400
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2614377020933166e-06,
+      "loss": 0.0405,
+      "step": 27401
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2609531663744324e-06,
+      "loss": 0.0375,
+      "step": 27402
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.260468717469875e-06,
+      "loss": 0.03,
+      "step": 27403
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2599843553844559e-06,
+      "loss": 0.0391,
+      "step": 27404
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.259500080122984e-06,
+      "loss": 0.0458,
+      "step": 27405
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2590158916902718e-06,
+      "loss": 0.0667,
+      "step": 27406
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.258531790091132e-06,
+      "loss": 0.0385,
+      "step": 27407
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.25804777533037e-06,
+      "loss": 0.0446,
+      "step": 27408
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2575638474127982e-06,
+      "loss": 0.0318,
+      "step": 27409
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2570800063432208e-06,
+      "loss": 0.0394,
+      "step": 27410
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2565962521264429e-06,
+      "loss": 0.0536,
+      "step": 27411
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2561125847672738e-06,
+      "loss": 0.0433,
+      "step": 27412
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2556290042705155e-06,
+      "loss": 0.0399,
+      "step": 27413
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2551455106409748e-06,
+      "loss": 0.0685,
+      "step": 27414
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2546621038834506e-06,
+      "loss": 0.0292,
+      "step": 27415
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2541787840027498e-06,
+      "loss": 0.0334,
+      "step": 27416
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2536955510036698e-06,
+      "loss": 0.0545,
+      "step": 27417
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2532124048910143e-06,
+      "loss": 0.0463,
+      "step": 27418
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.252729345669581e-06,
+      "loss": 0.0518,
+      "step": 27419
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2522463733441682e-06,
+      "loss": 0.0478,
+      "step": 27420
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.251763487919575e-06,
+      "loss": 0.0404,
+      "step": 27421
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.251280689400597e-06,
+      "loss": 0.0602,
+      "step": 27422
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2507979777920331e-06,
+      "loss": 0.0352,
+      "step": 27423
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2503153530986755e-06,
+      "loss": 0.0437,
+      "step": 27424
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2498328153253214e-06,
+      "loss": 0.0395,
+      "step": 27425
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2493503644767601e-06,
+      "loss": 0.0509,
+      "step": 27426
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2488680005577902e-06,
+      "loss": 0.0509,
+      "step": 27427
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2483857235732011e-06,
+      "loss": 0.0348,
+      "step": 27428
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.24790353352778e-06,
+      "loss": 0.0289,
+      "step": 27429
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2474214304263222e-06,
+      "loss": 0.0391,
+      "step": 27430
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2469394142736136e-06,
+      "loss": 0.0415,
+      "step": 27431
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2464574850744426e-06,
+      "loss": 0.0502,
+      "step": 27432
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.245975642833601e-06,
+      "loss": 0.0579,
+      "step": 27433
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2454938875558731e-06,
+      "loss": 0.0413,
+      "step": 27434
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2450122192460412e-06,
+      "loss": 0.0583,
+      "step": 27435
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2445306379088951e-06,
+      "loss": 0.0501,
+      "step": 27436
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2440491435492153e-06,
+      "loss": 0.05,
+      "step": 27437
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.243567736171788e-06,
+      "loss": 0.035,
+      "step": 27438
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2430864157813938e-06,
+      "loss": 0.0375,
+      "step": 27439
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2426051823828133e-06,
+      "loss": 0.0368,
+      "step": 27440
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2421240359808273e-06,
+      "loss": 0.0403,
+      "step": 27441
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2416429765802196e-06,
+      "loss": 0.0442,
+      "step": 27442
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2411620041857652e-06,
+      "loss": 0.0407,
+      "step": 27443
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2406811188022416e-06,
+      "loss": 0.0438,
+      "step": 27444
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2402003204344293e-06,
+      "loss": 0.0451,
+      "step": 27445
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2397196090871e-06,
+      "loss": 0.0559,
+      "step": 27446
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2392389847650333e-06,
+      "loss": 0.052,
+      "step": 27447
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.238758447473003e-06,
+      "loss": 0.0451,
+      "step": 27448
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.23827799721578e-06,
+      "loss": 0.0425,
+      "step": 27449
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.237797633998138e-06,
+      "loss": 0.0474,
+      "step": 27450
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2373173578248531e-06,
+      "loss": 0.0638,
+      "step": 27451
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.236837168700693e-06,
+      "loss": 0.0631,
+      "step": 27452
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.236357066630426e-06,
+      "loss": 0.0454,
+      "step": 27453
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2358770516188257e-06,
+      "loss": 0.0463,
+      "step": 27454
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2353971236706564e-06,
+      "loss": 0.0266,
+      "step": 27455
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2349172827906875e-06,
+      "loss": 0.0335,
+      "step": 27456
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.234437528983692e-06,
+      "loss": 0.0422,
+      "step": 27457
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2339578622544246e-06,
+      "loss": 0.0328,
+      "step": 27458
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2334782826076552e-06,
+      "loss": 0.0456,
+      "step": 27459
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.232998790048152e-06,
+      "loss": 0.0482,
+      "step": 27460
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2325193845806727e-06,
+      "loss": 0.0361,
+      "step": 27461
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.232040066209983e-06,
+      "loss": 0.0462,
+      "step": 27462
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2315608349408438e-06,
+      "loss": 0.0331,
+      "step": 27463
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2310816907780144e-06,
+      "loss": 0.0387,
+      "step": 27464
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2306026337262544e-06,
+      "loss": 0.0646,
+      "step": 27465
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.230123663790327e-06,
+      "loss": 0.0536,
+      "step": 27466
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2296447809749867e-06,
+      "loss": 0.0367,
+      "step": 27467
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.229165985284989e-06,
+      "loss": 0.0459,
+      "step": 27468
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2286872767250958e-06,
+      "loss": 0.0425,
+      "step": 27469
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.228208655300057e-06,
+      "loss": 0.0624,
+      "step": 27470
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2277301210146298e-06,
+      "loss": 0.0437,
+      "step": 27471
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2272516738735718e-06,
+      "loss": 0.0315,
+      "step": 27472
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2267733138816284e-06,
+      "loss": 0.0329,
+      "step": 27473
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2262950410435558e-06,
+      "loss": 0.0408,
+      "step": 27474
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2258168553641058e-06,
+      "loss": 0.0427,
+      "step": 27475
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.225338756848029e-06,
+      "loss": 0.0423,
+      "step": 27476
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2248607455000706e-06,
+      "loss": 0.0544,
+      "step": 27477
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2243828213249842e-06,
+      "loss": 0.0676,
+      "step": 27478
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.223904984327514e-06,
+      "loss": 0.0423,
+      "step": 27479
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2234272345124087e-06,
+      "loss": 0.0535,
+      "step": 27480
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2229495718844175e-06,
+      "loss": 0.0373,
+      "step": 27481
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2224719964482778e-06,
+      "loss": 0.0352,
+      "step": 27482
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.221994508208738e-06,
+      "loss": 0.0497,
+      "step": 27483
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2215171071705446e-06,
+      "loss": 0.057,
+      "step": 27484
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2210397933384344e-06,
+      "loss": 0.0422,
+      "step": 27485
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.220562566717154e-06,
+      "loss": 0.0447,
+      "step": 27486
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2200854273114415e-06,
+      "loss": 0.0612,
+      "step": 27487
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2196083751260357e-06,
+      "loss": 0.0363,
+      "step": 27488
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.219131410165678e-06,
+      "loss": 0.061,
+      "step": 27489
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2186545324351074e-06,
+      "loss": 0.0602,
+      "step": 27490
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2181777419390596e-06,
+      "loss": 0.0436,
+      "step": 27491
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2177010386822695e-06,
+      "loss": 0.0537,
+      "step": 27492
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2172244226694762e-06,
+      "loss": 0.0289,
+      "step": 27493
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.21674789390541e-06,
+      "loss": 0.0432,
+      "step": 27494
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2162714523948104e-06,
+      "loss": 0.0521,
+      "step": 27495
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2157950981424083e-06,
+      "loss": 0.0485,
+      "step": 27496
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2153188311529319e-06,
+      "loss": 0.0475,
+      "step": 27497
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2148426514311152e-06,
+      "loss": 0.0264,
+      "step": 27498
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.214366558981691e-06,
+      "loss": 0.0502,
+      "step": 27499
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.213890553809387e-06,
+      "loss": 0.0525,
+      "step": 27500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2134146359189303e-06,
+      "loss": 0.0418,
+      "step": 27501
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2129388053150515e-06,
+      "loss": 0.043,
+      "step": 27502
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2124630620024747e-06,
+      "loss": 0.0576,
+      "step": 27503
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2119874059859283e-06,
+      "loss": 0.0343,
+      "step": 27504
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2115118372701373e-06,
+      "loss": 0.0363,
+      "step": 27505
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2110363558598238e-06,
+      "loss": 0.0348,
+      "step": 27506
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2105609617597113e-06,
+      "loss": 0.0332,
+      "step": 27507
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2100856549745266e-06,
+      "loss": 0.0467,
+      "step": 27508
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2096104355089877e-06,
+      "loss": 0.0485,
+      "step": 27509
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2091353033678178e-06,
+      "loss": 0.0454,
+      "step": 27510
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.208660258555735e-06,
+      "loss": 0.0463,
+      "step": 27511
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.208185301077457e-06,
+      "loss": 0.0502,
+      "step": 27512
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2077104309377064e-06,
+      "loss": 0.0483,
+      "step": 27513
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2072356481411961e-06,
+      "loss": 0.0488,
+      "step": 27514
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.206760952692647e-06,
+      "loss": 0.0476,
+      "step": 27515
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.206286344596771e-06,
+      "loss": 0.0497,
+      "step": 27516
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2058118238582862e-06,
+      "loss": 0.0462,
+      "step": 27517
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.205337390481902e-06,
+      "loss": 0.0395,
+      "step": 27518
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2048630444723374e-06,
+      "loss": 0.0687,
+      "step": 27519
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2043887858343007e-06,
+      "loss": 0.0294,
+      "step": 27520
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2039146145725023e-06,
+      "loss": 0.0307,
+      "step": 27521
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2034405306916562e-06,
+      "loss": 0.0366,
+      "step": 27522
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2029665341964681e-06,
+      "loss": 0.0439,
+      "step": 27523
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.202492625091648e-06,
+      "loss": 0.033,
+      "step": 27524
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2020188033819092e-06,
+      "loss": 0.0381,
+      "step": 27525
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2015450690719488e-06,
+      "loss": 0.0429,
+      "step": 27526
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2010714221664787e-06,
+      "loss": 0.0373,
+      "step": 27527
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2005978626702064e-06,
+      "loss": 0.0517,
+      "step": 27528
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2001243905878323e-06,
+      "loss": 0.0493,
+      "step": 27529
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1996510059240585e-06,
+      "loss": 0.043,
+      "step": 27530
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1991777086835932e-06,
+      "loss": 0.0425,
+      "step": 27531
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1987044988711328e-06,
+      "loss": 0.0381,
+      "step": 27532
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.19823137649138e-06,
+      "loss": 0.0553,
+      "step": 27533
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.197758341549039e-06,
+      "loss": 0.0586,
+      "step": 27534
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1972853940488017e-06,
+      "loss": 0.0526,
+      "step": 27535
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1968125339953696e-06,
+      "loss": 0.0515,
+      "step": 27536
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1963397613934436e-06,
+      "loss": 0.0423,
+      "step": 27537
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1958670762477142e-06,
+      "loss": 0.0422,
+      "step": 27538
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1953944785628823e-06,
+      "loss": 0.0854,
+      "step": 27539
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1949219683436396e-06,
+      "loss": 0.0469,
+      "step": 27540
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1944495455946792e-06,
+      "loss": 0.0283,
+      "step": 27541
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.193977210320697e-06,
+      "loss": 0.0379,
+      "step": 27542
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1935049625263851e-06,
+      "loss": 0.0423,
+      "step": 27543
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.193032802216434e-06,
+      "loss": 0.048,
+      "step": 27544
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1925607293955322e-06,
+      "loss": 0.0423,
+      "step": 27545
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1920887440683725e-06,
+      "loss": 0.0382,
+      "step": 27546
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1916168462396405e-06,
+      "loss": 0.0443,
+      "step": 27547
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1911450359140253e-06,
+      "loss": 0.0467,
+      "step": 27548
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1906733130962189e-06,
+      "loss": 0.0603,
+      "step": 27549
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1902016777908998e-06,
+      "loss": 0.038,
+      "step": 27550
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1897301300027552e-06,
+      "loss": 0.0499,
+      "step": 27551
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1892586697364728e-06,
+      "loss": 0.045,
+      "step": 27552
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.188787296996733e-06,
+      "loss": 0.0349,
+      "step": 27553
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1883160117882197e-06,
+      "loss": 0.0447,
+      "step": 27554
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1878448141156162e-06,
+      "loss": 0.0403,
+      "step": 27555
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1873737039835986e-06,
+      "loss": 0.0347,
+      "step": 27556
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1869026813968499e-06,
+      "loss": 0.0399,
+      "step": 27557
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.186431746360054e-06,
+      "loss": 0.0453,
+      "step": 27558
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1859608988778804e-06,
+      "loss": 0.0475,
+      "step": 27559
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1854901389550099e-06,
+      "loss": 0.0498,
+      "step": 27560
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1850194665961223e-06,
+      "loss": 0.0576,
+      "step": 27561
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1845488818058892e-06,
+      "loss": 0.0609,
+      "step": 27562
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1840783845889892e-06,
+      "loss": 0.0329,
+      "step": 27563
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1836079749500939e-06,
+      "loss": 0.0289,
+      "step": 27564
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1831376528938742e-06,
+      "loss": 0.0273,
+      "step": 27565
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1826674184250052e-06,
+      "loss": 0.0588,
+      "step": 27566
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.18219727154816e-06,
+      "loss": 0.0388,
+      "step": 27567
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1817272122680068e-06,
+      "loss": 0.0528,
+      "step": 27568
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.181257240589212e-06,
+      "loss": 0.0434,
+      "step": 27569
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1807873565164507e-06,
+      "loss": 0.0444,
+      "step": 27570
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.180317560054386e-06,
+      "loss": 0.0589,
+      "step": 27571
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1798478512076871e-06,
+      "loss": 0.0264,
+      "step": 27572
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1793782299810209e-06,
+      "loss": 0.0304,
+      "step": 27573
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1789086963790486e-06,
+      "loss": 0.0536,
+      "step": 27574
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1784392504064378e-06,
+      "loss": 0.0373,
+      "step": 27575
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1779698920678529e-06,
+      "loss": 0.0547,
+      "step": 27576
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.177500621367953e-06,
+      "loss": 0.0555,
+      "step": 27577
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1770314383114045e-06,
+      "loss": 0.052,
+      "step": 27578
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.176562342902866e-06,
+      "loss": 0.0304,
+      "step": 27579
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.176093335146995e-06,
+      "loss": 0.0381,
+      "step": 27580
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1756244150484553e-06,
+      "loss": 0.044,
+      "step": 27581
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1751555826118998e-06,
+      "loss": 0.0349,
+      "step": 27582
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1746868378419918e-06,
+      "loss": 0.0381,
+      "step": 27583
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1742181807433828e-06,
+      "loss": 0.0369,
+      "step": 27584
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1737496113207336e-06,
+      "loss": 0.036,
+      "step": 27585
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1732811295786928e-06,
+      "loss": 0.0513,
+      "step": 27586
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.172812735521921e-06,
+      "loss": 0.0456,
+      "step": 27587
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1723444291550666e-06,
+      "loss": 0.0355,
+      "step": 27588
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.171876210482782e-06,
+      "loss": 0.0372,
+      "step": 27589
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1714080795097217e-06,
+      "loss": 0.0592,
+      "step": 27590
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.170940036240531e-06,
+      "loss": 0.0422,
+      "step": 27591
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1704720806798653e-06,
+      "loss": 0.0334,
+      "step": 27592
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1700042128323685e-06,
+      "loss": 0.0412,
+      "step": 27593
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1695364327026915e-06,
+      "loss": 0.0292,
+      "step": 27594
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1690687402954792e-06,
+      "loss": 0.047,
+      "step": 27595
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1686011356153804e-06,
+      "loss": 0.0273,
+      "step": 27596
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.168133618667039e-06,
+      "loss": 0.0494,
+      "step": 27597
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1676661894550967e-06,
+      "loss": 0.0469,
+      "step": 27598
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1671988479842012e-06,
+      "loss": 0.0366,
+      "step": 27599
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1667315942589918e-06,
+      "loss": 0.0427,
+      "step": 27600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1662644282841106e-06,
+      "loss": 0.0399,
+      "step": 27601
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1657973500642028e-06,
+      "loss": 0.0356,
+      "step": 27602
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1653303596039034e-06,
+      "loss": 0.0491,
+      "step": 27603
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.164863456907852e-06,
+      "loss": 0.0603,
+      "step": 27604
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1643966419806907e-06,
+      "loss": 0.0363,
+      "step": 27605
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.163929914827051e-06,
+      "loss": 0.0585,
+      "step": 27606
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1634632754515751e-06,
+      "loss": 0.037,
+      "step": 27607
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1629967238588956e-06,
+      "loss": 0.0287,
+      "step": 27608
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.162530260053647e-06,
+      "loss": 0.0626,
+      "step": 27609
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.162063884040463e-06,
+      "loss": 0.0327,
+      "step": 27610
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.16159759582398e-06,
+      "loss": 0.0525,
+      "step": 27611
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1611313954088277e-06,
+      "loss": 0.0505,
+      "step": 27612
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1606652827996345e-06,
+      "loss": 0.0253,
+      "step": 27613
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1601992580010367e-06,
+      "loss": 0.0509,
+      "step": 27614
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1597333210176587e-06,
+      "loss": 0.0285,
+      "step": 27615
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1592674718541331e-06,
+      "loss": 0.0527,
+      "step": 27616
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.158801710515085e-06,
+      "loss": 0.0308,
+      "step": 27617
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.158336037005141e-06,
+      "loss": 0.0437,
+      "step": 27618
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1578704513289285e-06,
+      "loss": 0.0505,
+      "step": 27619
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1574049534910737e-06,
+      "loss": 0.0543,
+      "step": 27620
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1569395434962004e-06,
+      "loss": 0.0389,
+      "step": 27621
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.156474221348929e-06,
+      "loss": 0.0455,
+      "step": 27622
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.156008987053886e-06,
+      "loss": 0.034,
+      "step": 27623
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1555438406156894e-06,
+      "loss": 0.0387,
+      "step": 27624
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1550787820389619e-06,
+      "loss": 0.0654,
+      "step": 27625
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.154613811328328e-06,
+      "loss": 0.0598,
+      "step": 27626
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1541489284883966e-06,
+      "loss": 0.053,
+      "step": 27627
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1536841335237935e-06,
+      "loss": 0.0588,
+      "step": 27628
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.153219426439135e-06,
+      "loss": 0.0342,
+      "step": 27629
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1527548072390349e-06,
+      "loss": 0.0355,
+      "step": 27630
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1522902759281117e-06,
+      "loss": 0.0459,
+      "step": 27631
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1518258325109799e-06,
+      "loss": 0.0518,
+      "step": 27632
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1513614769922499e-06,
+      "loss": 0.0553,
+      "step": 27633
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1508972093765369e-06,
+      "loss": 0.0435,
+      "step": 27634
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1504330296684552e-06,
+      "loss": 0.0413,
+      "step": 27635
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.149968937872613e-06,
+      "loss": 0.048,
+      "step": 27636
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1495049339936204e-06,
+      "loss": 0.0488,
+      "step": 27637
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.149041018036089e-06,
+      "loss": 0.049,
+      "step": 27638
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.148577190004625e-06,
+      "loss": 0.0382,
+      "step": 27639
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1481134499038371e-06,
+      "loss": 0.0501,
+      "step": 27640
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1476497977383361e-06,
+      "loss": 0.0278,
+      "step": 27641
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1471862335127204e-06,
+      "loss": 0.041,
+      "step": 27642
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1467227572315998e-06,
+      "loss": 0.0302,
+      "step": 27643
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1462593688995781e-06,
+      "loss": 0.042,
+      "step": 27644
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1457960685212599e-06,
+      "loss": 0.031,
+      "step": 27645
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1453328561012423e-06,
+      "loss": 0.0458,
+      "step": 27646
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1448697316441337e-06,
+      "loss": 0.0403,
+      "step": 27647
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1444066951545285e-06,
+      "loss": 0.0439,
+      "step": 27648
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1439437466370307e-06,
+      "loss": 0.0512,
+      "step": 27649
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.143480886096242e-06,
+      "loss": 0.0365,
+      "step": 27650
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1430181135367523e-06,
+      "loss": 0.0336,
+      "step": 27651
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1425554289631635e-06,
+      "loss": 0.036,
+      "step": 27652
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.142092832380074e-06,
+      "loss": 0.0457,
+      "step": 27653
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1416303237920745e-06,
+      "loss": 0.0494,
+      "step": 27654
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1411679032037636e-06,
+      "loss": 0.0346,
+      "step": 27655
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1407055706197344e-06,
+      "loss": 0.0419,
+      "step": 27656
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1402433260445767e-06,
+      "loss": 0.0576,
+      "step": 27657
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1397811694828842e-06,
+      "loss": 0.0537,
+      "step": 27658
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1393191009392501e-06,
+      "loss": 0.0394,
+      "step": 27659
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1388571204182641e-06,
+      "loss": 0.0481,
+      "step": 27660
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1383952279245114e-06,
+      "loss": 0.0525,
+      "step": 27661
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.137933423462585e-06,
+      "loss": 0.0539,
+      "step": 27662
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.13747170703707e-06,
+      "loss": 0.0344,
+      "step": 27663
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.137010078652555e-06,
+      "loss": 0.0586,
+      "step": 27664
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1365485383136255e-06,
+      "loss": 0.0314,
+      "step": 27665
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.136087086024864e-06,
+      "loss": 0.0454,
+      "step": 27666
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1356257217908562e-06,
+      "loss": 0.0521,
+      "step": 27667
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.135164445616187e-06,
+      "loss": 0.0585,
+      "step": 27668
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1347032575054385e-06,
+      "loss": 0.0478,
+      "step": 27669
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.134242157463188e-06,
+      "loss": 0.0677,
+      "step": 27670
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1337811454940218e-06,
+      "loss": 0.0363,
+      "step": 27671
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1333202216025153e-06,
+      "loss": 0.0446,
+      "step": 27672
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.13285938579325e-06,
+      "loss": 0.0375,
+      "step": 27673
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.132398638070803e-06,
+      "loss": 0.0427,
+      "step": 27674
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1319379784397499e-06,
+      "loss": 0.0461,
+      "step": 27675
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1314774069046675e-06,
+      "loss": 0.0448,
+      "step": 27676
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1310169234701352e-06,
+      "loss": 0.0809,
+      "step": 27677
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1305565281407215e-06,
+      "loss": 0.0471,
+      "step": 27678
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1300962209210042e-06,
+      "loss": 0.0336,
+      "step": 27679
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.129636001815555e-06,
+      "loss": 0.0461,
+      "step": 27680
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1291758708289436e-06,
+      "loss": 0.0374,
+      "step": 27681
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.128715827965744e-06,
+      "loss": 0.061,
+      "step": 27682
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1282558732305238e-06,
+      "loss": 0.0415,
+      "step": 27683
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1277960066278549e-06,
+      "loss": 0.0457,
+      "step": 27684
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1273362281623045e-06,
+      "loss": 0.0306,
+      "step": 27685
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1268765378384371e-06,
+      "loss": 0.0413,
+      "step": 27686
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1264169356608223e-06,
+      "loss": 0.0599,
+      "step": 27687
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1259574216340274e-06,
+      "loss": 0.041,
+      "step": 27688
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1254979957626156e-06,
+      "loss": 0.0501,
+      "step": 27689
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1250386580511486e-06,
+      "loss": 0.0503,
+      "step": 27690
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1245794085041928e-06,
+      "loss": 0.0478,
+      "step": 27691
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1241202471263068e-06,
+      "loss": 0.0369,
+      "step": 27692
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1236611739220549e-06,
+      "loss": 0.0614,
+      "step": 27693
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.123202188896001e-06,
+      "loss": 0.0511,
+      "step": 27694
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.122743292052697e-06,
+      "loss": 0.0346,
+      "step": 27695
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1222844833967038e-06,
+      "loss": 0.0381,
+      "step": 27696
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1218257629325836e-06,
+      "loss": 0.0427,
+      "step": 27697
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.12136713066489e-06,
+      "loss": 0.0389,
+      "step": 27698
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1209085865981772e-06,
+      "loss": 0.0354,
+      "step": 27699
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1204501307370053e-06,
+      "loss": 0.0472,
+      "step": 27700
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1199917630859247e-06,
+      "loss": 0.0433,
+      "step": 27701
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1195334836494886e-06,
+      "loss": 0.0585,
+      "step": 27702
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1190752924322557e-06,
+      "loss": 0.0355,
+      "step": 27703
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1186171894387698e-06,
+      "loss": 0.0425,
+      "step": 27704
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1181591746735842e-06,
+      "loss": 0.0385,
+      "step": 27705
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1177012481412518e-06,
+      "loss": 0.0272,
+      "step": 27706
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1172434098463182e-06,
+      "loss": 0.048,
+      "step": 27707
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1167856597933347e-06,
+      "loss": 0.0456,
+      "step": 27708
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1163279979868468e-06,
+      "loss": 0.0219,
+      "step": 27709
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1158704244314e-06,
+      "loss": 0.0458,
+      "step": 27710
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1154129391315393e-06,
+      "loss": 0.038,
+      "step": 27711
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1149555420918146e-06,
+      "loss": 0.0644,
+      "step": 27712
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1144982333167653e-06,
+      "loss": 0.0289,
+      "step": 27713
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1140410128109336e-06,
+      "loss": 0.0354,
+      "step": 27714
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1135838805788646e-06,
+      "loss": 0.0422,
+      "step": 27715
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.113126836625097e-06,
+      "loss": 0.0306,
+      "step": 27716
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1126698809541714e-06,
+      "loss": 0.0317,
+      "step": 27717
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1122130135706321e-06,
+      "loss": 0.0642,
+      "step": 27718
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.111756234479009e-06,
+      "loss": 0.0379,
+      "step": 27719
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1112995436838448e-06,
+      "loss": 0.044,
+      "step": 27720
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1108429411896782e-06,
+      "loss": 0.0458,
+      "step": 27721
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1103864270010401e-06,
+      "loss": 0.0473,
+      "step": 27722
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1099300011224713e-06,
+      "loss": 0.0426,
+      "step": 27723
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1094736635585013e-06,
+      "loss": 0.0441,
+      "step": 27724
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1090174143136635e-06,
+      "loss": 0.0442,
+      "step": 27725
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1085612533924928e-06,
+      "loss": 0.0603,
+      "step": 27726
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1081051807995224e-06,
+      "loss": 0.0538,
+      "step": 27727
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1076491965392766e-06,
+      "loss": 0.0464,
+      "step": 27728
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1071933006162883e-06,
+      "loss": 0.0376,
+      "step": 27729
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1067374930350905e-06,
+      "loss": 0.0485,
+      "step": 27730
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.106281773800204e-06,
+      "loss": 0.0432,
+      "step": 27731
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.105826142916162e-06,
+      "loss": 0.0448,
+      "step": 27732
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1053706003874887e-06,
+      "loss": 0.0428,
+      "step": 27733
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.104915146218707e-06,
+      "loss": 0.0583,
+      "step": 27734
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1044597804143441e-06,
+      "loss": 0.0522,
+      "step": 27735
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1040045029789248e-06,
+      "loss": 0.0458,
+      "step": 27736
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1035493139169694e-06,
+      "loss": 0.0409,
+      "step": 27737
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1030942132329992e-06,
+      "loss": 0.0293,
+      "step": 27738
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.102639200931539e-06,
+      "loss": 0.0532,
+      "step": 27739
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1021842770171044e-06,
+      "loss": 0.046,
+      "step": 27740
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.101729441494218e-06,
+      "loss": 0.0431,
+      "step": 27741
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1012746943673968e-06,
+      "loss": 0.0297,
+      "step": 27742
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.100820035641157e-06,
+      "loss": 0.0374,
+      "step": 27743
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.100365465320017e-06,
+      "loss": 0.0412,
+      "step": 27744
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.099910983408493e-06,
+      "loss": 0.044,
+      "step": 27745
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0994565899110976e-06,
+      "loss": 0.0341,
+      "step": 27746
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0990022848323477e-06,
+      "loss": 0.0546,
+      "step": 27747
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0985480681767557e-06,
+      "loss": 0.0344,
+      "step": 27748
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0980939399488311e-06,
+      "loss": 0.0504,
+      "step": 27749
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.097639900153088e-06,
+      "loss": 0.0545,
+      "step": 27750
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.097185948794035e-06,
+      "loss": 0.0385,
+      "step": 27751
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0967320858761843e-06,
+      "loss": 0.0817,
+      "step": 27752
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.096278311404041e-06,
+      "loss": 0.0363,
+      "step": 27753
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.095824625382117e-06,
+      "loss": 0.0391,
+      "step": 27754
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0953710278149155e-06,
+      "loss": 0.0496,
+      "step": 27755
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.094917518706945e-06,
+      "loss": 0.0513,
+      "step": 27756
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0944640980627108e-06,
+      "loss": 0.0531,
+      "step": 27757
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.094010765886715e-06,
+      "loss": 0.0347,
+      "step": 27758
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0935575221834638e-06,
+      "loss": 0.0412,
+      "step": 27759
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0931043669574559e-06,
+      "loss": 0.0517,
+      "step": 27760
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0926513002131979e-06,
+      "loss": 0.0364,
+      "step": 27761
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0921983219551858e-06,
+      "loss": 0.0566,
+      "step": 27762
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.091745432187924e-06,
+      "loss": 0.0428,
+      "step": 27763
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0912926309159077e-06,
+      "loss": 0.039,
+      "step": 27764
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.090839918143638e-06,
+      "loss": 0.0653,
+      "step": 27765
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0903872938756111e-06,
+      "loss": 0.0369,
+      "step": 27766
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.0618,
+      "step": 27767
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.089482310870269e-06,
+      "loss": 0.0363,
+      "step": 27768
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0890299521419433e-06,
+      "loss": 0.0448,
+      "step": 27769
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0885776819358396e-06,
+      "loss": 0.035,
+      "step": 27770
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.088125500256455e-06,
+      "loss": 0.038,
+      "step": 27771
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0876734071082772e-06,
+      "loss": 0.0329,
+      "step": 27772
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0872214024957973e-06,
+      "loss": 0.0571,
+      "step": 27773
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.086769486423508e-06,
+      "loss": 0.061,
+      "step": 27774
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0863176588958957e-06,
+      "loss": 0.0501,
+      "step": 27775
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0858659199174514e-06,
+      "loss": 0.0468,
+      "step": 27776
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0854142694926627e-06,
+      "loss": 0.0403,
+      "step": 27777
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0849627076260138e-06,
+      "loss": 0.0503,
+      "step": 27778
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.084511234321991e-06,
+      "loss": 0.0418,
+      "step": 27779
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.084059849585084e-06,
+      "loss": 0.043,
+      "step": 27780
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0836085534197716e-06,
+      "loss": 0.0456,
+      "step": 27781
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0831573458305379e-06,
+      "loss": 0.0391,
+      "step": 27782
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.082706226821868e-06,
+      "loss": 0.05,
+      "step": 27783
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0822551963982386e-06,
+      "loss": 0.0501,
+      "step": 27784
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0818042545641349e-06,
+      "loss": 0.0504,
+      "step": 27785
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0813534013240356e-06,
+      "loss": 0.0512,
+      "step": 27786
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.080902636682417e-06,
+      "loss": 0.0721,
+      "step": 27787
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.080451960643758e-06,
+      "loss": 0.0408,
+      "step": 27788
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.080001373212538e-06,
+      "loss": 0.0421,
+      "step": 27789
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0795508743932326e-06,
+      "loss": 0.0351,
+      "step": 27790
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0791004641903124e-06,
+      "loss": 0.0486,
+      "step": 27791
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0786501426082585e-06,
+      "loss": 0.0435,
+      "step": 27792
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0781999096515393e-06,
+      "loss": 0.0468,
+      "step": 27793
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0777497653246295e-06,
+      "loss": 0.0383,
+      "step": 27794
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0772997096320049e-06,
+      "loss": 0.0562,
+      "step": 27795
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0768497425781278e-06,
+      "loss": 0.0444,
+      "step": 27796
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0763998641674734e-06,
+      "loss": 0.0447,
+      "step": 27797
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0759500744045115e-06,
+      "loss": 0.0344,
+      "step": 27798
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0755003732937075e-06,
+      "loss": 0.0414,
+      "step": 27799
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0750507608395322e-06,
+      "loss": 0.0394,
+      "step": 27800
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0746012370464498e-06,
+      "loss": 0.0318,
+      "step": 27801
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0741518019189246e-06,
+      "loss": 0.0396,
+      "step": 27802
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0737024554614228e-06,
+      "loss": 0.0335,
+      "step": 27803
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.07325319767841e-06,
+      "loss": 0.0344,
+      "step": 27804
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.072804028574348e-06,
+      "loss": 0.0311,
+      "step": 27805
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0723549481536966e-06,
+      "loss": 0.055,
+      "step": 27806
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.07190595642092e-06,
+      "loss": 0.0325,
+      "step": 27807
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0714570533804768e-06,
+      "loss": 0.0485,
+      "step": 27808
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.071008239036826e-06,
+      "loss": 0.0548,
+      "step": 27809
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0705595133944312e-06,
+      "loss": 0.0388,
+      "step": 27810
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0701108764577427e-06,
+      "loss": 0.0681,
+      "step": 27811
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0696623282312201e-06,
+      "loss": 0.0565,
+      "step": 27812
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0692138687193222e-06,
+      "loss": 0.0331,
+      "step": 27813
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0687654979265004e-06,
+      "loss": 0.0667,
+      "step": 27814
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0683172158572097e-06,
+      "loss": 0.0361,
+      "step": 27815
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.067869022515905e-06,
+      "loss": 0.0491,
+      "step": 27816
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.067420917907036e-06,
+      "loss": 0.0615,
+      "step": 27817
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.066972902035055e-06,
+      "loss": 0.0491,
+      "step": 27818
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0665249749044171e-06,
+      "loss": 0.059,
+      "step": 27819
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0660771365195644e-06,
+      "loss": 0.04,
+      "step": 27820
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0656293868849498e-06,
+      "loss": 0.038,
+      "step": 27821
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0651817260050235e-06,
+      "loss": 0.0617,
+      "step": 27822
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0647341538842281e-06,
+      "loss": 0.0432,
+      "step": 27823
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0642866705270127e-06,
+      "loss": 0.0392,
+      "step": 27824
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0638392759378236e-06,
+      "loss": 0.0333,
+      "step": 27825
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0633919701211005e-06,
+      "loss": 0.0439,
+      "step": 27826
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0629447530812898e-06,
+      "loss": 0.0554,
+      "step": 27827
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0624976248228358e-06,
+      "loss": 0.0576,
+      "step": 27828
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0620505853501795e-06,
+      "loss": 0.0482,
+      "step": 27829
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0616036346677595e-06,
+      "loss": 0.0444,
+      "step": 27830
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0611567727800187e-06,
+      "loss": 0.0306,
+      "step": 27831
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0607099996913938e-06,
+      "loss": 0.0465,
+      "step": 27832
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0602633154063258e-06,
+      "loss": 0.0282,
+      "step": 27833
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0598167199292508e-06,
+      "loss": 0.0288,
+      "step": 27834
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0593702132646023e-06,
+      "loss": 0.0468,
+      "step": 27835
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0589237954168196e-06,
+      "loss": 0.0378,
+      "step": 27836
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0584774663903385e-06,
+      "loss": 0.038,
+      "step": 27837
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0580312261895908e-06,
+      "loss": 0.0244,
+      "step": 27838
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0575850748190075e-06,
+      "loss": 0.0402,
+      "step": 27839
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.057139012283025e-06,
+      "loss": 0.0449,
+      "step": 27840
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0566930385860707e-06,
+      "loss": 0.0469,
+      "step": 27841
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.056247153732578e-06,
+      "loss": 0.0587,
+      "step": 27842
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0558013577269754e-06,
+      "loss": 0.038,
+      "step": 27843
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0553556505736894e-06,
+      "loss": 0.0257,
+      "step": 27844
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0549100322771498e-06,
+      "loss": 0.0274,
+      "step": 27845
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0544645028417822e-06,
+      "loss": 0.0482,
+      "step": 27846
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0540190622720137e-06,
+      "loss": 0.0642,
+      "step": 27847
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.05357371057227e-06,
+      "loss": 0.0466,
+      "step": 27848
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0531284477469739e-06,
+      "loss": 0.0345,
+      "step": 27849
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0526832738005477e-06,
+      "loss": 0.0383,
+      "step": 27850
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0522381887374156e-06,
+      "loss": 0.0389,
+      "step": 27851
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0517931925619974e-06,
+      "loss": 0.0588,
+      "step": 27852
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0513482852787172e-06,
+      "loss": 0.051,
+      "step": 27853
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0509034668919916e-06,
+      "loss": 0.0317,
+      "step": 27854
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0504587374062392e-06,
+      "loss": 0.0346,
+      "step": 27855
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0500140968258777e-06,
+      "loss": 0.0306,
+      "step": 27856
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.049569545155329e-06,
+      "loss": 0.0431,
+      "step": 27857
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.049125082399004e-06,
+      "loss": 0.0362,
+      "step": 27858
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0486807085613193e-06,
+      "loss": 0.0408,
+      "step": 27859
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0482364236466913e-06,
+      "loss": 0.0433,
+      "step": 27860
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0477922276595297e-06,
+      "loss": 0.0428,
+      "step": 27861
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0473481206042502e-06,
+      "loss": 0.0401,
+      "step": 27862
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0469041024852677e-06,
+      "loss": 0.058,
+      "step": 27863
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0464601733069856e-06,
+      "loss": 0.0432,
+      "step": 27864
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0460163330738182e-06,
+      "loss": 0.0275,
+      "step": 27865
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0455725817901752e-06,
+      "loss": 0.0677,
+      "step": 27866
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0451289194604642e-06,
+      "loss": 0.0393,
+      "step": 27867
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0446853460890904e-06,
+      "loss": 0.0452,
+      "step": 27868
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.044241861680464e-06,
+      "loss": 0.0249,
+      "step": 27869
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0437984662389865e-06,
+      "loss": 0.0526,
+      "step": 27870
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.043355159769065e-06,
+      "loss": 0.0446,
+      "step": 27871
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0429119422751067e-06,
+      "loss": 0.0412,
+      "step": 27872
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.042468813761508e-06,
+      "loss": 0.048,
+      "step": 27873
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0420257742326722e-06,
+      "loss": 0.05,
+      "step": 27874
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0415828236930059e-06,
+      "loss": 0.0401,
+      "step": 27875
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0411399621469032e-06,
+      "loss": 0.0459,
+      "step": 27876
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0406971895987683e-06,
+      "loss": 0.0504,
+      "step": 27877
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0402545060529967e-06,
+      "loss": 0.0604,
+      "step": 27878
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.039811911513986e-06,
+      "loss": 0.0403,
+      "step": 27879
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0393694059861336e-06,
+      "loss": 0.05,
+      "step": 27880
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0389269894738373e-06,
+      "loss": 0.0361,
+      "step": 27881
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0384846619814904e-06,
+      "loss": 0.043,
+      "step": 27882
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0380424235134844e-06,
+      "loss": 0.0411,
+      "step": 27883
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0376002740742187e-06,
+      "loss": 0.0389,
+      "step": 27884
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0371582136680792e-06,
+      "loss": 0.0647,
+      "step": 27885
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0367162422994602e-06,
+      "loss": 0.0482,
+      "step": 27886
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0362743599727565e-06,
+      "loss": 0.0782,
+      "step": 27887
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0358325666923496e-06,
+      "loss": 0.0409,
+      "step": 27888
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0353908624626318e-06,
+      "loss": 0.0578,
+      "step": 27889
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.034949247287994e-06,
+      "loss": 0.0493,
+      "step": 27890
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0345077211728195e-06,
+      "loss": 0.0512,
+      "step": 27891
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0340662841214966e-06,
+      "loss": 0.0312,
+      "step": 27892
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0336249361384109e-06,
+      "loss": 0.0596,
+      "step": 27893
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0331836772279424e-06,
+      "loss": 0.0405,
+      "step": 27894
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0327425073944785e-06,
+      "loss": 0.0575,
+      "step": 27895
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0323014266424048e-06,
+      "loss": 0.0417,
+      "step": 27896
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0318604349760952e-06,
+      "loss": 0.0513,
+      "step": 27897
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0314195323999354e-06,
+      "loss": 0.0536,
+      "step": 27898
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0309787189183062e-06,
+      "loss": 0.034,
+      "step": 27899
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.030537994535583e-06,
+      "loss": 0.0341,
+      "step": 27900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.030097359256148e-06,
+      "loss": 0.0481,
+      "step": 27901
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0296568130843766e-06,
+      "loss": 0.0302,
+      "step": 27902
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0292163560246439e-06,
+      "loss": 0.0449,
+      "step": 27903
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0287759880813263e-06,
+      "loss": 0.0544,
+      "step": 27904
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.028335709258801e-06,
+      "loss": 0.0521,
+      "step": 27905
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0278955195614394e-06,
+      "loss": 0.059,
+      "step": 27906
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0274554189936126e-06,
+      "loss": 0.045,
+      "step": 27907
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.027015407559696e-06,
+      "loss": 0.0355,
+      "step": 27908
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0265754852640587e-06,
+      "loss": 0.0452,
+      "step": 27909
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0261356521110733e-06,
+      "loss": 0.0506,
+      "step": 27910
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0256959081051066e-06,
+      "loss": 0.0418,
+      "step": 27911
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.025256253250526e-06,
+      "loss": 0.0536,
+      "step": 27912
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0248166875517018e-06,
+      "loss": 0.0456,
+      "step": 27913
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0243772110130012e-06,
+      "loss": 0.0285,
+      "step": 27914
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0239378236387864e-06,
+      "loss": 0.0496,
+      "step": 27915
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0234985254334262e-06,
+      "loss": 0.0505,
+      "step": 27916
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0230593164012836e-06,
+      "loss": 0.0403,
+      "step": 27917
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0226201965467186e-06,
+      "loss": 0.0471,
+      "step": 27918
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0221811658740977e-06,
+      "loss": 0.0406,
+      "step": 27919
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0217422243877795e-06,
+      "loss": 0.0239,
+      "step": 27920
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.021303372092126e-06,
+      "loss": 0.0402,
+      "step": 27921
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0208646089914952e-06,
+      "loss": 0.0477,
+      "step": 27922
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.020425935090248e-06,
+      "loss": 0.0426,
+      "step": 27923
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0199873503927404e-06,
+      "loss": 0.0618,
+      "step": 27924
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0195488549033305e-06,
+      "loss": 0.0416,
+      "step": 27925
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0191104486263748e-06,
+      "loss": 0.0443,
+      "step": 27926
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0186721315662251e-06,
+      "loss": 0.043,
+      "step": 27927
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0182339037272393e-06,
+      "loss": 0.0638,
+      "step": 27928
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0177957651137682e-06,
+      "loss": 0.0428,
+      "step": 27929
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0173577157301685e-06,
+      "loss": 0.0413,
+      "step": 27930
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0169197555807852e-06,
+      "loss": 0.0271,
+      "step": 27931
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0164818846699764e-06,
+      "loss": 0.061,
+      "step": 27932
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.016044103002085e-06,
+      "loss": 0.0369,
+      "step": 27933
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0156064105814655e-06,
+      "loss": 0.0459,
+      "step": 27934
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0151688074124645e-06,
+      "loss": 0.038,
+      "step": 27935
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0147312934994258e-06,
+      "loss": 0.042,
+      "step": 27936
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0142938688467009e-06,
+      "loss": 0.0433,
+      "step": 27937
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0138565334586304e-06,
+      "loss": 0.0322,
+      "step": 27938
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.013419287339561e-06,
+      "loss": 0.049,
+      "step": 27939
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.012982130493838e-06,
+      "loss": 0.0272,
+      "step": 27940
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0125450629258026e-06,
+      "loss": 0.0386,
+      "step": 27941
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0121080846397946e-06,
+      "loss": 0.0601,
+      "step": 27942
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0116711956401593e-06,
+      "loss": 0.0508,
+      "step": 27943
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0112343959312321e-06,
+      "loss": 0.0662,
+      "step": 27944
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0107976855173574e-06,
+      "loss": 0.0575,
+      "step": 27945
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0103610644028693e-06,
+      "loss": 0.032,
+      "step": 27946
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0099245325921058e-06,
+      "loss": 0.0516,
+      "step": 27947
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0094880900894033e-06,
+      "loss": 0.0439,
+      "step": 27948
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0090517368991003e-06,
+      "loss": 0.0416,
+      "step": 27949
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0086154730255305e-06,
+      "loss": 0.0388,
+      "step": 27950
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0081792984730243e-06,
+      "loss": 0.0403,
+      "step": 27951
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0077432132459198e-06,
+      "loss": 0.079,
+      "step": 27952
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0073072173485442e-06,
+      "loss": 0.0345,
+      "step": 27953
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0068713107852336e-06,
+      "loss": 0.0408,
+      "step": 27954
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.006435493560315e-06,
+      "loss": 0.0381,
+      "step": 27955
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0059997656781173e-06,
+      "loss": 0.041,
+      "step": 27956
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0055641271429706e-06,
+      "loss": 0.0412,
+      "step": 27957
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0051285779592046e-06,
+      "loss": 0.0514,
+      "step": 27958
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0046931181311426e-06,
+      "loss": 0.0437,
+      "step": 27959
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.004257747663111e-06,
+      "loss": 0.0416,
+      "step": 27960
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0038224665594375e-06,
+      "loss": 0.0444,
+      "step": 27961
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.003387274824441e-06,
+      "loss": 0.0362,
+      "step": 27962
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0029521724624492e-06,
+      "loss": 0.0456,
+      "step": 27963
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0025171594777872e-06,
+      "loss": 0.0623,
+      "step": 27964
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0020822358747683e-06,
+      "loss": 0.0286,
+      "step": 27965
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.001647401657717e-06,
+      "loss": 0.0464,
+      "step": 27966
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0012126568309567e-06,
+      "loss": 0.0557,
+      "step": 27967
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.000778001398799e-06,
+      "loss": 0.0451,
+      "step": 27968
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0003434353655683e-06,
+      "loss": 0.0403,
+      "step": 27969
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.99908958735578e-07,
+      "loss": 0.0346,
+      "step": 27970
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.99474571513145e-07,
+      "loss": 0.0336,
+      "step": 27971
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.990402737025829e-07,
+      "loss": 0.0508,
+      "step": 27972
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.986060653082096e-07,
+      "loss": 0.0529,
+      "step": 27973
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.981719463343376e-07,
+      "loss": 0.0373,
+      "step": 27974
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.977379167852763e-07,
+      "loss": 0.0351,
+      "step": 27975
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.973039766653414e-07,
+      "loss": 0.0688,
+      "step": 27976
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.968701259788405e-07,
+      "loss": 0.0349,
+      "step": 27977
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.964363647300845e-07,
+      "loss": 0.0414,
+      "step": 27978
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.960026929233867e-07,
+      "loss": 0.0456,
+      "step": 27979
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.955691105630482e-07,
+      "loss": 0.0478,
+      "step": 27980
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.95135617653379e-07,
+      "loss": 0.0494,
+      "step": 27981
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.947022141986872e-07,
+      "loss": 0.0589,
+      "step": 27982
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.94268900203278e-07,
+      "loss": 0.0507,
+      "step": 27983
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.938356756714517e-07,
+      "loss": 0.0511,
+      "step": 27984
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.934025406075176e-07,
+      "loss": 0.051,
+      "step": 27985
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.929694950157754e-07,
+      "loss": 0.046,
+      "step": 27986
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.925365389005258e-07,
+      "loss": 0.0317,
+      "step": 27987
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.921036722660772e-07,
+      "loss": 0.059,
+      "step": 27988
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.916708951167198e-07,
+      "loss": 0.053,
+      "step": 27989
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.91238207456756e-07,
+      "loss": 0.0356,
+      "step": 27990
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.908056092904893e-07,
+      "loss": 0.0391,
+      "step": 27991
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.903731006222106e-07,
+      "loss": 0.039,
+      "step": 27992
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.899406814562208e-07,
+      "loss": 0.0487,
+      "step": 27993
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.89508351796814e-07,
+      "loss": 0.0498,
+      "step": 27994
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.89076111648284e-07,
+      "loss": 0.0502,
+      "step": 27995
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.886439610149246e-07,
+      "loss": 0.0556,
+      "step": 27996
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.882118999010315e-07,
+      "loss": 0.0459,
+      "step": 27997
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.877799283108968e-07,
+      "loss": 0.0411,
+      "step": 27998
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.87348046248807e-07,
+      "loss": 0.0339,
+      "step": 27999
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.86916253719058e-07,
+      "loss": 0.0498,
+      "step": 28000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.864845507259336e-07,
+      "loss": 0.036,
+      "step": 28001
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.860529372737293e-07,
+      "loss": 0.0576,
+      "step": 28002
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.856214133667274e-07,
+      "loss": 0.0477,
+      "step": 28003
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.851899790092156e-07,
+      "loss": 0.044,
+      "step": 28004
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.847586342054804e-07,
+      "loss": 0.0351,
+      "step": 28005
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.843273789598095e-07,
+      "loss": 0.0607,
+      "step": 28006
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.83896213276484e-07,
+      "loss": 0.0552,
+      "step": 28007
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.834651371597848e-07,
+      "loss": 0.0368,
+      "step": 28008
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.830341506139995e-07,
+      "loss": 0.0491,
+      "step": 28009
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.82603253643406e-07,
+      "loss": 0.0367,
+      "step": 28010
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.821724462522874e-07,
+      "loss": 0.045,
+      "step": 28011
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.817417284449217e-07,
+      "loss": 0.0593,
+      "step": 28012
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.81311100225587e-07,
+      "loss": 0.0581,
+      "step": 28013
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.808805615985627e-07,
+      "loss": 0.0426,
+      "step": 28014
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.804501125681243e-07,
+      "loss": 0.0453,
+      "step": 28015
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.80019753138548e-07,
+      "loss": 0.0553,
+      "step": 28016
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.795894833141128e-07,
+      "loss": 0.0242,
+      "step": 28017
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.791593030990899e-07,
+      "loss": 0.0208,
+      "step": 28018
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.7872921249775e-07,
+      "loss": 0.0677,
+      "step": 28019
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.78299211514372e-07,
+      "loss": 0.0462,
+      "step": 28020
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.778693001532213e-07,
+      "loss": 0.0311,
+      "step": 28021
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.774394784185736e-07,
+      "loss": 0.0287,
+      "step": 28022
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.770097463146966e-07,
+      "loss": 0.0453,
+      "step": 28023
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.765801038458577e-07,
+      "loss": 0.04,
+      "step": 28024
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.76150551016326e-07,
+      "loss": 0.0518,
+      "step": 28025
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.757210878303724e-07,
+      "loss": 0.0602,
+      "step": 28026
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.752917142922603e-07,
+      "loss": 0.0423,
+      "step": 28027
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.748624304062525e-07,
+      "loss": 0.0392,
+      "step": 28028
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.744332361766184e-07,
+      "loss": 0.0394,
+      "step": 28029
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.740041316076165e-07,
+      "loss": 0.0259,
+      "step": 28030
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.735751167035134e-07,
+      "loss": 0.0526,
+      "step": 28031
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.731461914685735e-07,
+      "loss": 0.0446,
+      "step": 28032
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.727173559070503e-07,
+      "loss": 0.03,
+      "step": 28033
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.722886100232076e-07,
+      "loss": 0.043,
+      "step": 28034
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.71859953821307e-07,
+      "loss": 0.0393,
+      "step": 28035
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.714313873056048e-07,
+      "loss": 0.043,
+      "step": 28036
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.710029104803565e-07,
+      "loss": 0.0359,
+      "step": 28037
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.705745233498209e-07,
+      "loss": 0.0388,
+      "step": 28038
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.701462259182526e-07,
+      "loss": 0.0663,
+      "step": 28039
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.697180181899058e-07,
+      "loss": 0.0462,
+      "step": 28040
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.692899001690404e-07,
+      "loss": 0.0741,
+      "step": 28041
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.688618718598996e-07,
+      "loss": 0.0579,
+      "step": 28042
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.684339332667403e-07,
+      "loss": 0.044,
+      "step": 28043
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.680060843938154e-07,
+      "loss": 0.0472,
+      "step": 28044
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.675783252453718e-07,
+      "loss": 0.0412,
+      "step": 28045
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.671506558256615e-07,
+      "loss": 0.0486,
+      "step": 28046
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.667230761389323e-07,
+      "loss": 0.0365,
+      "step": 28047
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.662955861894307e-07,
+      "loss": 0.0356,
+      "step": 28048
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.658681859814023e-07,
+      "loss": 0.0422,
+      "step": 28049
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.65440875519098e-07,
+      "loss": 0.0446,
+      "step": 28050
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.65013654806759e-07,
+      "loss": 0.0428,
+      "step": 28051
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.645865238486285e-07,
+      "loss": 0.0453,
+      "step": 28052
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.64159482648952e-07,
+      "loss": 0.0444,
+      "step": 28053
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.637325312119695e-07,
+      "loss": 0.0456,
+      "step": 28054
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.633056695419229e-07,
+      "loss": 0.0336,
+      "step": 28055
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.62878897643058e-07,
+      "loss": 0.0423,
+      "step": 28056
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.624522155196048e-07,
+      "loss": 0.0553,
+      "step": 28057
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.620256231758084e-07,
+      "loss": 0.0316,
+      "step": 28058
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.615991206159058e-07,
+      "loss": 0.0256,
+      "step": 28059
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.611727078441325e-07,
+      "loss": 0.0419,
+      "step": 28060
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.60746384864727e-07,
+      "loss": 0.0322,
+      "step": 28061
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.603201516819215e-07,
+      "loss": 0.0377,
+      "step": 28062
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.598940082999509e-07,
+      "loss": 0.0512,
+      "step": 28063
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.59467954723049e-07,
+      "loss": 0.0326,
+      "step": 28064
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.590419909554516e-07,
+      "loss": 0.0451,
+      "step": 28065
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.58616117001383e-07,
+      "loss": 0.0638,
+      "step": 28066
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.581903328650788e-07,
+      "loss": 0.0264,
+      "step": 28067
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.577646385507688e-07,
+      "loss": 0.0456,
+      "step": 28068
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.573390340626786e-07,
+      "loss": 0.0593,
+      "step": 28069
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.569135194050415e-07,
+      "loss": 0.0315,
+      "step": 28070
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.56488094582081e-07,
+      "loss": 0.0545,
+      "step": 28071
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.560627595980222e-07,
+      "loss": 0.0544,
+      "step": 28072
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.556375144570928e-07,
+      "loss": 0.0461,
+      "step": 28073
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.552123591635176e-07,
+      "loss": 0.0388,
+      "step": 28074
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.547872937215187e-07,
+      "loss": 0.0398,
+      "step": 28075
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.54362318135318e-07,
+      "loss": 0.061,
+      "step": 28076
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.539374324091389e-07,
+      "loss": 0.0456,
+      "step": 28077
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.535126365472003e-07,
+      "loss": 0.04,
+      "step": 28078
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.530879305537266e-07,
+      "loss": 0.0444,
+      "step": 28079
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.526633144329322e-07,
+      "loss": 0.0337,
+      "step": 28080
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.52238788189036e-07,
+      "loss": 0.0199,
+      "step": 28081
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.518143518262557e-07,
+      "loss": 0.0512,
+      "step": 28082
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.51390005348809e-07,
+      "loss": 0.0434,
+      "step": 28083
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.509657487609103e-07,
+      "loss": 0.0389,
+      "step": 28084
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.505415820667751e-07,
+      "loss": 0.0431,
+      "step": 28085
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.501175052706169e-07,
+      "loss": 0.0467,
+      "step": 28086
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.496935183766454e-07,
+      "loss": 0.038,
+      "step": 28087
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.492696213890773e-07,
+      "loss": 0.0369,
+      "step": 28088
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.488458143121193e-07,
+      "loss": 0.0667,
+      "step": 28089
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.484220971499869e-07,
+      "loss": 0.0458,
+      "step": 28090
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.479984699068823e-07,
+      "loss": 0.0299,
+      "step": 28091
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.475749325870209e-07,
+      "loss": 0.05,
+      "step": 28092
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.47151485194604e-07,
+      "loss": 0.0226,
+      "step": 28093
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.467281277338436e-07,
+      "loss": 0.0538,
+      "step": 28094
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.463048602089431e-07,
+      "loss": 0.0485,
+      "step": 28095
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.458816826241046e-07,
+      "loss": 0.0349,
+      "step": 28096
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.454585949835359e-07,
+      "loss": 0.0604,
+      "step": 28097
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.450355972914372e-07,
+      "loss": 0.0473,
+      "step": 28098
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.446126895520124e-07,
+      "loss": 0.0613,
+      "step": 28099
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.44189871769462e-07,
+      "loss": 0.0551,
+      "step": 28100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.437671439479868e-07,
+      "loss": 0.0542,
+      "step": 28101
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.433445060917845e-07,
+      "loss": 0.0315,
+      "step": 28102
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.429219582050575e-07,
+      "loss": 0.0475,
+      "step": 28103
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.424995002920001e-07,
+      "loss": 0.0398,
+      "step": 28104
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.420771323568079e-07,
+      "loss": 0.0401,
+      "step": 28105
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.416548544036808e-07,
+      "loss": 0.0402,
+      "step": 28106
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.412326664368099e-07,
+      "loss": 0.0462,
+      "step": 28107
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.408105684603897e-07,
+      "loss": 0.0347,
+      "step": 28108
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.403885604786178e-07,
+      "loss": 0.0494,
+      "step": 28109
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.399666424956821e-07,
+      "loss": 0.0434,
+      "step": 28110
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.395448145157749e-07,
+      "loss": 0.0457,
+      "step": 28111
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.39123076543087e-07,
+      "loss": 0.0393,
+      "step": 28112
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.387014285818064e-07,
+      "loss": 0.0443,
+      "step": 28113
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.382798706361251e-07,
+      "loss": 0.0517,
+      "step": 28114
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.3785840271023e-07,
+      "loss": 0.049,
+      "step": 28115
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.374370248083042e-07,
+      "loss": 0.0343,
+      "step": 28116
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.370157369345367e-07,
+      "loss": 0.0417,
+      "step": 28117
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.365945390931142e-07,
+      "loss": 0.0594,
+      "step": 28118
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.361734312882187e-07,
+      "loss": 0.0447,
+      "step": 28119
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.357524135240326e-07,
+      "loss": 0.0321,
+      "step": 28120
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.353314858047413e-07,
+      "loss": 0.0475,
+      "step": 28121
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.349106481345227e-07,
+      "loss": 0.0462,
+      "step": 28122
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.344899005175612e-07,
+      "loss": 0.0444,
+      "step": 28123
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.340692429580345e-07,
+      "loss": 0.044,
+      "step": 28124
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.336486754601204e-07,
+      "loss": 0.0301,
+      "step": 28125
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.332281980279977e-07,
+      "loss": 0.0455,
+      "step": 28126
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.328078106658467e-07,
+      "loss": 0.0523,
+      "step": 28127
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.323875133778392e-07,
+      "loss": 0.0486,
+      "step": 28128
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.319673061681511e-07,
+      "loss": 0.039,
+      "step": 28129
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.315471890409599e-07,
+      "loss": 0.0323,
+      "step": 28130
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.311271620004336e-07,
+      "loss": 0.0536,
+      "step": 28131
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.307072250507499e-07,
+      "loss": 0.0351,
+      "step": 28132
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.302873781960808e-07,
+      "loss": 0.0431,
+      "step": 28133
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.29867621440591e-07,
+      "loss": 0.0501,
+      "step": 28134
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.294479547884561e-07,
+      "loss": 0.0392,
+      "step": 28135
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.290283782438436e-07,
+      "loss": 0.0474,
+      "step": 28136
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.286088918109193e-07,
+      "loss": 0.0604,
+      "step": 28137
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.281894954938553e-07,
+      "loss": 0.0436,
+      "step": 28138
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.277701892968138e-07,
+      "loss": 0.0651,
+      "step": 28139
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.273509732239605e-07,
+      "loss": 0.0481,
+      "step": 28140
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.269318472794608e-07,
+      "loss": 0.0405,
+      "step": 28141
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.265128114674804e-07,
+      "loss": 0.0533,
+      "step": 28142
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.260938657921792e-07,
+      "loss": 0.0409,
+      "step": 28143
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.256750102577183e-07,
+      "loss": 0.0331,
+      "step": 28144
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.252562448682623e-07,
+      "loss": 0.0444,
+      "step": 28145
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.248375696279677e-07,
+      "loss": 0.0573,
+      "step": 28146
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.244189845409968e-07,
+      "loss": 0.0537,
+      "step": 28147
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.240004896115051e-07,
+      "loss": 0.0427,
+      "step": 28148
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.235820848436494e-07,
+      "loss": 0.0432,
+      "step": 28149
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.231637702415885e-07,
+      "loss": 0.0326,
+      "step": 28150
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.227455458094791e-07,
+      "loss": 0.0268,
+      "step": 28151
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.223274115514736e-07,
+      "loss": 0.0376,
+      "step": 28152
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.219093674717239e-07,
+      "loss": 0.0414,
+      "step": 28153
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.214914135743879e-07,
+      "loss": 0.0621,
+      "step": 28154
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.210735498636125e-07,
+      "loss": 0.0378,
+      "step": 28155
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.206557763435519e-07,
+      "loss": 0.0406,
+      "step": 28156
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.202380930183585e-07,
+      "loss": 0.0325,
+      "step": 28157
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.198204998921745e-07,
+      "loss": 0.0543,
+      "step": 28158
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.194029969691532e-07,
+      "loss": 0.051,
+      "step": 28159
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.189855842534423e-07,
+      "loss": 0.0274,
+      "step": 28160
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.185682617491865e-07,
+      "loss": 0.0353,
+      "step": 28161
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.181510294605344e-07,
+      "loss": 0.0322,
+      "step": 28162
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.177338873916286e-07,
+      "loss": 0.0395,
+      "step": 28163
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.173168355466111e-07,
+      "loss": 0.0479,
+      "step": 28164
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.168998739296264e-07,
+      "loss": 0.0291,
+      "step": 28165
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.164830025448202e-07,
+      "loss": 0.0484,
+      "step": 28166
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.160662213963311e-07,
+      "loss": 0.0325,
+      "step": 28167
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.156495304882973e-07,
+      "loss": 0.0514,
+      "step": 28168
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.152329298248619e-07,
+      "loss": 0.0298,
+      "step": 28169
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.148164194101595e-07,
+      "loss": 0.0508,
+      "step": 28170
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.14399999248331e-07,
+      "loss": 0.0306,
+      "step": 28171
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.139836693435134e-07,
+      "loss": 0.0399,
+      "step": 28172
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.135674296998387e-07,
+      "loss": 0.0536,
+      "step": 28173
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.131512803214471e-07,
+      "loss": 0.0418,
+      "step": 28174
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.127352212124663e-07,
+      "loss": 0.0548,
+      "step": 28175
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.123192523770364e-07,
+      "loss": 0.056,
+      "step": 28176
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.11903373819284e-07,
+      "loss": 0.0445,
+      "step": 28177
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.114875855433447e-07,
+      "loss": 0.0332,
+      "step": 28178
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.110718875533453e-07,
+      "loss": 0.0644,
+      "step": 28179
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.106562798534191e-07,
+      "loss": 0.0381,
+      "step": 28180
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.102407624476939e-07,
+      "loss": 0.0447,
+      "step": 28181
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.09825335340293e-07,
+      "loss": 0.0461,
+      "step": 28182
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.094099985353488e-07,
+      "loss": 0.0452,
+      "step": 28183
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.089947520369835e-07,
+      "loss": 0.035,
+      "step": 28184
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.085795958493249e-07,
+      "loss": 0.0429,
+      "step": 28185
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.081645299764963e-07,
+      "loss": 0.0425,
+      "step": 28186
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.077495544226222e-07,
+      "loss": 0.0414,
+      "step": 28187
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.073346691918217e-07,
+      "loss": 0.0547,
+      "step": 28188
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.06919874288219e-07,
+      "loss": 0.0728,
+      "step": 28189
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.065051697159332e-07,
+      "loss": 0.0521,
+      "step": 28190
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.060905554790866e-07,
+      "loss": 0.0308,
+      "step": 28191
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.056760315817959e-07,
+      "loss": 0.0515,
+      "step": 28192
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.052615980281776e-07,
+      "loss": 0.0317,
+      "step": 28193
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.048472548223497e-07,
+      "loss": 0.0601,
+      "step": 28194
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.044330019684311e-07,
+      "loss": 0.045,
+      "step": 28195
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.040188394705362e-07,
+      "loss": 0.0427,
+      "step": 28196
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.036047673327753e-07,
+      "loss": 0.046,
+      "step": 28197
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.031907855592659e-07,
+      "loss": 0.0444,
+      "step": 28198
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.027768941541182e-07,
+      "loss": 0.0374,
+      "step": 28199
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.023630931214444e-07,
+      "loss": 0.0516,
+      "step": 28200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.019493824653591e-07,
+      "loss": 0.0511,
+      "step": 28201
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.015357621899657e-07,
+      "loss": 0.0367,
+      "step": 28202
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.011222322993762e-07,
+      "loss": 0.0611,
+      "step": 28203
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.007087927976999e-07,
+      "loss": 0.0418,
+      "step": 28204
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.002954436890421e-07,
+      "loss": 0.0692,
+      "step": 28205
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.998821849775085e-07,
+      "loss": 0.0507,
+      "step": 28206
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.994690166672071e-07,
+      "loss": 0.0498,
+      "step": 28207
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.990559387622388e-07,
+      "loss": 0.0578,
+      "step": 28208
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.986429512667094e-07,
+      "loss": 0.0364,
+      "step": 28209
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.982300541847255e-07,
+      "loss": 0.0435,
+      "step": 28210
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.978172475203795e-07,
+      "loss": 0.0451,
+      "step": 28211
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.974045312777779e-07,
+      "loss": 0.0342,
+      "step": 28212
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.969919054610222e-07,
+      "loss": 0.0624,
+      "step": 28213
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.965793700742075e-07,
+      "loss": 0.036,
+      "step": 28214
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.961669251214356e-07,
+      "loss": 0.0383,
+      "step": 28215
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.957545706068016e-07,
+      "loss": 0.0373,
+      "step": 28216
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.953423065344003e-07,
+      "loss": 0.0317,
+      "step": 28217
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.949301329083293e-07,
+      "loss": 0.0389,
+      "step": 28218
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.945180497326855e-07,
+      "loss": 0.0279,
+      "step": 28219
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.941060570115589e-07,
+      "loss": 0.0398,
+      "step": 28220
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.936941547490419e-07,
+      "loss": 0.0356,
+      "step": 28221
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.932823429492299e-07,
+      "loss": 0.034,
+      "step": 28222
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.928706216162097e-07,
+      "loss": 0.0289,
+      "step": 28223
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.924589907540738e-07,
+      "loss": 0.0405,
+      "step": 28224
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.920474503669141e-07,
+      "loss": 0.0232,
+      "step": 28225
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.91636000458812e-07,
+      "loss": 0.0584,
+      "step": 28226
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.912246410338599e-07,
+      "loss": 0.0554,
+      "step": 28227
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.908133720961443e-07,
+      "loss": 0.0632,
+      "step": 28228
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.904021936497475e-07,
+      "loss": 0.0545,
+      "step": 28229
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.899911056987576e-07,
+      "loss": 0.0378,
+      "step": 28230
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.895801082472577e-07,
+      "loss": 0.0548,
+      "step": 28231
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.891692012993281e-07,
+      "loss": 0.0739,
+      "step": 28232
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.88758384859052e-07,
+      "loss": 0.037,
+      "step": 28233
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.883476589305151e-07,
+      "loss": 0.0334,
+      "step": 28234
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.879370235177909e-07,
+      "loss": 0.037,
+      "step": 28235
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.875264786249605e-07,
+      "loss": 0.0588,
+      "step": 28236
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.87116024256105e-07,
+      "loss": 0.0515,
+      "step": 28237
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.867056604152979e-07,
+      "loss": 0.044,
+      "step": 28238
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.862953871066205e-07,
+      "loss": 0.0496,
+      "step": 28239
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.858852043341459e-07,
+      "loss": 0.0405,
+      "step": 28240
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.854751121019467e-07,
+      "loss": 0.0282,
+      "step": 28241
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.850651104140984e-07,
+      "loss": 0.0597,
+      "step": 28242
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.846551992746777e-07,
+      "loss": 0.0429,
+      "step": 28243
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.842453786877525e-07,
+      "loss": 0.0536,
+      "step": 28244
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.838356486573941e-07,
+      "loss": 0.0489,
+      "step": 28245
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.834260091876756e-07,
+      "loss": 0.0458,
+      "step": 28246
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.830164602826618e-07,
+      "loss": 0.0508,
+      "step": 28247
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.826070019464261e-07,
+      "loss": 0.0413,
+      "step": 28248
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.821976341830341e-07,
+      "loss": 0.045,
+      "step": 28249
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.817883569965502e-07,
+      "loss": 0.0514,
+      "step": 28250
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.813791703910424e-07,
+      "loss": 0.0368,
+      "step": 28251
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.809700743705773e-07,
+      "loss": 0.0572,
+      "step": 28252
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.805610689392163e-07,
+      "loss": 0.0517,
+      "step": 28253
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.801521541010239e-07,
+      "loss": 0.0389,
+      "step": 28254
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.797433298600622e-07,
+      "loss": 0.035,
+      "step": 28255
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.793345962203892e-07,
+      "loss": 0.0458,
+      "step": 28256
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.789259531860706e-07,
+      "loss": 0.0415,
+      "step": 28257
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.785174007611619e-07,
+      "loss": 0.0511,
+      "step": 28258
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.781089389497244e-07,
+      "loss": 0.0418,
+      "step": 28259
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.777005677558126e-07,
+      "loss": 0.0282,
+      "step": 28260
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.772922871834865e-07,
+      "loss": 0.0372,
+      "step": 28261
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.768840972367997e-07,
+      "loss": 0.0595,
+      "step": 28262
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.76475997919809e-07,
+      "loss": 0.0397,
+      "step": 28263
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.760679892365687e-07,
+      "loss": 0.0468,
+      "step": 28264
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.756600711911289e-07,
+      "loss": 0.049,
+      "step": 28265
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.752522437875455e-07,
+      "loss": 0.0411,
+      "step": 28266
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.748445070298661e-07,
+      "loss": 0.0405,
+      "step": 28267
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.744368609221465e-07,
+      "loss": 0.0525,
+      "step": 28268
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.740293054684301e-07,
+      "loss": 0.0294,
+      "step": 28269
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.736218406727715e-07,
+      "loss": 0.0379,
+      "step": 28270
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.732144665392128e-07,
+      "loss": 0.0416,
+      "step": 28271
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.728071830718065e-07,
+      "loss": 0.0527,
+      "step": 28272
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.72399990274595e-07,
+      "loss": 0.033,
+      "step": 28273
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.719928881516226e-07,
+      "loss": 0.043,
+      "step": 28274
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.715858767069374e-07,
+      "loss": 0.0327,
+      "step": 28275
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.711789559445782e-07,
+      "loss": 0.0477,
+      "step": 28276
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.707721258685897e-07,
+      "loss": 0.0512,
+      "step": 28277
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.703653864830142e-07,
+      "loss": 0.0522,
+      "step": 28278
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.699587377918928e-07,
+      "loss": 0.0371,
+      "step": 28279
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.695521797992612e-07,
+      "loss": 0.056,
+      "step": 28280
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.691457125091618e-07,
+      "loss": 0.0552,
+      "step": 28281
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.687393359256302e-07,
+      "loss": 0.0446,
+      "step": 28282
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.683330500527066e-07,
+      "loss": 0.0551,
+      "step": 28283
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.679268548944242e-07,
+      "loss": 0.0511,
+      "step": 28284
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.675207504548178e-07,
+      "loss": 0.0432,
+      "step": 28285
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.671147367379218e-07,
+      "loss": 0.0602,
+      "step": 28286
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.667088137477741e-07,
+      "loss": 0.0482,
+      "step": 28287
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.663029814884028e-07,
+      "loss": 0.0469,
+      "step": 28288
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.658972399638377e-07,
+      "loss": 0.048,
+      "step": 28289
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.654915891781146e-07,
+      "loss": 0.0653,
+      "step": 28290
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.650860291352592e-07,
+      "loss": 0.0359,
+      "step": 28291
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.646805598393037e-07,
+      "loss": 0.0506,
+      "step": 28292
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.642751812942741e-07,
+      "loss": 0.0463,
+      "step": 28293
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.638698935041956e-07,
+      "loss": 0.0363,
+      "step": 28294
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.634646964730975e-07,
+      "loss": 0.0583,
+      "step": 28295
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.630595902050043e-07,
+      "loss": 0.0408,
+      "step": 28296
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.626545747039406e-07,
+      "loss": 0.0435,
+      "step": 28297
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.622496499739275e-07,
+      "loss": 0.0471,
+      "step": 28298
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.618448160189918e-07,
+      "loss": 0.0411,
+      "step": 28299
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.614400728431505e-07,
+      "loss": 0.0352,
+      "step": 28300
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.610354204504257e-07,
+      "loss": 0.0675,
+      "step": 28301
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.60630858844842e-07,
+      "loss": 0.0336,
+      "step": 28302
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.602263880304118e-07,
+      "loss": 0.0351,
+      "step": 28303
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.598220080111542e-07,
+      "loss": 0.058,
+      "step": 28304
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.594177187910913e-07,
+      "loss": 0.0409,
+      "step": 28305
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.590135203742333e-07,
+      "loss": 0.0368,
+      "step": 28306
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.586094127645995e-07,
+      "loss": 0.0432,
+      "step": 28307
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.58205395966204e-07,
+      "loss": 0.0391,
+      "step": 28308
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.578014699830562e-07,
+      "loss": 0.0539,
+      "step": 28309
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.573976348191726e-07,
+      "loss": 0.0319,
+      "step": 28310
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.569938904785668e-07,
+      "loss": 0.0291,
+      "step": 28311
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.565902369652456e-07,
+      "loss": 0.0524,
+      "step": 28312
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.561866742832192e-07,
+      "loss": 0.0405,
+      "step": 28313
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.557832024364998e-07,
+      "loss": 0.0313,
+      "step": 28314
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.55379821429092e-07,
+      "loss": 0.0414,
+      "step": 28315
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.54976531265006e-07,
+      "loss": 0.0447,
+      "step": 28316
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.545733319482464e-07,
+      "loss": 0.0572,
+      "step": 28317
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.541702234828164e-07,
+      "loss": 0.046,
+      "step": 28318
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.537672058727242e-07,
+      "loss": 0.0447,
+      "step": 28319
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.533642791219731e-07,
+      "loss": 0.0471,
+      "step": 28320
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.529614432345645e-07,
+      "loss": 0.0745,
+      "step": 28321
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.525586982144995e-07,
+      "loss": 0.0531,
+      "step": 28322
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.521560440657817e-07,
+      "loss": 0.0325,
+      "step": 28323
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.517534807924078e-07,
+      "loss": 0.0694,
+      "step": 28324
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.513510083983778e-07,
+      "loss": 0.026,
+      "step": 28325
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.509486268876943e-07,
+      "loss": 0.0351,
+      "step": 28326
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.505463362643485e-07,
+      "loss": 0.0438,
+      "step": 28327
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.501441365323382e-07,
+      "loss": 0.0347,
+      "step": 28328
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.497420276956614e-07,
+      "loss": 0.0411,
+      "step": 28329
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.493400097583104e-07,
+      "loss": 0.0459,
+      "step": 28330
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.48938082724281e-07,
+      "loss": 0.0466,
+      "step": 28331
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.485362465975644e-07,
+      "loss": 0.0596,
+      "step": 28332
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.481345013821506e-07,
+      "loss": 0.0414,
+      "step": 28333
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.477328470820345e-07,
+      "loss": 0.0346,
+      "step": 28334
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.473312837012027e-07,
+      "loss": 0.044,
+      "step": 28335
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.469298112436475e-07,
+      "loss": 0.0565,
+      "step": 28336
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.465284297133547e-07,
+      "loss": 0.0339,
+      "step": 28337
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.461271391143134e-07,
+      "loss": 0.0587,
+      "step": 28338
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.45725939450508e-07,
+      "loss": 0.0336,
+      "step": 28339
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.453248307259266e-07,
+      "loss": 0.039,
+      "step": 28340
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.449238129445525e-07,
+      "loss": 0.0489,
+      "step": 28341
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.445228861103682e-07,
+      "loss": 0.0338,
+      "step": 28342
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.441220502273595e-07,
+      "loss": 0.0405,
+      "step": 28343
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.437213052995041e-07,
+      "loss": 0.0339,
+      "step": 28344
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.433206513307878e-07,
+      "loss": 0.0513,
+      "step": 28345
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.429200883251865e-07,
+      "loss": 0.0395,
+      "step": 28346
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.425196162866834e-07,
+      "loss": 0.0368,
+      "step": 28347
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.42119235219253e-07,
+      "loss": 0.037,
+      "step": 28348
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.41718945126877e-07,
+      "loss": 0.0455,
+      "step": 28349
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.413187460135286e-07,
+      "loss": 0.0563,
+      "step": 28350
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.409186378831824e-07,
+      "loss": 0.0306,
+      "step": 28351
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.405186207398175e-07,
+      "loss": 0.0426,
+      "step": 28352
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.401186945874018e-07,
+      "loss": 0.0223,
+      "step": 28353
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.397188594299133e-07,
+      "loss": 0.0383,
+      "step": 28354
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.393191152713231e-07,
+      "loss": 0.0516,
+      "step": 28355
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.389194621156016e-07,
+      "loss": 0.0355,
+      "step": 28356
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.385198999667166e-07,
+      "loss": 0.0405,
+      "step": 28357
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.381204288286415e-07,
+      "loss": 0.0414,
+      "step": 28358
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.377210487053411e-07,
+      "loss": 0.0423,
+      "step": 28359
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.373217596007866e-07,
+      "loss": 0.0362,
+      "step": 28360
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.369225615189424e-07,
+      "loss": 0.0428,
+      "step": 28361
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.365234544637724e-07,
+      "loss": 0.0342,
+      "step": 28362
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.361244384392431e-07,
+      "loss": 0.03,
+      "step": 28363
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.357255134493203e-07,
+      "loss": 0.0617,
+      "step": 28364
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.353266794979653e-07,
+      "loss": 0.0476,
+      "step": 28365
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.349279365891383e-07,
+      "loss": 0.0579,
+      "step": 28366
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.345292847268039e-07,
+      "loss": 0.0281,
+      "step": 28367
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.341307239149177e-07,
+      "loss": 0.037,
+      "step": 28368
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.337322541574433e-07,
+      "loss": 0.0529,
+      "step": 28369
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.333338754583408e-07,
+      "loss": 0.0382,
+      "step": 28370
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.329355878215606e-07,
+      "loss": 0.0503,
+      "step": 28371
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.325373912510637e-07,
+      "loss": 0.0356,
+      "step": 28372
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.321392857508071e-07,
+      "loss": 0.0365,
+      "step": 28373
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.317412713247442e-07,
+      "loss": 0.0341,
+      "step": 28374
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.313433479768262e-07,
+      "loss": 0.0649,
+      "step": 28375
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.309455157110113e-07,
+      "loss": 0.0501,
+      "step": 28376
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.305477745312463e-07,
+      "loss": 0.0428,
+      "step": 28377
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.301501244414856e-07,
+      "loss": 0.0659,
+      "step": 28378
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.297525654456817e-07,
+      "loss": 0.046,
+      "step": 28379
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.293550975477782e-07,
+      "loss": 0.0736,
+      "step": 28380
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.289577207517264e-07,
+      "loss": 0.0567,
+      "step": 28381
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.28560435061475e-07,
+      "loss": 0.0541,
+      "step": 28382
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.281632404809692e-07,
+      "loss": 0.0533,
+      "step": 28383
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.277661370141554e-07,
+      "loss": 0.0502,
+      "step": 28384
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.273691246649795e-07,
+      "loss": 0.0612,
+      "step": 28385
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.269722034373817e-07,
+      "loss": 0.0411,
+      "step": 28386
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.265753733353077e-07,
+      "loss": 0.0284,
+      "step": 28387
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.261786343627021e-07,
+      "loss": 0.0435,
+      "step": 28388
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.257819865235029e-07,
+      "loss": 0.0522,
+      "step": 28389
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.25385429821649e-07,
+      "loss": 0.0398,
+      "step": 28390
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.249889642610831e-07,
+      "loss": 0.0298,
+      "step": 28391
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.245925898457419e-07,
+      "loss": 0.0511,
+      "step": 28392
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.241963065795633e-07,
+      "loss": 0.0465,
+      "step": 28393
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.238001144664876e-07,
+      "loss": 0.0397,
+      "step": 28394
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.234040135104437e-07,
+      "loss": 0.027,
+      "step": 28395
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.230080037153698e-07,
+      "loss": 0.0352,
+      "step": 28396
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.226120850852026e-07,
+      "loss": 0.038,
+      "step": 28397
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.222162576238701e-07,
+      "loss": 0.0374,
+      "step": 28398
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.218205213353092e-07,
+      "loss": 0.035,
+      "step": 28399
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.214248762234501e-07,
+      "loss": 0.042,
+      "step": 28400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.210293222922194e-07,
+      "loss": 0.0729,
+      "step": 28401
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.206338595455498e-07,
+      "loss": 0.0323,
+      "step": 28402
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.202384879873726e-07,
+      "loss": 0.0357,
+      "step": 28403
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.19843207621609e-07,
+      "loss": 0.0468,
+      "step": 28404
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.194480184521891e-07,
+      "loss": 0.0538,
+      "step": 28405
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.190529204830399e-07,
+      "loss": 0.0374,
+      "step": 28406
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.186579137180839e-07,
+      "loss": 0.0516,
+      "step": 28407
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.182629981612477e-07,
+      "loss": 0.0358,
+      "step": 28408
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.178681738164529e-07,
+      "loss": 0.0478,
+      "step": 28409
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.174734406876206e-07,
+      "loss": 0.0303,
+      "step": 28410
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.170787987786732e-07,
+      "loss": 0.0371,
+      "step": 28411
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.166842480935333e-07,
+      "loss": 0.0327,
+      "step": 28412
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.162897886361177e-07,
+      "loss": 0.0426,
+      "step": 28413
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.158954204103431e-07,
+      "loss": 0.048,
+      "step": 28414
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.155011434201332e-07,
+      "loss": 0.0556,
+      "step": 28415
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.151069576693981e-07,
+      "loss": 0.0301,
+      "step": 28416
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.147128631620593e-07,
+      "loss": 0.0442,
+      "step": 28417
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.143188599020291e-07,
+      "loss": 0.0582,
+      "step": 28418
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.139249478932199e-07,
+      "loss": 0.0727,
+      "step": 28419
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.135311271395474e-07,
+      "loss": 0.047,
+      "step": 28420
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.131373976449241e-07,
+      "loss": 0.0571,
+      "step": 28421
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.127437594132592e-07,
+      "loss": 0.0404,
+      "step": 28422
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.123502124484661e-07,
+      "loss": 0.0417,
+      "step": 28423
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.119567567544518e-07,
+      "loss": 0.0637,
+      "step": 28424
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.115633923351252e-07,
+      "loss": 0.0494,
+      "step": 28425
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.111701191943955e-07,
+      "loss": 0.0307,
+      "step": 28426
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.107769373361662e-07,
+      "loss": 0.0323,
+      "step": 28427
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.103838467643477e-07,
+      "loss": 0.0494,
+      "step": 28428
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.099908474828411e-07,
+      "loss": 0.0337,
+      "step": 28429
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.095979394955544e-07,
+      "loss": 0.058,
+      "step": 28430
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.092051228063857e-07,
+      "loss": 0.0501,
+      "step": 28431
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.08812397419243e-07,
+      "loss": 0.0504,
+      "step": 28432
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.084197633380241e-07,
+      "loss": 0.0452,
+      "step": 28433
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.080272205666284e-07,
+      "loss": 0.0453,
+      "step": 28434
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.076347691089592e-07,
+      "loss": 0.043,
+      "step": 28435
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.072424089689112e-07,
+      "loss": 0.0575,
+      "step": 28436
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.068501401503859e-07,
+      "loss": 0.0282,
+      "step": 28437
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.064579626572755e-07,
+      "loss": 0.0357,
+      "step": 28438
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.060658764934815e-07,
+      "loss": 0.0481,
+      "step": 28439
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.05673881662894e-07,
+      "loss": 0.0428,
+      "step": 28440
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.05281978169411e-07,
+      "loss": 0.0525,
+      "step": 28441
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.04890166016924e-07,
+      "loss": 0.0509,
+      "step": 28442
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.04498445209324e-07,
+      "loss": 0.0374,
+      "step": 28443
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.04106815750505e-07,
+      "loss": 0.05,
+      "step": 28444
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.037152776443535e-07,
+      "loss": 0.0433,
+      "step": 28445
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.033238308947621e-07,
+      "loss": 0.0535,
+      "step": 28446
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.029324755056211e-07,
+      "loss": 0.0402,
+      "step": 28447
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.025412114808162e-07,
+      "loss": 0.0386,
+      "step": 28448
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.02150038824232e-07,
+      "loss": 0.0387,
+      "step": 28449
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.017589575397588e-07,
+      "loss": 0.0294,
+      "step": 28450
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.013679676312768e-07,
+      "loss": 0.0337,
+      "step": 28451
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.009770691026753e-07,
+      "loss": 0.0468,
+      "step": 28452
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.005862619578353e-07,
+      "loss": 0.039,
+      "step": 28453
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.001955462006361e-07,
+      "loss": 0.0325,
+      "step": 28454
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.998049218349624e-07,
+      "loss": 0.0466,
+      "step": 28455
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.994143888646955e-07,
+      "loss": 0.0625,
+      "step": 28456
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.990239472937134e-07,
+      "loss": 0.0369,
+      "step": 28457
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.986335971258941e-07,
+      "loss": 0.0426,
+      "step": 28458
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.982433383651166e-07,
+      "loss": 0.0319,
+      "step": 28459
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.97853171015257e-07,
+      "loss": 0.0284,
+      "step": 28460
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.974630950801931e-07,
+      "loss": 0.0397,
+      "step": 28461
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.970731105637974e-07,
+      "loss": 0.0498,
+      "step": 28462
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.966832174699446e-07,
+      "loss": 0.0423,
+      "step": 28463
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.962934158025082e-07,
+      "loss": 0.0334,
+      "step": 28464
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.959037055653617e-07,
+      "loss": 0.03,
+      "step": 28465
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.955140867623756e-07,
+      "loss": 0.0414,
+      "step": 28466
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.9512455939742e-07,
+      "loss": 0.0328,
+      "step": 28467
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.947351234743649e-07,
+      "loss": 0.0571,
+      "step": 28468
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.943457789970776e-07,
+      "loss": 0.0474,
+      "step": 28469
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.939565259694281e-07,
+      "loss": 0.0556,
+      "step": 28470
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.935673643952857e-07,
+      "loss": 0.0405,
+      "step": 28471
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.931782942785083e-07,
+      "loss": 0.0547,
+      "step": 28472
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.927893156229672e-07,
+      "loss": 0.049,
+      "step": 28473
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.924004284325259e-07,
+      "loss": 0.0472,
+      "step": 28474
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.920116327110449e-07,
+      "loss": 0.0414,
+      "step": 28475
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.9162292846239e-07,
+      "loss": 0.027,
+      "step": 28476
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.912343156904212e-07,
+      "loss": 0.0502,
+      "step": 28477
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.908457943989977e-07,
+      "loss": 0.0477,
+      "step": 28478
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.9045736459198e-07,
+      "loss": 0.0286,
+      "step": 28479
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.90069026273228e-07,
+      "loss": 0.0448,
+      "step": 28480
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.896807794465988e-07,
+      "loss": 0.0318,
+      "step": 28481
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.892926241159482e-07,
+      "loss": 0.063,
+      "step": 28482
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.889045602851331e-07,
+      "loss": 0.0451,
+      "step": 28483
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.885165879580081e-07,
+      "loss": 0.0438,
+      "step": 28484
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.88128707138428e-07,
+      "loss": 0.0391,
+      "step": 28485
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.877409178302464e-07,
+      "loss": 0.0484,
+      "step": 28486
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.873532200373135e-07,
+      "loss": 0.0491,
+      "step": 28487
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.869656137634807e-07,
+      "loss": 0.0356,
+      "step": 28488
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.865780990126027e-07,
+      "loss": 0.0516,
+      "step": 28489
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.861906757885263e-07,
+      "loss": 0.0521,
+      "step": 28490
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.858033440950985e-07,
+      "loss": 0.0367,
+      "step": 28491
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.854161039361696e-07,
+      "loss": 0.0316,
+      "step": 28492
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.850289553155854e-07,
+      "loss": 0.0396,
+      "step": 28493
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.846418982371929e-07,
+      "loss": 0.0429,
+      "step": 28494
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.842549327048366e-07,
+      "loss": 0.0469,
+      "step": 28495
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.838680587223579e-07,
+      "loss": 0.0429,
+      "step": 28496
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.834812762936028e-07,
+      "loss": 0.0549,
+      "step": 28497
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.830945854224159e-07,
+      "loss": 0.0515,
+      "step": 28498
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.827079861126341e-07,
+      "loss": 0.0383,
+      "step": 28499
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.82321478368101e-07,
+      "loss": 0.0519,
+      "step": 28500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.819350621926559e-07,
+      "loss": 0.0521,
+      "step": 28501
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.815487375901342e-07,
+      "loss": 0.0365,
+      "step": 28502
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.811625045643778e-07,
+      "loss": 0.0502,
+      "step": 28503
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.807763631192211e-07,
+      "loss": 0.0322,
+      "step": 28504
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.803903132585023e-07,
+      "loss": 0.0518,
+      "step": 28505
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.800043549860525e-07,
+      "loss": 0.0508,
+      "step": 28506
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.796184883057112e-07,
+      "loss": 0.0453,
+      "step": 28507
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.792327132213063e-07,
+      "loss": 0.0402,
+      "step": 28508
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.788470297366746e-07,
+      "loss": 0.0474,
+      "step": 28509
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.784614378556454e-07,
+      "loss": 0.0441,
+      "step": 28510
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.780759375820479e-07,
+      "loss": 0.0375,
+      "step": 28511
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.776905289197145e-07,
+      "loss": 0.0412,
+      "step": 28512
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.7730521187247e-07,
+      "loss": 0.051,
+      "step": 28513
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.769199864441479e-07,
+      "loss": 0.0329,
+      "step": 28514
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.765348526385685e-07,
+      "loss": 0.0563,
+      "step": 28515
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.761498104595633e-07,
+      "loss": 0.0402,
+      "step": 28516
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.757648599109535e-07,
+      "loss": 0.0782,
+      "step": 28517
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.753800009965673e-07,
+      "loss": 0.0501,
+      "step": 28518
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.749952337202249e-07,
+      "loss": 0.0483,
+      "step": 28519
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.746105580857466e-07,
+      "loss": 0.029,
+      "step": 28520
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.742259740969594e-07,
+      "loss": 0.0438,
+      "step": 28521
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.738414817576778e-07,
+      "loss": 0.0539,
+      "step": 28522
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.734570810717246e-07,
+      "loss": 0.0496,
+      "step": 28523
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.730727720429199e-07,
+      "loss": 0.0428,
+      "step": 28524
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.726885546750796e-07,
+      "loss": 0.0263,
+      "step": 28525
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.723044289720194e-07,
+      "loss": 0.0462,
+      "step": 28526
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.719203949375575e-07,
+      "loss": 0.0395,
+      "step": 28527
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.715364525755065e-07,
+      "loss": 0.0462,
+      "step": 28528
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.711526018896843e-07,
+      "loss": 0.0504,
+      "step": 28529
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.707688428839011e-07,
+      "loss": 0.0572,
+      "step": 28530
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.703851755619684e-07,
+      "loss": 0.0456,
+      "step": 28531
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.700015999276977e-07,
+      "loss": 0.037,
+      "step": 28532
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.696181159849036e-07,
+      "loss": 0.0415,
+      "step": 28533
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.692347237373932e-07,
+      "loss": 0.0518,
+      "step": 28534
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.688514231889721e-07,
+      "loss": 0.0349,
+      "step": 28535
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.684682143434529e-07,
+      "loss": 0.0497,
+      "step": 28536
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.680850972046383e-07,
+      "loss": 0.037,
+      "step": 28537
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.677020717763362e-07,
+      "loss": 0.0299,
+      "step": 28538
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.673191380623546e-07,
+      "loss": 0.0384,
+      "step": 28539
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.669362960664916e-07,
+      "loss": 0.0407,
+      "step": 28540
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.665535457925532e-07,
+      "loss": 0.0665,
+      "step": 28541
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.66170887244343e-07,
+      "loss": 0.0347,
+      "step": 28542
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.657883204256622e-07,
+      "loss": 0.0368,
+      "step": 28543
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.654058453403079e-07,
+      "loss": 0.0414,
+      "step": 28544
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.650234619920826e-07,
+      "loss": 0.0483,
+      "step": 28545
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.646411703847834e-07,
+      "loss": 0.0596,
+      "step": 28546
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.642589705222092e-07,
+      "loss": 0.0338,
+      "step": 28547
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.638768624081594e-07,
+      "loss": 0.0332,
+      "step": 28548
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.634948460464242e-07,
+      "loss": 0.0273,
+      "step": 28549
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.631129214408006e-07,
+      "loss": 0.0502,
+      "step": 28550
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.627310885950845e-07,
+      "loss": 0.0307,
+      "step": 28551
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.623493475130672e-07,
+      "loss": 0.0517,
+      "step": 28552
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.619676981985436e-07,
+      "loss": 0.0501,
+      "step": 28553
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.615861406553016e-07,
+      "loss": 0.0481,
+      "step": 28554
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.0468,
+      "step": 28555
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.608233008978272e-07,
+      "loss": 0.058,
+      "step": 28556
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.604420186911742e-07,
+      "loss": 0.0716,
+      "step": 28557
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.60060828270961e-07,
+      "loss": 0.0516,
+      "step": 28558
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.596797296409719e-07,
+      "loss": 0.0478,
+      "step": 28559
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.592987228049964e-07,
+      "loss": 0.0286,
+      "step": 28560
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.589178077668158e-07,
+      "loss": 0.0549,
+      "step": 28561
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.58536984530216e-07,
+      "loss": 0.061,
+      "step": 28562
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.58156253098985e-07,
+      "loss": 0.0445,
+      "step": 28563
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.577756134768955e-07,
+      "loss": 0.0344,
+      "step": 28564
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.573950656677342e-07,
+      "loss": 0.0432,
+      "step": 28565
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.570146096752828e-07,
+      "loss": 0.0322,
+      "step": 28566
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.56634245503316e-07,
+      "loss": 0.0642,
+      "step": 28567
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.562539731556185e-07,
+      "loss": 0.0545,
+      "step": 28568
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.558737926359638e-07,
+      "loss": 0.0408,
+      "step": 28569
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.554937039481281e-07,
+      "loss": 0.053,
+      "step": 28570
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.551137070958881e-07,
+      "loss": 0.0366,
+      "step": 28571
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.547338020830231e-07,
+      "loss": 0.0466,
+      "step": 28572
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.543539889133e-07,
+      "loss": 0.0425,
+      "step": 28573
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.539742675904949e-07,
+      "loss": 0.0494,
+      "step": 28574
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.535946381183812e-07,
+      "loss": 0.0437,
+      "step": 28575
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.532151005007282e-07,
+      "loss": 0.0423,
+      "step": 28576
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.528356547413095e-07,
+      "loss": 0.0236,
+      "step": 28577
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.524563008438923e-07,
+      "loss": 0.0513,
+      "step": 28578
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.520770388122422e-07,
+      "loss": 0.0482,
+      "step": 28579
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.516978686501309e-07,
+      "loss": 0.0641,
+      "step": 28580
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.513187903613251e-07,
+      "loss": 0.0438,
+      "step": 28581
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.509398039495896e-07,
+      "loss": 0.0501,
+      "step": 28582
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.505609094186871e-07,
+      "loss": 0.0274,
+      "step": 28583
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.501821067723858e-07,
+      "loss": 0.0394,
+      "step": 28584
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.498033960144435e-07,
+      "loss": 0.0517,
+      "step": 28585
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.494247771486285e-07,
+      "loss": 0.0642,
+      "step": 28586
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.490462501786988e-07,
+      "loss": 0.0471,
+      "step": 28587
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.486678151084125e-07,
+      "loss": 0.0545,
+      "step": 28588
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.482894719415313e-07,
+      "loss": 0.0426,
+      "step": 28589
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.479112206818162e-07,
+      "loss": 0.0387,
+      "step": 28590
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.4753306133302e-07,
+      "loss": 0.0437,
+      "step": 28591
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.471549938989032e-07,
+      "loss": 0.0456,
+      "step": 28592
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.467770183832201e-07,
+      "loss": 0.0364,
+      "step": 28593
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.463991347897238e-07,
+      "loss": 0.0378,
+      "step": 28594
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.460213431221719e-07,
+      "loss": 0.0406,
+      "step": 28595
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.456436433843129e-07,
+      "loss": 0.0407,
+      "step": 28596
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.452660355799036e-07,
+      "loss": 0.0475,
+      "step": 28597
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.448885197126909e-07,
+      "loss": 0.0406,
+      "step": 28598
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.445110957864287e-07,
+      "loss": 0.0469,
+      "step": 28599
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.44133763804864e-07,
+      "loss": 0.0353,
+      "step": 28600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.43756523771747e-07,
+      "loss": 0.0431,
+      "step": 28601
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.433793756908236e-07,
+      "loss": 0.0529,
+      "step": 28602
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.43002319565841e-07,
+      "loss": 0.042,
+      "step": 28603
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.426253554005447e-07,
+      "loss": 0.043,
+      "step": 28604
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.422484831986798e-07,
+      "loss": 0.0312,
+      "step": 28605
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.41871702963991e-07,
+      "loss": 0.0365,
+      "step": 28606
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.414950147002187e-07,
+      "loss": 0.0509,
+      "step": 28607
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.411184184111087e-07,
+      "loss": 0.0346,
+      "step": 28608
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.407419141003979e-07,
+      "loss": 0.0498,
+      "step": 28609
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.403655017718303e-07,
+      "loss": 0.0323,
+      "step": 28610
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.399891814291438e-07,
+      "loss": 0.0341,
+      "step": 28611
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.396129530760743e-07,
+      "loss": 0.0495,
+      "step": 28612
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.392368167163643e-07,
+      "loss": 0.0351,
+      "step": 28613
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.388607723537456e-07,
+      "loss": 0.0446,
+      "step": 28614
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.384848199919558e-07,
+      "loss": 0.0336,
+      "step": 28615
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.381089596347313e-07,
+      "loss": 0.0494,
+      "step": 28616
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.377331912858043e-07,
+      "loss": 0.0342,
+      "step": 28617
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.373575149489065e-07,
+      "loss": 0.0355,
+      "step": 28618
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.369819306277725e-07,
+      "loss": 0.0516,
+      "step": 28619
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.366064383261307e-07,
+      "loss": 0.0333,
+      "step": 28620
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.362310380477133e-07,
+      "loss": 0.0456,
+      "step": 28621
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.358557297962499e-07,
+      "loss": 0.0516,
+      "step": 28622
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.354805135754662e-07,
+      "loss": 0.0549,
+      "step": 28623
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.351053893890902e-07,
+      "loss": 0.0374,
+      "step": 28624
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.347303572408515e-07,
+      "loss": 0.0565,
+      "step": 28625
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.343554171344735e-07,
+      "loss": 0.0383,
+      "step": 28626
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.339805690736801e-07,
+      "loss": 0.0383,
+      "step": 28627
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.336058130621971e-07,
+      "loss": 0.0323,
+      "step": 28628
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.332311491037447e-07,
+      "loss": 0.0349,
+      "step": 28629
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.32856577202048e-07,
+      "loss": 0.0346,
+      "step": 28630
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.32482097360826e-07,
+      "loss": 0.0388,
+      "step": 28631
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.32107709583798e-07,
+      "loss": 0.0401,
+      "step": 28632
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.317334138746845e-07,
+      "loss": 0.0503,
+      "step": 28633
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.313592102372058e-07,
+      "loss": 0.0407,
+      "step": 28634
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.309850986750766e-07,
+      "loss": 0.0362,
+      "step": 28635
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.306110791920118e-07,
+      "loss": 0.0322,
+      "step": 28636
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.302371517917306e-07,
+      "loss": 0.0625,
+      "step": 28637
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.298633164779457e-07,
+      "loss": 0.0508,
+      "step": 28638
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.294895732543694e-07,
+      "loss": 0.0312,
+      "step": 28639
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.291159221247212e-07,
+      "loss": 0.0426,
+      "step": 28640
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.287423630927038e-07,
+      "loss": 0.0408,
+      "step": 28641
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.283688961620328e-07,
+      "loss": 0.0628,
+      "step": 28642
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.279955213364187e-07,
+      "loss": 0.0602,
+      "step": 28643
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.276222386195686e-07,
+      "loss": 0.0302,
+      "step": 28644
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.272490480151939e-07,
+      "loss": 0.0473,
+      "step": 28645
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.268759495269994e-07,
+      "loss": 0.056,
+      "step": 28646
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.265029431586912e-07,
+      "loss": 0.0261,
+      "step": 28647
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.26130028913975e-07,
+      "loss": 0.0359,
+      "step": 28648
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.257572067965569e-07,
+      "loss": 0.0444,
+      "step": 28649
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.253844768101403e-07,
+      "loss": 0.0488,
+      "step": 28650
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.250118389584259e-07,
+      "loss": 0.0544,
+      "step": 28651
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.246392932451174e-07,
+      "loss": 0.0312,
+      "step": 28652
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.242668396739139e-07,
+      "loss": 0.0475,
+      "step": 28653
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.238944782485191e-07,
+      "loss": 0.0449,
+      "step": 28654
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.23522208972628e-07,
+      "loss": 0.0554,
+      "step": 28655
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.231500318499396e-07,
+      "loss": 0.0304,
+      "step": 28656
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.227779468841523e-07,
+      "loss": 0.0432,
+      "step": 28657
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.224059540789618e-07,
+      "loss": 0.0457,
+      "step": 28658
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.220340534380654e-07,
+      "loss": 0.0409,
+      "step": 28659
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.216622449651533e-07,
+      "loss": 0.0329,
+      "step": 28660
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.212905286639238e-07,
+      "loss": 0.0426,
+      "step": 28661
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.209189045380649e-07,
+      "loss": 0.0475,
+      "step": 28662
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.205473725912726e-07,
+      "loss": 0.0389,
+      "step": 28663
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.201759328272373e-07,
+      "loss": 0.0396,
+      "step": 28664
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.198045852496438e-07,
+      "loss": 0.0568,
+      "step": 28665
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.194333298621869e-07,
+      "loss": 0.0362,
+      "step": 28666
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.190621666685526e-07,
+      "loss": 0.0318,
+      "step": 28667
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.186910956724269e-07,
+      "loss": 0.0289,
+      "step": 28668
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.183201168774989e-07,
+      "loss": 0.0649,
+      "step": 28669
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.179492302874525e-07,
+      "loss": 0.0356,
+      "step": 28670
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.175784359059701e-07,
+      "loss": 0.0437,
+      "step": 28671
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.172077337367389e-07,
+      "loss": 0.0604,
+      "step": 28672
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.16837123783437e-07,
+      "loss": 0.0499,
+      "step": 28673
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.164666060497516e-07,
+      "loss": 0.0291,
+      "step": 28674
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.160961805393574e-07,
+      "loss": 0.0372,
+      "step": 28675
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.157258472559403e-07,
+      "loss": 0.0297,
+      "step": 28676
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.15355606203173e-07,
+      "loss": 0.0383,
+      "step": 28677
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.149854573847392e-07,
+      "loss": 0.0306,
+      "step": 28678
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.146154008043138e-07,
+      "loss": 0.0486,
+      "step": 28679
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.142454364655704e-07,
+      "loss": 0.0369,
+      "step": 28680
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.138755643721884e-07,
+      "loss": 0.032,
+      "step": 28681
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.135057845278392e-07,
+      "loss": 0.0569,
+      "step": 28682
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.131360969361989e-07,
+      "loss": 0.0491,
+      "step": 28683
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.127665016009355e-07,
+      "loss": 0.0478,
+      "step": 28684
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.123969985257262e-07,
+      "loss": 0.0436,
+      "step": 28685
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.120275877142369e-07,
+      "loss": 0.0272,
+      "step": 28686
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.116582691701412e-07,
+      "loss": 0.0378,
+      "step": 28687
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.112890428971053e-07,
+      "loss": 0.0582,
+      "step": 28688
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.109199088987973e-07,
+      "loss": 0.0344,
+      "step": 28689
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.105508671788864e-07,
+      "loss": 0.0382,
+      "step": 28690
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.101819177410352e-07,
+      "loss": 0.0309,
+      "step": 28691
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.098130605889109e-07,
+      "loss": 0.0551,
+      "step": 28692
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.094442957261793e-07,
+      "loss": 0.0527,
+      "step": 28693
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.090756231565021e-07,
+      "loss": 0.0438,
+      "step": 28694
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.087070428835397e-07,
+      "loss": 0.033,
+      "step": 28695
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.083385549109578e-07,
+      "loss": 0.0671,
+      "step": 28696
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.079701592424127e-07,
+      "loss": 0.0493,
+      "step": 28697
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.076018558815678e-07,
+      "loss": 0.0361,
+      "step": 28698
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.072336448320816e-07,
+      "loss": 0.0389,
+      "step": 28699
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.068655260976076e-07,
+      "loss": 0.0435,
+      "step": 28700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.064974996818053e-07,
+      "loss": 0.0579,
+      "step": 28701
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.061295655883338e-07,
+      "loss": 0.0286,
+      "step": 28702
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.057617238208447e-07,
+      "loss": 0.0351,
+      "step": 28703
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.053939743829919e-07,
+      "loss": 0.0308,
+      "step": 28704
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.05026317278431e-07,
+      "loss": 0.0382,
+      "step": 28705
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.046587525108118e-07,
+      "loss": 0.0456,
+      "step": 28706
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.042912800837865e-07,
+      "loss": 0.0401,
+      "step": 28707
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.03923900001009e-07,
+      "loss": 0.0298,
+      "step": 28708
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.03556612266123e-07,
+      "loss": 0.0377,
+      "step": 28709
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.031894168827802e-07,
+      "loss": 0.0395,
+      "step": 28710
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.028223138546297e-07,
+      "loss": 0.0414,
+      "step": 28711
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.024553031853165e-07,
+      "loss": 0.0338,
+      "step": 28712
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.020883848784854e-07,
+      "loss": 0.0517,
+      "step": 28713
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.017215589377846e-07,
+      "loss": 0.0453,
+      "step": 28714
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.013548253668556e-07,
+      "loss": 0.047,
+      "step": 28715
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.009881841693411e-07,
+      "loss": 0.0553,
+      "step": 28716
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.006216353488881e-07,
+      "loss": 0.0398,
+      "step": 28717
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.002551789091316e-07,
+      "loss": 0.0332,
+      "step": 28718
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.998888148537153e-07,
+      "loss": 0.0564,
+      "step": 28719
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.995225431862796e-07,
+      "loss": 0.0471,
+      "step": 28720
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.991563639104593e-07,
+      "loss": 0.0501,
+      "step": 28721
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.987902770298971e-07,
+      "loss": 0.0643,
+      "step": 28722
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.984242825482257e-07,
+      "loss": 0.0439,
+      "step": 28723
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.980583804690821e-07,
+      "loss": 0.0377,
+      "step": 28724
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.976925707961002e-07,
+      "loss": 0.045,
+      "step": 28725
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.973268535329181e-07,
+      "loss": 0.0363,
+      "step": 28726
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.96961228683165e-07,
+      "loss": 0.0473,
+      "step": 28727
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.965956962504727e-07,
+      "loss": 0.0337,
+      "step": 28728
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.962302562384759e-07,
+      "loss": 0.0502,
+      "step": 28729
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.958649086507996e-07,
+      "loss": 0.0468,
+      "step": 28730
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.954996534910774e-07,
+      "loss": 0.0502,
+      "step": 28731
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.951344907629398e-07,
+      "loss": 0.0336,
+      "step": 28732
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.947694204700083e-07,
+      "loss": 0.0483,
+      "step": 28733
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.944044426159125e-07,
+      "loss": 0.0367,
+      "step": 28734
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.940395572042791e-07,
+      "loss": 0.057,
+      "step": 28735
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.936747642387309e-07,
+      "loss": 0.0451,
+      "step": 28736
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.93310063722894e-07,
+      "loss": 0.0411,
+      "step": 28737
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.929454556603899e-07,
+      "loss": 0.0404,
+      "step": 28738
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.925809400548389e-07,
+      "loss": 0.042,
+      "step": 28739
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.922165169098649e-07,
+      "loss": 0.0455,
+      "step": 28740
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.918521862290905e-07,
+      "loss": 0.0433,
+      "step": 28741
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.914879480161285e-07,
+      "loss": 0.0744,
+      "step": 28742
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.911238022746003e-07,
+      "loss": 0.0727,
+      "step": 28743
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.907597490081253e-07,
+      "loss": 0.0431,
+      "step": 28744
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.903957882203171e-07,
+      "loss": 0.0496,
+      "step": 28745
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.900319199147932e-07,
+      "loss": 0.0406,
+      "step": 28746
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.896681440951691e-07,
+      "loss": 0.0716,
+      "step": 28747
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.893044607650545e-07,
+      "loss": 0.0344,
+      "step": 28748
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.889408699280653e-07,
+      "loss": 0.0576,
+      "step": 28749
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.885773715878141e-07,
+      "loss": 0.0352,
+      "step": 28750
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.882139657479114e-07,
+      "loss": 0.0437,
+      "step": 28751
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.878506524119644e-07,
+      "loss": 0.0572,
+      "step": 28752
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.874874315835867e-07,
+      "loss": 0.0543,
+      "step": 28753
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.871243032663832e-07,
+      "loss": 0.0403,
+      "step": 28754
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.867612674639634e-07,
+      "loss": 0.0456,
+      "step": 28755
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.863983241799333e-07,
+      "loss": 0.0324,
+      "step": 28756
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.860354734178964e-07,
+      "loss": 0.0405,
+      "step": 28757
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.856727151814591e-07,
+      "loss": 0.0466,
+      "step": 28758
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.85310049474226e-07,
+      "loss": 0.0551,
+      "step": 28759
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.849474762997966e-07,
+      "loss": 0.044,
+      "step": 28760
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.845849956617768e-07,
+      "loss": 0.0601,
+      "step": 28761
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.842226075637659e-07,
+      "loss": 0.0407,
+      "step": 28762
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.838603120093613e-07,
+      "loss": 0.0504,
+      "step": 28763
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.834981090021664e-07,
+      "loss": 0.0313,
+      "step": 28764
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.831359985457753e-07,
+      "loss": 0.0426,
+      "step": 28765
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.827739806437883e-07,
+      "loss": 0.0477,
+      "step": 28766
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.824120552997993e-07,
+      "loss": 0.0532,
+      "step": 28767
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.820502225174064e-07,
+      "loss": 0.0466,
+      "step": 28768
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.816884823002012e-07,
+      "loss": 0.0529,
+      "step": 28769
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.813268346517809e-07,
+      "loss": 0.0439,
+      "step": 28770
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.809652795757349e-07,
+      "loss": 0.0361,
+      "step": 28771
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.806038170756546e-07,
+      "loss": 0.0414,
+      "step": 28772
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.802424471551339e-07,
+      "loss": 0.0355,
+      "step": 28773
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.798811698177588e-07,
+      "loss": 0.0401,
+      "step": 28774
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.79519985067123e-07,
+      "loss": 0.0427,
+      "step": 28775
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.791588929068094e-07,
+      "loss": 0.0436,
+      "step": 28776
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.787978933404094e-07,
+      "loss": 0.0366,
+      "step": 28777
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.784369863715056e-07,
+      "loss": 0.0239,
+      "step": 28778
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.780761720036866e-07,
+      "loss": 0.034,
+      "step": 28779
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.777154502405347e-07,
+      "loss": 0.0291,
+      "step": 28780
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.77354821085634e-07,
+      "loss": 0.0415,
+      "step": 28781
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.76994284542567e-07,
+      "loss": 0.0247,
+      "step": 28782
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.766338406149142e-07,
+      "loss": 0.0587,
+      "step": 28783
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.762734893062572e-07,
+      "loss": 0.043,
+      "step": 28784
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.759132306201788e-07,
+      "loss": 0.0409,
+      "step": 28785
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.755530645602514e-07,
+      "loss": 0.0508,
+      "step": 28786
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.751929911300559e-07,
+      "loss": 0.0403,
+      "step": 28787
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.748330103331723e-07,
+      "loss": 0.0558,
+      "step": 28788
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.744731221731726e-07,
+      "loss": 0.077,
+      "step": 28789
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.741133266536349e-07,
+      "loss": 0.0558,
+      "step": 28790
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.737536237781317e-07,
+      "loss": 0.0319,
+      "step": 28791
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.733940135502348e-07,
+      "loss": 0.0467,
+      "step": 28792
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.730344959735191e-07,
+      "loss": 0.0412,
+      "step": 28793
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.726750710515562e-07,
+      "loss": 0.0484,
+      "step": 28794
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.723157387879164e-07,
+      "loss": 0.0475,
+      "step": 28795
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.719564991861671e-07,
+      "loss": 0.0335,
+      "step": 28796
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.715973522498798e-07,
+      "loss": 0.052,
+      "step": 28797
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.712382979826204e-07,
+      "loss": 0.0472,
+      "step": 28798
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.708793363879573e-07,
+      "loss": 0.0611,
+      "step": 28799
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.705204674694566e-07,
+      "loss": 0.0387,
+      "step": 28800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.701616912306807e-07,
+      "loss": 0.0393,
+      "step": 28801
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.698030076751949e-07,
+      "loss": 0.0425,
+      "step": 28802
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.694444168065651e-07,
+      "loss": 0.035,
+      "step": 28803
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.690859186283505e-07,
+      "loss": 0.0449,
+      "step": 28804
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.687275131441129e-07,
+      "loss": 0.0294,
+      "step": 28805
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.683692003574138e-07,
+      "loss": 0.0536,
+      "step": 28806
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.680109802718115e-07,
+      "loss": 0.0721,
+      "step": 28807
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.676528528908643e-07,
+      "loss": 0.0409,
+      "step": 28808
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.672948182181338e-07,
+      "loss": 0.0307,
+      "step": 28809
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.669368762571704e-07,
+      "loss": 0.044,
+      "step": 28810
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.665790270115346e-07,
+      "loss": 0.0438,
+      "step": 28811
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.662212704847804e-07,
+      "loss": 0.0582,
+      "step": 28812
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.658636066804591e-07,
+      "loss": 0.0377,
+      "step": 28813
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.655060356021281e-07,
+      "loss": 0.0535,
+      "step": 28814
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.651485572533379e-07,
+      "loss": 0.0536,
+      "step": 28815
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.647911716376365e-07,
+      "loss": 0.0303,
+      "step": 28816
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.64433878758578e-07,
+      "loss": 0.0438,
+      "step": 28817
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.640766786197117e-07,
+      "loss": 0.0353,
+      "step": 28818
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.637195712245848e-07,
+      "loss": 0.0265,
+      "step": 28819
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.633625565767443e-07,
+      "loss": 0.0374,
+      "step": 28820
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.630056346797387e-07,
+      "loss": 0.0237,
+      "step": 28821
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.626488055371105e-07,
+      "loss": 0.0386,
+      "step": 28822
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.622920691524081e-07,
+      "loss": 0.0489,
+      "step": 28823
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.619354255291743e-07,
+      "loss": 0.0256,
+      "step": 28824
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.615788746709495e-07,
+      "loss": 0.0371,
+      "step": 28825
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.612224165812775e-07,
+      "loss": 0.0437,
+      "step": 28826
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.60866051263701e-07,
+      "loss": 0.0525,
+      "step": 28827
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.605097787217584e-07,
+      "loss": 0.0387,
+      "step": 28828
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.601535989589891e-07,
+      "loss": 0.0675,
+      "step": 28829
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.597975119789324e-07,
+      "loss": 0.0319,
+      "step": 28830
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.594415177851232e-07,
+      "loss": 0.0334,
+      "step": 28831
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.590856163811011e-07,
+      "loss": 0.0477,
+      "step": 28832
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.587298077703996e-07,
+      "loss": 0.039,
+      "step": 28833
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.583740919565529e-07,
+      "loss": 0.0548,
+      "step": 28834
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.580184689430958e-07,
+      "loss": 0.0463,
+      "step": 28835
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.576629387335631e-07,
+      "loss": 0.0334,
+      "step": 28836
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.573075013314823e-07,
+      "loss": 0.0439,
+      "step": 28837
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.56952156740388e-07,
+      "loss": 0.029,
+      "step": 28838
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.565969049638099e-07,
+      "loss": 0.0243,
+      "step": 28839
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.562417460052728e-07,
+      "loss": 0.0425,
+      "step": 28840
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.558866798683105e-07,
+      "loss": 0.0645,
+      "step": 28841
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.55531706556447e-07,
+      "loss": 0.0483,
+      "step": 28842
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.551768260732095e-07,
+      "loss": 0.0441,
+      "step": 28843
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.548220384221226e-07,
+      "loss": 0.0604,
+      "step": 28844
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.54467343606714e-07,
+      "loss": 0.035,
+      "step": 28845
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.541127416305015e-07,
+      "loss": 0.0668,
+      "step": 28846
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.537582324970137e-07,
+      "loss": 0.0429,
+      "step": 28847
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.5340381620977e-07,
+      "loss": 0.0554,
+      "step": 28848
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.530494927722885e-07,
+      "loss": 0.0446,
+      "step": 28849
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.526952621880933e-07,
+      "loss": 0.0309,
+      "step": 28850
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.523411244607004e-07,
+      "loss": 0.0521,
+      "step": 28851
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.5198707959363e-07,
+      "loss": 0.0515,
+      "step": 28852
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.516331275903964e-07,
+      "loss": 0.0524,
+      "step": 28853
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.5127926845452e-07,
+      "loss": 0.036,
+      "step": 28854
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.509255021895111e-07,
+      "loss": 0.0596,
+      "step": 28855
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.505718287988883e-07,
+      "loss": 0.0324,
+      "step": 28856
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.502182482861641e-07,
+      "loss": 0.0489,
+      "step": 28857
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.49864760654848e-07,
+      "loss": 0.0355,
+      "step": 28858
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.495113659084551e-07,
+      "loss": 0.0307,
+      "step": 28859
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.491580640504925e-07,
+      "loss": 0.0466,
+      "step": 28860
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.488048550844728e-07,
+      "loss": 0.0685,
+      "step": 28861
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.484517390139056e-07,
+      "loss": 0.0448,
+      "step": 28862
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.48098715842298e-07,
+      "loss": 0.0381,
+      "step": 28863
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.47745785573154e-07,
+      "loss": 0.0666,
+      "step": 28864
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.473929482099827e-07,
+      "loss": 0.0488,
+      "step": 28865
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.470402037562883e-07,
+      "loss": 0.0474,
+      "step": 28866
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.466875522155758e-07,
+      "loss": 0.0772,
+      "step": 28867
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.463349935913476e-07,
+      "loss": 0.0578,
+      "step": 28868
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.459825278871046e-07,
+      "loss": 0.0446,
+      "step": 28869
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.456301551063504e-07,
+      "loss": 0.0393,
+      "step": 28870
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.452778752525868e-07,
+      "loss": 0.0524,
+      "step": 28871
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.449256883293098e-07,
+      "loss": 0.0428,
+      "step": 28872
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.4457359434002e-07,
+      "loss": 0.0446,
+      "step": 28873
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.442215932882157e-07,
+      "loss": 0.0409,
+      "step": 28874
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.438696851773906e-07,
+      "loss": 0.0676,
+      "step": 28875
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.435178700110434e-07,
+      "loss": 0.0367,
+      "step": 28876
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.43166147792672e-07,
+      "loss": 0.0362,
+      "step": 28877
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.428145185257628e-07,
+      "loss": 0.0525,
+      "step": 28878
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.42462982213814e-07,
+      "loss": 0.0482,
+      "step": 28879
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.421115388603183e-07,
+      "loss": 0.0448,
+      "step": 28880
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.417601884687652e-07,
+      "loss": 0.0507,
+      "step": 28881
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.414089310426441e-07,
+      "loss": 0.0371,
+      "step": 28882
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.410577665854467e-07,
+      "loss": 0.0478,
+      "step": 28883
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.407066951006591e-07,
+      "loss": 0.0636,
+      "step": 28884
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.403557165917695e-07,
+      "loss": 0.0358,
+      "step": 28885
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.400048310622697e-07,
+      "loss": 0.0503,
+      "step": 28886
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.396540385156369e-07,
+      "loss": 0.0483,
+      "step": 28887
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.393033389553604e-07,
+      "loss": 0.0354,
+      "step": 28888
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.389527323849254e-07,
+      "loss": 0.0367,
+      "step": 28889
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.386022188078111e-07,
+      "loss": 0.0373,
+      "step": 28890
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.382517982275038e-07,
+      "loss": 0.0376,
+      "step": 28891
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.379014706474817e-07,
+      "loss": 0.0479,
+      "step": 28892
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.375512360712233e-07,
+      "loss": 0.0399,
+      "step": 28893
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.372010945022111e-07,
+      "loss": 0.0459,
+      "step": 28894
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.368510459439248e-07,
+      "loss": 0.0367,
+      "step": 28895
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.365010903998381e-07,
+      "loss": 0.0698,
+      "step": 28896
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.361512278734283e-07,
+      "loss": 0.0339,
+      "step": 28897
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.358014583681726e-07,
+      "loss": 0.0505,
+      "step": 28898
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.354517818875427e-07,
+      "loss": 0.04,
+      "step": 28899
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.351021984350147e-07,
+      "loss": 0.0581,
+      "step": 28900
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.347527080140648e-07,
+      "loss": 0.0429,
+      "step": 28901
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.344033106281566e-07,
+      "loss": 0.0346,
+      "step": 28902
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.340540062807654e-07,
+      "loss": 0.0472,
+      "step": 28903
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.337047949753638e-07,
+      "loss": 0.0449,
+      "step": 28904
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.333556767154159e-07,
+      "loss": 0.0474,
+      "step": 28905
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.330066515043943e-07,
+      "loss": 0.044,
+      "step": 28906
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.326577193457628e-07,
+      "loss": 0.0364,
+      "step": 28907
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.323088802429889e-07,
+      "loss": 0.0463,
+      "step": 28908
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.319601341995385e-07,
+      "loss": 0.0495,
+      "step": 28909
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.316114812188778e-07,
+      "loss": 0.0405,
+      "step": 28910
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.312629213044663e-07,
+      "loss": 0.0206,
+      "step": 28911
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.309144544597679e-07,
+      "loss": 0.0536,
+      "step": 28912
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.305660806882463e-07,
+      "loss": 0.0499,
+      "step": 28913
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.302177999933601e-07,
+      "loss": 0.041,
+      "step": 28914
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.298696123785719e-07,
+      "loss": 0.0354,
+      "step": 28915
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.29521517847338e-07,
+      "loss": 0.0483,
+      "step": 28916
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.291735164031165e-07,
+      "loss": 0.034,
+      "step": 28917
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.288256080493638e-07,
+      "loss": 0.0478,
+      "step": 28918
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.284777927895403e-07,
+      "loss": 0.0521,
+      "step": 28919
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.281300706270977e-07,
+      "loss": 0.0462,
+      "step": 28920
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.277824415654887e-07,
+      "loss": 0.0526,
+      "step": 28921
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.274349056081719e-07,
+      "loss": 0.0454,
+      "step": 28922
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.270874627585944e-07,
+      "loss": 0.0613,
+      "step": 28923
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.26740113020211e-07,
+      "loss": 0.0504,
+      "step": 28924
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.263928563964727e-07,
+      "loss": 0.0388,
+      "step": 28925
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.260456928908254e-07,
+      "loss": 0.0488,
+      "step": 28926
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.256986225067196e-07,
+      "loss": 0.0611,
+      "step": 28927
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.253516452476071e-07,
+      "loss": 0.0544,
+      "step": 28928
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.250047611169285e-07,
+      "loss": 0.0257,
+      "step": 28929
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.246579701181354e-07,
+      "loss": 0.0525,
+      "step": 28930
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.243112722546696e-07,
+      "loss": 0.0352,
+      "step": 28931
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.239646675299737e-07,
+      "loss": 0.0416,
+      "step": 28932
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.236181559474952e-07,
+      "loss": 0.059,
+      "step": 28933
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.232717375106722e-07,
+      "loss": 0.0321,
+      "step": 28934
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2292541222295e-07,
+      "loss": 0.0576,
+      "step": 28935
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.225791800877656e-07,
+      "loss": 0.053,
+      "step": 28936
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.222330411085619e-07,
+      "loss": 0.0364,
+      "step": 28937
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.21886995288773e-07,
+      "loss": 0.0435,
+      "step": 28938
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.215410426318414e-07,
+      "loss": 0.0569,
+      "step": 28939
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.211951831412011e-07,
+      "loss": 0.0441,
+      "step": 28940
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.208494168202872e-07,
+      "loss": 0.0338,
+      "step": 28941
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.205037436725369e-07,
+      "loss": 0.0429,
+      "step": 28942
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.201581637013809e-07,
+      "loss": 0.0665,
+      "step": 28943
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.198126769102563e-07,
+      "loss": 0.0375,
+      "step": 28944
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.194672833025927e-07,
+      "loss": 0.0364,
+      "step": 28945
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.191219828818206e-07,
+      "loss": 0.0478,
+      "step": 28946
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.187767756513707e-07,
+      "loss": 0.0385,
+      "step": 28947
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.184316616146757e-07,
+      "loss": 0.0522,
+      "step": 28948
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.180866407751595e-07,
+      "loss": 0.0399,
+      "step": 28949
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.177417131362507e-07,
+      "loss": 0.0441,
+      "step": 28950
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.173968787013784e-07,
+      "loss": 0.059,
+      "step": 28951
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.170521374739635e-07,
+      "loss": 0.0615,
+      "step": 28952
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.167074894574343e-07,
+      "loss": 0.0597,
+      "step": 28953
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.163629346552169e-07,
+      "loss": 0.0399,
+      "step": 28954
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.160184730707264e-07,
+      "loss": 0.0518,
+      "step": 28955
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.156741047073911e-07,
+      "loss": 0.0585,
+      "step": 28956
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.153298295686305e-07,
+      "loss": 0.0587,
+      "step": 28957
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.14985647657862e-07,
+      "loss": 0.059,
+      "step": 28958
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.146415589785093e-07,
+      "loss": 0.0495,
+      "step": 28959
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.142975635339888e-07,
+      "loss": 0.0385,
+      "step": 28960
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.139536613277153e-07,
+      "loss": 0.0424,
+      "step": 28961
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.136098523631073e-07,
+      "loss": 0.0273,
+      "step": 28962
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.13266136643581e-07,
+      "loss": 0.0406,
+      "step": 28963
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.129225141725504e-07,
+      "loss": 0.0438,
+      "step": 28964
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.125789849534269e-07,
+      "loss": 0.0629,
+      "step": 28965
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.122355489896259e-07,
+      "loss": 0.0558,
+      "step": 28966
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.118922062845567e-07,
+      "loss": 0.0519,
+      "step": 28967
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.115489568416333e-07,
+      "loss": 0.0408,
+      "step": 28968
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.11205800664263e-07,
+      "loss": 0.0449,
+      "step": 28969
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.108627377558551e-07,
+      "loss": 0.0486,
+      "step": 28970
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.105197681198172e-07,
+      "loss": 0.0234,
+      "step": 28971
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.101768917595585e-07,
+      "loss": 0.0369,
+      "step": 28972
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.098341086784842e-07,
+      "loss": 0.0628,
+      "step": 28973
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.094914188799972e-07,
+      "loss": 0.0408,
+      "step": 28974
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.091488223675058e-07,
+      "loss": 0.0527,
+      "step": 28975
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.088063191444093e-07,
+      "loss": 0.0315,
+      "step": 28976
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.084639092141121e-07,
+      "loss": 0.0515,
+      "step": 28977
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.081215925800188e-07,
+      "loss": 0.0345,
+      "step": 28978
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.077793692455247e-07,
+      "loss": 0.0446,
+      "step": 28979
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.074372392140315e-07,
+      "loss": 0.0539,
+      "step": 28980
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.070952024889387e-07,
+      "loss": 0.0593,
+      "step": 28981
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.067532590736436e-07,
+      "loss": 0.0461,
+      "step": 28982
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.064114089715434e-07,
+      "loss": 0.0476,
+      "step": 28983
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.060696521860343e-07,
+      "loss": 0.0539,
+      "step": 28984
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.057279887205092e-07,
+      "loss": 0.0353,
+      "step": 28985
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.053864185783642e-07,
+      "loss": 0.061,
+      "step": 28986
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.050449417629933e-07,
+      "loss": 0.0346,
+      "step": 28987
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.047035582777871e-07,
+      "loss": 0.0459,
+      "step": 28988
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.043622681261363e-07,
+      "loss": 0.0548,
+      "step": 28989
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.040210713114336e-07,
+      "loss": 0.0462,
+      "step": 28990
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.036799678370664e-07,
+      "loss": 0.0366,
+      "step": 28991
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.033389577064252e-07,
+      "loss": 0.0441,
+      "step": 28992
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.029980409228964e-07,
+      "loss": 0.0351,
+      "step": 28993
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.026572174898659e-07,
+      "loss": 0.0543,
+      "step": 28994
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.023164874107202e-07,
+      "loss": 0.0638,
+      "step": 28995
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.019758506888451e-07,
+      "loss": 0.0551,
+      "step": 28996
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.016353073276237e-07,
+      "loss": 0.0661,
+      "step": 28997
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.012948573304389e-07,
+      "loss": 0.048,
+      "step": 28998
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.009545007006734e-07,
+      "loss": 0.0404,
+      "step": 28999
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.006142374417057e-07,
+      "loss": 0.0572,
+      "step": 29000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.002740675569207e-07,
+      "loss": 0.0583,
+      "step": 29001
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.999339910496948e-07,
+      "loss": 0.0456,
+      "step": 29002
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.995940079234042e-07,
+      "loss": 0.0505,
+      "step": 29003
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.992541181814293e-07,
+      "loss": 0.0321,
+      "step": 29004
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.989143218271487e-07,
+      "loss": 0.0331,
+      "step": 29005
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.985746188639319e-07,
+      "loss": 0.0264,
+      "step": 29006
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.982350092951594e-07,
+      "loss": 0.0434,
+      "step": 29007
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.97895493124202e-07,
+      "loss": 0.0553,
+      "step": 29008
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.975560703544326e-07,
+      "loss": 0.0291,
+      "step": 29009
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.972167409892238e-07,
+      "loss": 0.0382,
+      "step": 29010
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.968775050319442e-07,
+      "loss": 0.0397,
+      "step": 29011
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.965383624859677e-07,
+      "loss": 0.0468,
+      "step": 29012
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.961993133546595e-07,
+      "loss": 0.0456,
+      "step": 29013
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.958603576413913e-07,
+      "loss": 0.0419,
+      "step": 29014
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.95521495349527e-07,
+      "loss": 0.0236,
+      "step": 29015
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.951827264824361e-07,
+      "loss": 0.041,
+      "step": 29016
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.948440510434828e-07,
+      "loss": 0.0416,
+      "step": 29017
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.945054690360286e-07,
+      "loss": 0.047,
+      "step": 29018
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.94166980463442e-07,
+      "loss": 0.0342,
+      "step": 29019
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.938285853290804e-07,
+      "loss": 0.0443,
+      "step": 29020
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.934902836363099e-07,
+      "loss": 0.0446,
+      "step": 29021
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.931520753884889e-07,
+      "loss": 0.0341,
+      "step": 29022
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.928139605889782e-07,
+      "loss": 0.0231,
+      "step": 29023
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.924759392411339e-07,
+      "loss": 0.0474,
+      "step": 29024
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.921380113483188e-07,
+      "loss": 0.0335,
+      "step": 29025
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.91800176913887e-07,
+      "loss": 0.04,
+      "step": 29026
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.914624359411925e-07,
+      "loss": 0.0632,
+      "step": 29027
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.911247884335958e-07,
+      "loss": 0.0483,
+      "step": 29028
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.907872343944454e-07,
+      "loss": 0.054,
+      "step": 29029
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.904497738270975e-07,
+      "loss": 0.0546,
+      "step": 29030
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.901124067349051e-07,
+      "loss": 0.0312,
+      "step": 29031
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.897751331212198e-07,
+      "loss": 0.0258,
+      "step": 29032
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.89437952989389e-07,
+      "loss": 0.0444,
+      "step": 29033
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.891008663427667e-07,
+      "loss": 0.0361,
+      "step": 29034
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.887638731846967e-07,
+      "loss": 0.0317,
+      "step": 29035
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.884269735185299e-07,
+      "loss": 0.0589,
+      "step": 29036
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.880901673476136e-07,
+      "loss": 0.037,
+      "step": 29037
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.877534546752906e-07,
+      "loss": 0.0484,
+      "step": 29038
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.874168355049081e-07,
+      "loss": 0.0207,
+      "step": 29039
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.870803098398104e-07,
+      "loss": 0.0375,
+      "step": 29040
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.867438776833401e-07,
+      "loss": 0.0695,
+      "step": 29041
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.86407539038838e-07,
+      "loss": 0.0443,
+      "step": 29042
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.860712939096469e-07,
+      "loss": 0.0495,
+      "step": 29043
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.857351422991053e-07,
+      "loss": 0.0475,
+      "step": 29044
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.85399084210555e-07,
+      "loss": 0.0358,
+      "step": 29045
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.850631196473344e-07,
+      "loss": 0.0333,
+      "step": 29046
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.847272486127786e-07,
+      "loss": 0.0507,
+      "step": 29047
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.843914711102239e-07,
+      "loss": 0.0386,
+      "step": 29048
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.840557871430097e-07,
+      "loss": 0.0486,
+      "step": 29049
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.83720196714469e-07,
+      "loss": 0.0495,
+      "step": 29050
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.833846998279324e-07,
+      "loss": 0.0586,
+      "step": 29051
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.830492964867374e-07,
+      "loss": 0.0313,
+      "step": 29052
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.827139866942122e-07,
+      "loss": 0.0199,
+      "step": 29053
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.823787704536899e-07,
+      "loss": 0.0508,
+      "step": 29054
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.820436477685021e-07,
+      "loss": 0.0415,
+      "step": 29055
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.81708618641974e-07,
+      "loss": 0.0564,
+      "step": 29056
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.813736830774352e-07,
+      "loss": 0.0434,
+      "step": 29057
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.810388410782153e-07,
+      "loss": 0.0559,
+      "step": 29058
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.807040926476371e-07,
+      "loss": 0.0354,
+      "step": 29059
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.803694377890301e-07,
+      "loss": 0.0598,
+      "step": 29060
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.800348765057162e-07,
+      "loss": 0.0302,
+      "step": 29061
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.797004088010172e-07,
+      "loss": 0.0472,
+      "step": 29062
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.793660346782582e-07,
+      "loss": 0.0463,
+      "step": 29063
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.790317541407619e-07,
+      "loss": 0.0522,
+      "step": 29064
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.786975671918482e-07,
+      "loss": 0.051,
+      "step": 29065
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.783634738348343e-07,
+      "loss": 0.0505,
+      "step": 29066
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.78029474073043e-07,
+      "loss": 0.0336,
+      "step": 29067
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.776955679097885e-07,
+      "loss": 0.0412,
+      "step": 29068
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.773617553483901e-07,
+      "loss": 0.0592,
+      "step": 29069
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.770280363921665e-07,
+      "loss": 0.0378,
+      "step": 29070
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.766944110444273e-07,
+      "loss": 0.0361,
+      "step": 29071
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.763608793084885e-07,
+      "loss": 0.0424,
+      "step": 29072
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.760274411876677e-07,
+      "loss": 0.0317,
+      "step": 29073
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.756940966852709e-07,
+      "loss": 0.047,
+      "step": 29074
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.753608458046145e-07,
+      "loss": 0.0352,
+      "step": 29075
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.750276885490081e-07,
+      "loss": 0.0281,
+      "step": 29076
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.746946249217577e-07,
+      "loss": 0.0473,
+      "step": 29077
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.743616549261755e-07,
+      "loss": 0.0419,
+      "step": 29078
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.740287785655719e-07,
+      "loss": 0.029,
+      "step": 29079
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.736959958432464e-07,
+      "loss": 0.0553,
+      "step": 29080
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.733633067625089e-07,
+      "loss": 0.0459,
+      "step": 29081
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.730307113266664e-07,
+      "loss": 0.0557,
+      "step": 29082
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.726982095390177e-07,
+      "loss": 0.059,
+      "step": 29083
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.723658014028721e-07,
+      "loss": 0.0504,
+      "step": 29084
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.720334869215282e-07,
+      "loss": 0.0629,
+      "step": 29085
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.717012660982857e-07,
+      "loss": 0.0419,
+      "step": 29086
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.713691389364462e-07,
+      "loss": 0.0349,
+      "step": 29087
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.710371054393127e-07,
+      "loss": 0.0383,
+      "step": 29088
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.707051656101803e-07,
+      "loss": 0.0475,
+      "step": 29089
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.703733194523442e-07,
+      "loss": 0.0377,
+      "step": 29090
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.700415669691061e-07,
+      "loss": 0.0506,
+      "step": 29091
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.69709908163758e-07,
+      "loss": 0.0522,
+      "step": 29092
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.693783430395972e-07,
+      "loss": 0.0408,
+      "step": 29093
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.690468715999154e-07,
+      "loss": 0.0554,
+      "step": 29094
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.687154938480055e-07,
+      "loss": 0.0362,
+      "step": 29095
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.683842097871595e-07,
+      "loss": 0.0354,
+      "step": 29096
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.680530194206712e-07,
+      "loss": 0.0381,
+      "step": 29097
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.677219227518272e-07,
+      "loss": 0.0418,
+      "step": 29098
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.673909197839189e-07,
+      "loss": 0.0437,
+      "step": 29099
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.670600105202328e-07,
+      "loss": 0.0593,
+      "step": 29100
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.667291949640563e-07,
+      "loss": 0.0378,
+      "step": 29101
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.663984731186777e-07,
+      "loss": 0.0458,
+      "step": 29102
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.660678449873802e-07,
+      "loss": 0.0524,
+      "step": 29103
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.657373105734498e-07,
+      "loss": 0.0474,
+      "step": 29104
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.654068698801696e-07,
+      "loss": 0.0443,
+      "step": 29105
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.650765229108202e-07,
+      "loss": 0.0348,
+      "step": 29106
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.647462696686856e-07,
+      "loss": 0.0404,
+      "step": 29107
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.644161101570478e-07,
+      "loss": 0.0649,
+      "step": 29108
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.640860443791851e-07,
+      "loss": 0.0486,
+      "step": 29109
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.637560723383739e-07,
+      "loss": 0.0273,
+      "step": 29110
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.63426194037896e-07,
+      "loss": 0.0512,
+      "step": 29111
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.630964094810265e-07,
+      "loss": 0.0521,
+      "step": 29112
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.627667186710428e-07,
+      "loss": 0.0463,
+      "step": 29113
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.62437121611219e-07,
+      "loss": 0.0295,
+      "step": 29114
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.62107618304828e-07,
+      "loss": 0.0514,
+      "step": 29115
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.617782087551449e-07,
+      "loss": 0.0372,
+      "step": 29116
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.614488929654427e-07,
+      "loss": 0.0476,
+      "step": 29117
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.611196709389921e-07,
+      "loss": 0.0422,
+      "step": 29118
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.607905426790627e-07,
+      "loss": 0.0425,
+      "step": 29119
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.604615081889253e-07,
+      "loss": 0.0224,
+      "step": 29120
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.601325674718471e-07,
+      "loss": 0.0529,
+      "step": 29121
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.59803720531098e-07,
+      "loss": 0.0263,
+      "step": 29122
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.594749673699451e-07,
+      "loss": 0.0362,
+      "step": 29123
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.591463079916504e-07,
+      "loss": 0.0633,
+      "step": 29124
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.588177423994812e-07,
+      "loss": 0.0519,
+      "step": 29125
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.584892705967027e-07,
+      "loss": 0.0365,
+      "step": 29126
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.581608925865755e-07,
+      "loss": 0.034,
+      "step": 29127
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.578326083723651e-07,
+      "loss": 0.0441,
+      "step": 29128
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.575044179573297e-07,
+      "loss": 0.0505,
+      "step": 29129
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.57176321344729e-07,
+      "loss": 0.0327,
+      "step": 29130
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.568483185378248e-07,
+      "loss": 0.0538,
+      "step": 29131
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.565204095398757e-07,
+      "loss": 0.0329,
+      "step": 29132
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.561925943541368e-07,
+      "loss": 0.0316,
+      "step": 29133
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.558648729838644e-07,
+      "loss": 0.0533,
+      "step": 29134
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.555372454323182e-07,
+      "loss": 0.0535,
+      "step": 29135
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.552097117027477e-07,
+      "loss": 0.0428,
+      "step": 29136
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.548822717984115e-07,
+      "loss": 0.0487,
+      "step": 29137
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.545549257225602e-07,
+      "loss": 0.0384,
+      "step": 29138
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.542276734784435e-07,
+      "loss": 0.0518,
+      "step": 29139
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.539005150693144e-07,
+      "loss": 0.0579,
+      "step": 29140
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.535734504984247e-07,
+      "loss": 0.0475,
+      "step": 29141
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.532464797690207e-07,
+      "loss": 0.054,
+      "step": 29142
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.529196028843509e-07,
+      "loss": 0.058,
+      "step": 29143
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.525928198476648e-07,
+      "loss": 0.056,
+      "step": 29144
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.522661306622046e-07,
+      "loss": 0.0455,
+      "step": 29145
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.519395353312195e-07,
+      "loss": 0.0294,
+      "step": 29146
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.516130338579539e-07,
+      "loss": 0.0434,
+      "step": 29147
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.512866262456484e-07,
+      "loss": 0.0463,
+      "step": 29148
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.50960312497546e-07,
+      "loss": 0.0371,
+      "step": 29149
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.506340926168918e-07,
+      "loss": 0.0553,
+      "step": 29150
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.503079666069222e-07,
+      "loss": 0.0277,
+      "step": 29151
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.499819344708802e-07,
+      "loss": 0.0494,
+      "step": 29152
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.49655996212003e-07,
+      "loss": 0.0365,
+      "step": 29153
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.49330151833527e-07,
+      "loss": 0.0395,
+      "step": 29154
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.490044013386919e-07,
+      "loss": 0.0397,
+      "step": 29155
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.486787447307341e-07,
+      "loss": 0.0522,
+      "step": 29156
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.483531820128862e-07,
+      "loss": 0.0388,
+      "step": 29157
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.480277131883827e-07,
+      "loss": 0.0503,
+      "step": 29158
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.477023382604596e-07,
+      "loss": 0.0473,
+      "step": 29159
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.473770572323456e-07,
+      "loss": 0.0403,
+      "step": 29160
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.470518701072746e-07,
+      "loss": 0.0377,
+      "step": 29161
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.467267768884765e-07,
+      "loss": 0.0625,
+      "step": 29162
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.464017775791797e-07,
+      "loss": 0.0415,
+      "step": 29163
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.460768721826126e-07,
+      "loss": 0.0534,
+      "step": 29164
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.45752060702005e-07,
+      "loss": 0.0513,
+      "step": 29165
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.454273431405832e-07,
+      "loss": 0.0488,
+      "step": 29166
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.451027195015701e-07,
+      "loss": 0.046,
+      "step": 29167
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.447781897881943e-07,
+      "loss": 0.0378,
+      "step": 29168
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.444537540036765e-07,
+      "loss": 0.0308,
+      "step": 29169
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.441294121512419e-07,
+      "loss": 0.0556,
+      "step": 29170
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.438051642341125e-07,
+      "loss": 0.0442,
+      "step": 29171
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.434810102555065e-07,
+      "loss": 0.0515,
+      "step": 29172
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.431569502186463e-07,
+      "loss": 0.0331,
+      "step": 29173
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.428329841267532e-07,
+      "loss": 0.0418,
+      "step": 29174
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.425091119830417e-07,
+      "loss": 0.0412,
+      "step": 29175
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.421853337907312e-07,
+      "loss": 0.0413,
+      "step": 29176
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.418616495530393e-07,
+      "loss": 0.0405,
+      "step": 29177
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.415380592731778e-07,
+      "loss": 0.0469,
+      "step": 29178
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.412145629543652e-07,
+      "loss": 0.0377,
+      "step": 29179
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.408911605998112e-07,
+      "loss": 0.0383,
+      "step": 29180
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.405678522127322e-07,
+      "loss": 0.0408,
+      "step": 29181
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.402446377963378e-07,
+      "loss": 0.0441,
+      "step": 29182
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.399215173538408e-07,
+      "loss": 0.0349,
+      "step": 29183
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.395984908884477e-07,
+      "loss": 0.04,
+      "step": 29184
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.392755584033715e-07,
+      "loss": 0.0408,
+      "step": 29185
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.389527199018174e-07,
+      "loss": 0.0369,
+      "step": 29186
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.386299753869927e-07,
+      "loss": 0.0338,
+      "step": 29187
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.383073248621062e-07,
+      "loss": 0.0361,
+      "step": 29188
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.379847683303585e-07,
+      "loss": 0.0319,
+      "step": 29189
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.376623057949593e-07,
+      "loss": 0.0623,
+      "step": 29190
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.37339937259107e-07,
+      "loss": 0.056,
+      "step": 29191
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.370176627260082e-07,
+      "loss": 0.0339,
+      "step": 29192
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.366954821988602e-07,
+      "loss": 0.0468,
+      "step": 29193
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.363733956808681e-07,
+      "loss": 0.0432,
+      "step": 29194
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.360514031752284e-07,
+      "loss": 0.0317,
+      "step": 29195
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.357295046851407e-07,
+      "loss": 0.0644,
+      "step": 29196
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.354077002138025e-07,
+      "loss": 0.0409,
+      "step": 29197
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3508598976441e-07,
+      "loss": 0.0587,
+      "step": 29198
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.347643733401608e-07,
+      "loss": 0.0365,
+      "step": 29199
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.344428509442501e-07,
+      "loss": 0.0505,
+      "step": 29200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.341214225798708e-07,
+      "loss": 0.0339,
+      "step": 29201
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.338000882502137e-07,
+      "loss": 0.0258,
+      "step": 29202
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.334788479584751e-07,
+      "loss": 0.0285,
+      "step": 29203
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.331577017078437e-07,
+      "loss": 0.037,
+      "step": 29204
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.328366495015113e-07,
+      "loss": 0.0407,
+      "step": 29205
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.325156913426666e-07,
+      "loss": 0.0432,
+      "step": 29206
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.321948272344968e-07,
+      "loss": 0.0488,
+      "step": 29207
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.318740571801905e-07,
+      "loss": 0.0359,
+      "step": 29208
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.315533811829354e-07,
+      "loss": 0.0462,
+      "step": 29209
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.312327992459154e-07,
+      "loss": 0.046,
+      "step": 29210
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.309123113723158e-07,
+      "loss": 0.035,
+      "step": 29211
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.305919175653206e-07,
+      "loss": 0.0244,
+      "step": 29212
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.302716178281109e-07,
+      "loss": 0.0486,
+      "step": 29213
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.299514121638704e-07,
+      "loss": 0.0533,
+      "step": 29214
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.296313005757814e-07,
+      "loss": 0.0569,
+      "step": 29215
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.293112830670211e-07,
+      "loss": 0.0486,
+      "step": 29216
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.289913596407681e-07,
+      "loss": 0.0396,
+      "step": 29217
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.286715303002043e-07,
+      "loss": 0.0541,
+      "step": 29218
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.28351795048504e-07,
+      "loss": 0.0226,
+      "step": 29219
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.280321538888434e-07,
+      "loss": 0.0543,
+      "step": 29220
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.277126068243988e-07,
+      "loss": 0.0464,
+      "step": 29221
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.273931538583443e-07,
+      "loss": 0.0532,
+      "step": 29222
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.270737949938531e-07,
+      "loss": 0.0269,
+      "step": 29223
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.267545302341004e-07,
+      "loss": 0.0465,
+      "step": 29224
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.264353595822535e-07,
+      "loss": 0.0432,
+      "step": 29225
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.261162830414845e-07,
+      "loss": 0.043,
+      "step": 29226
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.257973006149653e-07,
+      "loss": 0.0416,
+      "step": 29227
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.25478412305861e-07,
+      "loss": 0.0443,
+      "step": 29228
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.251596181173435e-07,
+      "loss": 0.0442,
+      "step": 29229
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.248409180525782e-07,
+      "loss": 0.0454,
+      "step": 29230
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.24522312114728e-07,
+      "loss": 0.0341,
+      "step": 29231
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.242038003069616e-07,
+      "loss": 0.0272,
+      "step": 29232
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.238853826324431e-07,
+      "loss": 0.0564,
+      "step": 29233
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.235670590943353e-07,
+      "loss": 0.0361,
+      "step": 29234
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.232488296957983e-07,
+      "loss": 0.0486,
+      "step": 29235
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.229306944399959e-07,
+      "loss": 0.0384,
+      "step": 29236
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.226126533300869e-07,
+      "loss": 0.0371,
+      "step": 29237
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.222947063692308e-07,
+      "loss": 0.041,
+      "step": 29238
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.219768535605885e-07,
+      "loss": 0.0496,
+      "step": 29239
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.216590949073141e-07,
+      "loss": 0.0477,
+      "step": 29240
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.213414304125664e-07,
+      "loss": 0.0504,
+      "step": 29241
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.210238600795015e-07,
+      "loss": 0.0485,
+      "step": 29242
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.207063839112714e-07,
+      "loss": 0.0195,
+      "step": 29243
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.203890019110336e-07,
+      "loss": 0.0362,
+      "step": 29244
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.200717140819389e-07,
+      "loss": 0.0486,
+      "step": 29245
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.197545204271393e-07,
+      "loss": 0.048,
+      "step": 29246
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.194374209497844e-07,
+      "loss": 0.0438,
+      "step": 29247
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.191204156530294e-07,
+      "loss": 0.0438,
+      "step": 29248
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.188035045400175e-07,
+      "loss": 0.024,
+      "step": 29249
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.184866876138994e-07,
+      "loss": 0.0404,
+      "step": 29250
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.181699648778238e-07,
+      "loss": 0.0356,
+      "step": 29251
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.178533363349347e-07,
+      "loss": 0.0469,
+      "step": 29252
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.175368019883798e-07,
+      "loss": 0.0519,
+      "step": 29253
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.172203618413019e-07,
+      "loss": 0.0511,
+      "step": 29254
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.169040158968431e-07,
+      "loss": 0.0472,
+      "step": 29255
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.165877641581485e-07,
+      "loss": 0.031,
+      "step": 29256
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.162716066283602e-07,
+      "loss": 0.0338,
+      "step": 29257
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.159555433106178e-07,
+      "loss": 0.0541,
+      "step": 29258
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1563957420806e-07,
+      "loss": 0.0412,
+      "step": 29259
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.153236993238275e-07,
+      "loss": 0.041,
+      "step": 29260
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.150079186610568e-07,
+      "loss": 0.0407,
+      "step": 29261
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.146922322228865e-07,
+      "loss": 0.0445,
+      "step": 29262
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.143766400124517e-07,
+      "loss": 0.0429,
+      "step": 29263
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.140611420328867e-07,
+      "loss": 0.0347,
+      "step": 29264
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.137457382873257e-07,
+      "loss": 0.0467,
+      "step": 29265
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.134304287789049e-07,
+      "loss": 0.0338,
+      "step": 29266
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.131152135107531e-07,
+      "loss": 0.0416,
+      "step": 29267
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.128000924860044e-07,
+      "loss": 0.0399,
+      "step": 29268
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.124850657077873e-07,
+      "loss": 0.0298,
+      "step": 29269
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.121701331792306e-07,
+      "loss": 0.0467,
+      "step": 29270
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.11855294903466e-07,
+      "loss": 0.0496,
+      "step": 29271
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.115405508836169e-07,
+      "loss": 0.0398,
+      "step": 29272
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.112259011228149e-07,
+      "loss": 0.0413,
+      "step": 29273
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.109113456241833e-07,
+      "loss": 0.0251,
+      "step": 29274
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.105968843908449e-07,
+      "loss": 0.0444,
+      "step": 29275
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.102825174259251e-07,
+      "loss": 0.0439,
+      "step": 29276
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.099682447325493e-07,
+      "loss": 0.0449,
+      "step": 29277
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.096540663138371e-07,
+      "loss": 0.0332,
+      "step": 29278
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.093399821729084e-07,
+      "loss": 0.0537,
+      "step": 29279
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.090259923128871e-07,
+      "loss": 0.0439,
+      "step": 29280
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.087120967368875e-07,
+      "loss": 0.0375,
+      "step": 29281
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.083982954480326e-07,
+      "loss": 0.0443,
+      "step": 29282
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.080845884494379e-07,
+      "loss": 0.0505,
+      "step": 29283
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.077709757442173e-07,
+      "loss": 0.0533,
+      "step": 29284
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.074574573354896e-07,
+      "loss": 0.0663,
+      "step": 29285
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.071440332263689e-07,
+      "loss": 0.0323,
+      "step": 29286
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.068307034199682e-07,
+      "loss": 0.0615,
+      "step": 29287
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.065174679193984e-07,
+      "loss": 0.0509,
+      "step": 29288
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.06204326727775e-07,
+      "loss": 0.0579,
+      "step": 29289
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.058912798482052e-07,
+      "loss": 0.0362,
+      "step": 29290
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.055783272837988e-07,
+      "loss": 0.0384,
+      "step": 29291
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.052654690376701e-07,
+      "loss": 0.0416,
+      "step": 29292
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0495270511292e-07,
+      "loss": 0.0412,
+      "step": 29293
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.04640035512659e-07,
+      "loss": 0.0553,
+      "step": 29294
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.043274602399939e-07,
+      "loss": 0.0453,
+      "step": 29295
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.040149792980264e-07,
+      "loss": 0.042,
+      "step": 29296
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.037025926898654e-07,
+      "loss": 0.0266,
+      "step": 29297
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.033903004186125e-07,
+      "loss": 0.0317,
+      "step": 29298
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.030781024873665e-07,
+      "loss": 0.0312,
+      "step": 29299
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.027659988992328e-07,
+      "loss": 0.0362,
+      "step": 29300
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.024539896573122e-07,
+      "loss": 0.0522,
+      "step": 29301
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.02142074764701e-07,
+      "loss": 0.0348,
+      "step": 29302
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.01830254224499e-07,
+      "loss": 0.0896,
+      "step": 29303
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.015185280398061e-07,
+      "loss": 0.0337,
+      "step": 29304
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.012068962137151e-07,
+      "loss": 0.0431,
+      "step": 29305
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.008953587493259e-07,
+      "loss": 0.0421,
+      "step": 29306
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.005839156497317e-07,
+      "loss": 0.0403,
+      "step": 29307
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.002725669180242e-07,
+      "loss": 0.0391,
+      "step": 29308
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.999613125572989e-07,
+      "loss": 0.0423,
+      "step": 29309
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.996501525706477e-07,
+      "loss": 0.0311,
+      "step": 29310
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.993390869611614e-07,
+      "loss": 0.0348,
+      "step": 29311
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.990281157319288e-07,
+      "loss": 0.0427,
+      "step": 29312
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.987172388860407e-07,
+      "loss": 0.0445,
+      "step": 29313
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.984064564265845e-07,
+      "loss": 0.0382,
+      "step": 29314
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.980957683566468e-07,
+      "loss": 0.0457,
+      "step": 29315
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.977851746793183e-07,
+      "loss": 0.0278,
+      "step": 29316
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.974746753976789e-07,
+      "loss": 0.0346,
+      "step": 29317
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.971642705148161e-07,
+      "loss": 0.0319,
+      "step": 29318
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.968539600338129e-07,
+      "loss": 0.0351,
+      "step": 29319
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.965437439577514e-07,
+      "loss": 0.0502,
+      "step": 29320
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.962336222897146e-07,
+      "loss": 0.0436,
+      "step": 29321
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.959235950327834e-07,
+      "loss": 0.0582,
+      "step": 29322
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.956136621900342e-07,
+      "loss": 0.0497,
+      "step": 29323
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.953038237645491e-07,
+      "loss": 0.047,
+      "step": 29324
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.949940797594066e-07,
+      "loss": 0.0672,
+      "step": 29325
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.946844301776821e-07,
+      "loss": 0.0299,
+      "step": 29326
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.94374875022452e-07,
+      "loss": 0.0373,
+      "step": 29327
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.940654142967915e-07,
+      "loss": 0.0481,
+      "step": 29328
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.93756048003774e-07,
+      "loss": 0.0421,
+      "step": 29329
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.934467761464756e-07,
+      "loss": 0.0303,
+      "step": 29330
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.931375987279652e-07,
+      "loss": 0.0366,
+      "step": 29331
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.928285157513157e-07,
+      "loss": 0.0348,
+      "step": 29332
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.92519527219597e-07,
+      "loss": 0.048,
+      "step": 29333
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9221063313588e-07,
+      "loss": 0.0466,
+      "step": 29334
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.919018335032322e-07,
+      "loss": 0.0779,
+      "step": 29335
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.91593128324721e-07,
+      "loss": 0.0367,
+      "step": 29336
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.912845176034142e-07,
+      "loss": 0.0342,
+      "step": 29337
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.909760013423759e-07,
+      "loss": 0.0419,
+      "step": 29338
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.906675795446725e-07,
+      "loss": 0.0395,
+      "step": 29339
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.903592522133682e-07,
+      "loss": 0.0487,
+      "step": 29340
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.900510193515229e-07,
+      "loss": 0.0289,
+      "step": 29341
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.897428809622007e-07,
+      "loss": 0.057,
+      "step": 29342
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.042,
+      "step": 29343
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.891268876133714e-07,
+      "loss": 0.059,
+      "step": 29344
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.888190326599829e-07,
+      "loss": 0.0317,
+      "step": 29345
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.885112721913565e-07,
+      "loss": 0.0589,
+      "step": 29346
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.882036062105467e-07,
+      "loss": 0.0394,
+      "step": 29347
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.878960347206152e-07,
+      "loss": 0.0498,
+      "step": 29348
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.87588557724612e-07,
+      "loss": 0.0452,
+      "step": 29349
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.872811752255968e-07,
+      "loss": 0.0414,
+      "step": 29350
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.869738872266184e-07,
+      "loss": 0.0572,
+      "step": 29351
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.866666937307329e-07,
+      "loss": 0.0436,
+      "step": 29352
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.863595947409894e-07,
+      "loss": 0.0451,
+      "step": 29353
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.860525902604407e-07,
+      "loss": 0.0386,
+      "step": 29354
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.857456802921368e-07,
+      "loss": 0.0316,
+      "step": 29355
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.85438864839124e-07,
+      "loss": 0.0455,
+      "step": 29356
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.851321439044531e-07,
+      "loss": 0.0441,
+      "step": 29357
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.848255174911675e-07,
+      "loss": 0.0439,
+      "step": 29358
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.845189856023181e-07,
+      "loss": 0.0466,
+      "step": 29359
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.842125482409455e-07,
+      "loss": 0.0263,
+      "step": 29360
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.839062054100973e-07,
+      "loss": 0.0353,
+      "step": 29361
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.835999571128147e-07,
+      "loss": 0.0345,
+      "step": 29362
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.832938033521406e-07,
+      "loss": 0.0245,
+      "step": 29363
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.829877441311171e-07,
+      "loss": 0.0465,
+      "step": 29364
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.826817794527828e-07,
+      "loss": 0.0283,
+      "step": 29365
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.823759093201786e-07,
+      "loss": 0.0376,
+      "step": 29366
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.82070133736341e-07,
+      "loss": 0.0418,
+      "step": 29367
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.817644527043097e-07,
+      "loss": 0.055,
+      "step": 29368
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.814588662271225e-07,
+      "loss": 0.0252,
+      "step": 29369
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.811533743078134e-07,
+      "loss": 0.048,
+      "step": 29370
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.808479769494156e-07,
+      "loss": 0.0369,
+      "step": 29371
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.805426741549668e-07,
+      "loss": 0.049,
+      "step": 29372
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.802374659274955e-07,
+      "loss": 0.0497,
+      "step": 29373
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.799323522700383e-07,
+      "loss": 0.0476,
+      "step": 29374
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.796273331856227e-07,
+      "loss": 0.0355,
+      "step": 29375
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.793224086772785e-07,
+      "loss": 0.0417,
+      "step": 29376
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.790175787480366e-07,
+      "loss": 0.0282,
+      "step": 29377
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.787128434009259e-07,
+      "loss": 0.0236,
+      "step": 29378
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.784082026389735e-07,
+      "loss": 0.0539,
+      "step": 29379
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.78103656465203e-07,
+      "loss": 0.0442,
+      "step": 29380
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.777992048826419e-07,
+      "loss": 0.0586,
+      "step": 29381
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.774948478943142e-07,
+      "loss": 0.0436,
+      "step": 29382
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.771905855032432e-07,
+      "loss": 0.0532,
+      "step": 29383
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.768864177124543e-07,
+      "loss": 0.0602,
+      "step": 29384
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.76582344524964e-07,
+      "loss": 0.0771,
+      "step": 29385
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.762783659437964e-07,
+      "loss": 0.0527,
+      "step": 29386
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7597448197197025e-07,
+      "loss": 0.0436,
+      "step": 29387
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.756706926125054e-07,
+      "loss": 0.0371,
+      "step": 29388
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.753669978684183e-07,
+      "loss": 0.031,
+      "step": 29389
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7506339774272657e-07,
+      "loss": 0.0403,
+      "step": 29390
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.747598922384455e-07,
+      "loss": 0.0519,
+      "step": 29391
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7445648135859057e-07,
+      "loss": 0.0426,
+      "step": 29392
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7415316510617816e-07,
+      "loss": 0.0597,
+      "step": 29393
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.73849943484217e-07,
+      "loss": 0.0403,
+      "step": 29394
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.735468164957213e-07,
+      "loss": 0.0512,
+      "step": 29395
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.732437841437054e-07,
+      "loss": 0.0398,
+      "step": 29396
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7294084643117465e-07,
+      "loss": 0.0318,
+      "step": 29397
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7263800336114106e-07,
+      "loss": 0.0456,
+      "step": 29398
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7233525493661446e-07,
+      "loss": 0.0628,
+      "step": 29399
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.72032601160598e-07,
+      "loss": 0.0272,
+      "step": 29400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.717300420361015e-07,
+      "loss": 0.0389,
+      "step": 29401
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7142757756613144e-07,
+      "loss": 0.0523,
+      "step": 29402
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7112520775369096e-07,
+      "loss": 0.0474,
+      "step": 29403
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7082293260178326e-07,
+      "loss": 0.0434,
+      "step": 29404
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.705207521134125e-07,
+      "loss": 0.0512,
+      "step": 29405
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.702186662915786e-07,
+      "loss": 0.0337,
+      "step": 29406
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.6991667513928363e-07,
+      "loss": 0.0487,
+      "step": 29407
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.696147786595318e-07,
+      "loss": 0.0579,
+      "step": 29408
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6931297685531397e-07,
+      "loss": 0.0392,
+      "step": 29409
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.690112697296334e-07,
+      "loss": 0.0277,
+      "step": 29410
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.687096572854877e-07,
+      "loss": 0.0493,
+      "step": 29411
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6840813952586995e-07,
+      "loss": 0.0511,
+      "step": 29412
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.681067164537789e-07,
+      "loss": 0.037,
+      "step": 29413
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6780538807220776e-07,
+      "loss": 0.049,
+      "step": 29414
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6750415438414745e-07,
+      "loss": 0.0349,
+      "step": 29415
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.672030153925922e-07,
+      "loss": 0.0602,
+      "step": 29416
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.669019711005363e-07,
+      "loss": 0.0403,
+      "step": 29417
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6660102151096623e-07,
+      "loss": 0.0327,
+      "step": 29418
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6630016662687294e-07,
+      "loss": 0.0622,
+      "step": 29419
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.659994064512463e-07,
+      "loss": 0.0561,
+      "step": 29420
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6569874098707166e-07,
+      "loss": 0.0329,
+      "step": 29421
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6539817023734e-07,
+      "loss": 0.0561,
+      "step": 29422
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.650976942050334e-07,
+      "loss": 0.0354,
+      "step": 29423
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6479731289313824e-07,
+      "loss": 0.0509,
+      "step": 29424
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.644970263046367e-07,
+      "loss": 0.0429,
+      "step": 29425
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6419683444251626e-07,
+      "loss": 0.0427,
+      "step": 29426
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.638967373097547e-07,
+      "loss": 0.066,
+      "step": 29427
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.635967349093351e-07,
+      "loss": 0.0564,
+      "step": 29428
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.632968272442373e-07,
+      "loss": 0.034,
+      "step": 29429
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6299701431744e-07,
+      "loss": 0.028,
+      "step": 29430
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.626972961319243e-07,
+      "loss": 0.0266,
+      "step": 29431
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6239767269066536e-07,
+      "loss": 0.0367,
+      "step": 29432
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.620981439966377e-07,
+      "loss": 0.0372,
+      "step": 29433
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6179871005282097e-07,
+      "loss": 0.0326,
+      "step": 29434
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.614993708621862e-07,
+      "loss": 0.0451,
+      "step": 29435
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.612001264277088e-07,
+      "loss": 0.0331,
+      "step": 29436
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.609009767523631e-07,
+      "loss": 0.0372,
+      "step": 29437
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6060192183911887e-07,
+      "loss": 0.0505,
+      "step": 29438
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.60302961690946e-07,
+      "loss": 0.0295,
+      "step": 29439
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6000409631081654e-07,
+      "loss": 0.0548,
+      "step": 29440
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5970532570169815e-07,
+      "loss": 0.0417,
+      "step": 29441
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5940664986655947e-07,
+      "loss": 0.0403,
+      "step": 29442
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5910806880836713e-07,
+      "loss": 0.044,
+      "step": 29443
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5880958253008755e-07,
+      "loss": 0.0478,
+      "step": 29444
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.58511191034684e-07,
+      "loss": 0.0752,
+      "step": 29445
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5821289432512517e-07,
+      "loss": 0.035,
+      "step": 29446
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.579146924043709e-07,
+      "loss": 0.0384,
+      "step": 29447
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5761658527538443e-07,
+      "loss": 0.0382,
+      "step": 29448
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.573185729411267e-07,
+      "loss": 0.0385,
+      "step": 29449
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5702065540455865e-07,
+      "loss": 0.047,
+      "step": 29450
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5672283266864016e-07,
+      "loss": 0.0543,
+      "step": 29451
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5642510473632997e-07,
+      "loss": 0.0423,
+      "step": 29452
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5612747161058347e-07,
+      "loss": 0.0446,
+      "step": 29453
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.558299332943594e-07,
+      "loss": 0.0401,
+      "step": 29454
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.555324897906133e-07,
+      "loss": 0.0477,
+      "step": 29455
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.552351411023015e-07,
+      "loss": 0.0437,
+      "step": 29456
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.54937887232374e-07,
+      "loss": 0.0661,
+      "step": 29457
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5464072818378836e-07,
+      "loss": 0.0509,
+      "step": 29458
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5434366395949227e-07,
+      "loss": 0.0446,
+      "step": 29459
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5404669456243887e-07,
+      "loss": 0.0529,
+      "step": 29460
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.537498199955803e-07,
+      "loss": 0.0436,
+      "step": 29461
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5345304026186086e-07,
+      "loss": 0.0493,
+      "step": 29462
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.531563553642326e-07,
+      "loss": 0.0487,
+      "step": 29463
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5285976530564214e-07,
+      "loss": 0.0462,
+      "step": 29464
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.525632700890348e-07,
+      "loss": 0.0455,
+      "step": 29465
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.522668697173571e-07,
+      "loss": 0.067,
+      "step": 29466
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5197056419355346e-07,
+      "loss": 0.0356,
+      "step": 29467
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.516743535205659e-07,
+      "loss": 0.0622,
+      "step": 29468
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.513782377013376e-07,
+      "loss": 0.0395,
+      "step": 29469
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5108221673881404e-07,
+      "loss": 0.0311,
+      "step": 29470
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5078629063592837e-07,
+      "loss": 0.0467,
+      "step": 29471
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.504904593956261e-07,
+      "loss": 0.0496,
+      "step": 29472
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5019472302084476e-07,
+      "loss": 0.043,
+      "step": 29473
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.49899081514521e-07,
+      "loss": 0.0432,
+      "step": 29474
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4960353487959354e-07,
+      "loss": 0.0303,
+      "step": 29475
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.493080831189978e-07,
+      "loss": 0.0555,
+      "step": 29476
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.490127262356669e-07,
+      "loss": 0.0421,
+      "step": 29477
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.487174642325365e-07,
+      "loss": 0.0306,
+      "step": 29478
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4842229711254184e-07,
+      "loss": 0.0489,
+      "step": 29479
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4812722487861173e-07,
+      "loss": 0.0318,
+      "step": 29480
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4783224753367825e-07,
+      "loss": 0.0409,
+      "step": 29481
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4753736508067356e-07,
+      "loss": 0.0451,
+      "step": 29482
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.472425775225242e-07,
+      "loss": 0.051,
+      "step": 29483
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4694788486216e-07,
+      "loss": 0.0375,
+      "step": 29484
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.46653287102512e-07,
+      "loss": 0.041,
+      "step": 29485
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4635878424650005e-07,
+      "loss": 0.0406,
+      "step": 29486
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.460643762970529e-07,
+      "loss": 0.0382,
+      "step": 29487
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.457700632570972e-07,
+      "loss": 0.0216,
+      "step": 29488
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.454758451295527e-07,
+      "loss": 0.0323,
+      "step": 29489
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.451817219173471e-07,
+      "loss": 0.0386,
+      "step": 29490
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.448876936233981e-07,
+      "loss": 0.0487,
+      "step": 29491
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4459376025062675e-07,
+      "loss": 0.0459,
+      "step": 29492
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.44299921801955e-07,
+      "loss": 0.0263,
+      "step": 29493
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.440061782803018e-07,
+      "loss": 0.0453,
+      "step": 29494
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.437125296885847e-07,
+      "loss": 0.0438,
+      "step": 29495
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.434189760297203e-07,
+      "loss": 0.0357,
+      "step": 29496
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.431255173066251e-07,
+      "loss": 0.0459,
+      "step": 29497
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.428321535222135e-07,
+      "loss": 0.0463,
+      "step": 29498
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.425388846794021e-07,
+      "loss": 0.0451,
+      "step": 29499
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.422457107811029e-07,
+      "loss": 0.0474,
+      "step": 29500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.419526318302259e-07,
+      "loss": 0.0386,
+      "step": 29501
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.416596478296864e-07,
+      "loss": 0.0317,
+      "step": 29502
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.413667587823933e-07,
+      "loss": 0.0355,
+      "step": 29503
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.410739646912576e-07,
+      "loss": 0.0492,
+      "step": 29504
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.407812655591848e-07,
+      "loss": 0.0342,
+      "step": 29505
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4048866138908576e-07,
+      "loss": 0.0684,
+      "step": 29506
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.401961521838638e-07,
+      "loss": 0.0395,
+      "step": 29507
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3990373794642884e-07,
+      "loss": 0.0465,
+      "step": 29508
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.39611418679684e-07,
+      "loss": 0.0383,
+      "step": 29509
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.393191943865316e-07,
+      "loss": 0.0487,
+      "step": 29510
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.390270650698758e-07,
+      "loss": 0.0553,
+      "step": 29511
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.387350307326199e-07,
+      "loss": 0.0349,
+      "step": 29512
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.384430913776627e-07,
+      "loss": 0.0402,
+      "step": 29513
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.381512470079063e-07,
+      "loss": 0.0564,
+      "step": 29514
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.378594976262496e-07,
+      "loss": 0.0364,
+      "step": 29515
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.37567843235589e-07,
+      "loss": 0.0442,
+      "step": 29516
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.372762838388245e-07,
+      "loss": 0.0296,
+      "step": 29517
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.369848194388504e-07,
+      "loss": 0.0545,
+      "step": 29518
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3669345003856334e-07,
+      "loss": 0.0428,
+      "step": 29519
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.364021756408565e-07,
+      "loss": 0.0539,
+      "step": 29520
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3611099624862653e-07,
+      "loss": 0.0535,
+      "step": 29521
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.35819911864761e-07,
+      "loss": 0.0405,
+      "step": 29522
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.355289224921566e-07,
+      "loss": 0.0406,
+      "step": 29523
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.35238028133701e-07,
+      "loss": 0.0458,
+      "step": 29524
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3494722879228404e-07,
+      "loss": 0.0397,
+      "step": 29525
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3465652447079673e-07,
+      "loss": 0.0514,
+      "step": 29526
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3436591517212355e-07,
+      "loss": 0.0443,
+      "step": 29527
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3407540089915435e-07,
+      "loss": 0.0382,
+      "step": 29528
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3378498165477344e-07,
+      "loss": 0.0417,
+      "step": 29529
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3349465744186745e-07,
+      "loss": 0.0666,
+      "step": 29530
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3320442826331963e-07,
+      "loss": 0.0257,
+      "step": 29531
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3291429412201323e-07,
+      "loss": 0.0284,
+      "step": 29532
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.326242550208304e-07,
+      "loss": 0.0344,
+      "step": 29533
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.32334310962651e-07,
+      "loss": 0.0388,
+      "step": 29534
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.320444619503583e-07,
+      "loss": 0.0422,
+      "step": 29535
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3175470798682786e-07,
+      "loss": 0.0358,
+      "step": 29536
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.314650490749417e-07,
+      "loss": 0.0227,
+      "step": 29537
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3117548521757646e-07,
+      "loss": 0.0518,
+      "step": 29538
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3088601641760875e-07,
+      "loss": 0.0399,
+      "step": 29539
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.305966426779118e-07,
+      "loss": 0.0642,
+      "step": 29540
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.303073640013633e-07,
+      "loss": 0.0487,
+      "step": 29541
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3001818039083434e-07,
+      "loss": 0.0427,
+      "step": 29542
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.297290918492003e-07,
+      "loss": 0.0737,
+      "step": 29543
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.294400983793323e-07,
+      "loss": 0.035,
+      "step": 29544
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2915119998409914e-07,
+      "loss": 0.0471,
+      "step": 29545
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.288623966663719e-07,
+      "loss": 0.0454,
+      "step": 29546
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.285736884290226e-07,
+      "loss": 0.0553,
+      "step": 29547
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.282850752749157e-07,
+      "loss": 0.0322,
+      "step": 29548
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.279965572069178e-07,
+      "loss": 0.0443,
+      "step": 29549
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.277081342278988e-07,
+      "loss": 0.0421,
+      "step": 29550
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2741980634071867e-07,
+      "loss": 0.0355,
+      "step": 29551
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.271315735482462e-07,
+      "loss": 0.032,
+      "step": 29552
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.268434358533458e-07,
+      "loss": 0.0288,
+      "step": 29553
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2655539325887396e-07,
+      "loss": 0.0423,
+      "step": 29554
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.262674457676952e-07,
+      "loss": 0.0492,
+      "step": 29555
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.259795933826727e-07,
+      "loss": 0.0463,
+      "step": 29556
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.25691836106662e-07,
+      "loss": 0.0519,
+      "step": 29557
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.25404173942523e-07,
+      "loss": 0.033,
+      "step": 29558
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.251166068931145e-07,
+      "loss": 0.0396,
+      "step": 29559
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.24829134961291e-07,
+      "loss": 0.0469,
+      "step": 29560
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2454175814991005e-07,
+      "loss": 0.0502,
+      "step": 29561
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.242544764618273e-07,
+      "loss": 0.0422,
+      "step": 29562
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2396728989989364e-07,
+      "loss": 0.0504,
+      "step": 29563
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.236801984669636e-07,
+      "loss": 0.0506,
+      "step": 29564
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.233932021658904e-07,
+      "loss": 0.0299,
+      "step": 29565
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2310630099952287e-07,
+      "loss": 0.0327,
+      "step": 29566
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.228194949707132e-07,
+      "loss": 0.0458,
+      "step": 29567
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.225327840823101e-07,
+      "loss": 0.0438,
+      "step": 29568
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.222461683371593e-07,
+      "loss": 0.0421,
+      "step": 29569
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2195964773810937e-07,
+      "loss": 0.0337,
+      "step": 29570
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2167322228800934e-07,
+      "loss": 0.0372,
+      "step": 29571
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.213868919897024e-07,
+      "loss": 0.0305,
+      "step": 29572
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.211006568460307e-07,
+      "loss": 0.0429,
+      "step": 29573
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.208145168598421e-07,
+      "loss": 0.0372,
+      "step": 29574
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.205284720339764e-07,
+      "loss": 0.0253,
+      "step": 29575
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.202425223712747e-07,
+      "loss": 0.0571,
+      "step": 29576
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.199566678745826e-07,
+      "loss": 0.0631,
+      "step": 29577
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.196709085467321e-07,
+      "loss": 0.0414,
+      "step": 29578
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.193852443905666e-07,
+      "loss": 0.04,
+      "step": 29579
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1909967540892493e-07,
+      "loss": 0.0704,
+      "step": 29580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1881420160463924e-07,
+      "loss": 0.0569,
+      "step": 29581
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1852882298055175e-07,
+      "loss": 0.0475,
+      "step": 29582
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.182435395394924e-07,
+      "loss": 0.0514,
+      "step": 29583
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.179583512842966e-07,
+      "loss": 0.044,
+      "step": 29584
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.176732582177967e-07,
+      "loss": 0.0497,
+      "step": 29585
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.173882603428292e-07,
+      "loss": 0.0262,
+      "step": 29586
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1710335766221967e-07,
+      "loss": 0.0333,
+      "step": 29587
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.168185501788002e-07,
+      "loss": 0.0389,
+      "step": 29588
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.16533837895402e-07,
+      "loss": 0.0305,
+      "step": 29589
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.162492208148494e-07,
+      "loss": 0.0317,
+      "step": 29590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.159646989399757e-07,
+      "loss": 0.0383,
+      "step": 29591
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1568027227360197e-07,
+      "loss": 0.0458,
+      "step": 29592
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1539594081855593e-07,
+      "loss": 0.0265,
+      "step": 29593
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1511170457766313e-07,
+      "loss": 0.0493,
+      "step": 29594
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1482756355374463e-07,
+      "loss": 0.0306,
+      "step": 29595
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1454351774962485e-07,
+      "loss": 0.0494,
+      "step": 29596
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.142595671681249e-07,
+      "loss": 0.0433,
+      "step": 29597
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1397571181206686e-07,
+      "loss": 0.0667,
+      "step": 29598
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1369195168426856e-07,
+      "loss": 0.055,
+      "step": 29599
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.134082867875511e-07,
+      "loss": 0.0444,
+      "step": 29600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.13124717124731e-07,
+      "loss": 0.0401,
+      "step": 29601
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.128412426986239e-07,
+      "loss": 0.0443,
+      "step": 29602
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1255786351204975e-07,
+      "loss": 0.0367,
+      "step": 29603
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1227457956781845e-07,
+      "loss": 0.0411,
+      "step": 29604
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1199139086874895e-07,
+      "loss": 0.0378,
+      "step": 29605
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.117082974176523e-07,
+      "loss": 0.0425,
+      "step": 29606
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1142529921734174e-07,
+      "loss": 0.0361,
+      "step": 29607
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1114239627062626e-07,
+      "loss": 0.0557,
+      "step": 29608
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1085958858032017e-07,
+      "loss": 0.0509,
+      "step": 29609
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.10576876149229e-07,
+      "loss": 0.0526,
+      "step": 29610
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1029425898016285e-07,
+      "loss": 0.0307,
+      "step": 29611
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1001173707593047e-07,
+      "loss": 0.0568,
+      "step": 29612
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.097293104393363e-07,
+      "loss": 0.047,
+      "step": 29613
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.094469790731859e-07,
+      "loss": 0.0527,
+      "step": 29614
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0916474298028694e-07,
+      "loss": 0.0504,
+      "step": 29615
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.088826021634407e-07,
+      "loss": 0.0414,
+      "step": 29616
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.086005566254503e-07,
+      "loss": 0.0335,
+      "step": 29617
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.08318606369118e-07,
+      "loss": 0.0391,
+      "step": 29618
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0803675139724385e-07,
+      "loss": 0.0476,
+      "step": 29619
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0775499171263e-07,
+      "loss": 0.0578,
+      "step": 29620
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.074733273180731e-07,
+      "loss": 0.0414,
+      "step": 29621
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.071917582163709e-07,
+      "loss": 0.0467,
+      "step": 29622
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.069102844103223e-07,
+      "loss": 0.042,
+      "step": 29623
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0662890590272397e-07,
+      "loss": 0.0262,
+      "step": 29624
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.063476226963703e-07,
+      "loss": 0.0387,
+      "step": 29625
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.060664347940546e-07,
+      "loss": 0.0496,
+      "step": 29626
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0578534219857134e-07,
+      "loss": 0.0346,
+      "step": 29627
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.055043449127116e-07,
+      "loss": 0.0363,
+      "step": 29628
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0522344293926765e-07,
+      "loss": 0.0415,
+      "step": 29629
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0494263628103267e-07,
+      "loss": 0.0484,
+      "step": 29630
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0466192494079125e-07,
+      "loss": 0.0484,
+      "step": 29631
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.043813089213344e-07,
+      "loss": 0.0662,
+      "step": 29632
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.041007882254511e-07,
+      "loss": 0.0572,
+      "step": 29633
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0382036285592564e-07,
+      "loss": 0.0493,
+      "step": 29634
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0354003281554586e-07,
+      "loss": 0.0382,
+      "step": 29635
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.032597981070951e-07,
+      "loss": 0.0578,
+      "step": 29636
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0297965873335673e-07,
+      "loss": 0.0503,
+      "step": 29637
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0269961469711516e-07,
+      "loss": 0.0587,
+      "step": 29638
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.024196660011548e-07,
+      "loss": 0.0549,
+      "step": 29639
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.021398126482512e-07,
+      "loss": 0.073,
+      "step": 29640
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.018600546411866e-07,
+      "loss": 0.0527,
+      "step": 29641
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0158039198274103e-07,
+      "loss": 0.0494,
+      "step": 29642
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.013008246756922e-07,
+      "loss": 0.034,
+      "step": 29643
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0102135272281796e-07,
+      "loss": 0.0541,
+      "step": 29644
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.007419761268938e-07,
+      "loss": 0.0374,
+      "step": 29645
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0046269489069533e-07,
+      "loss": 0.0339,
+      "step": 29646
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0018350901699587e-07,
+      "loss": 0.0193,
+      "step": 29647
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.99904418508571e-07,
+      "loss": 0.05,
+      "step": 29648
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.996254233681929e-07,
+      "loss": 0.0463,
+      "step": 29649
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9934652359863045e-07,
+      "loss": 0.0505,
+      "step": 29650
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.990677192026571e-07,
+      "loss": 0.0402,
+      "step": 29651
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.987890101830405e-07,
+      "loss": 0.0297,
+      "step": 29652
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9851039654255075e-07,
+      "loss": 0.0466,
+      "step": 29653
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.982318782839578e-07,
+      "loss": 0.0606,
+      "step": 29654
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9795345541002395e-07,
+      "loss": 0.0599,
+      "step": 29655
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.976751279235169e-07,
+      "loss": 0.0305,
+      "step": 29656
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9739689582720233e-07,
+      "loss": 0.0333,
+      "step": 29657
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9711875912384347e-07,
+      "loss": 0.0342,
+      "step": 29658
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.968407178162037e-07,
+      "loss": 0.0365,
+      "step": 29659
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.965627719070453e-07,
+      "loss": 0.0344,
+      "step": 29660
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9628492139912823e-07,
+      "loss": 0.0386,
+      "step": 29661
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9600716629521253e-07,
+      "loss": 0.0283,
+      "step": 29662
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9572950659806044e-07,
+      "loss": 0.0365,
+      "step": 29663
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.954519423104286e-07,
+      "loss": 0.0235,
+      "step": 29664
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.951744734350715e-07,
+      "loss": 0.047,
+      "step": 29665
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9489709997475036e-07,
+      "loss": 0.0556,
+      "step": 29666
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9461982193221726e-07,
+      "loss": 0.0289,
+      "step": 29667
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.943426393102279e-07,
+      "loss": 0.0542,
+      "step": 29668
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9406555211153666e-07,
+      "loss": 0.052,
+      "step": 29669
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.937885603388925e-07,
+      "loss": 0.0355,
+      "step": 29670
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9351166399505093e-07,
+      "loss": 0.0387,
+      "step": 29671
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9323486308276205e-07,
+      "loss": 0.0529,
+      "step": 29672
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9295815760477584e-07,
+      "loss": 0.044,
+      "step": 29673
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.926815475638379e-07,
+      "loss": 0.0496,
+      "step": 29674
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.924050329627005e-07,
+      "loss": 0.0539,
+      "step": 29675
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9212861380410696e-07,
+      "loss": 0.0298,
+      "step": 29676
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9185229009080617e-07,
+      "loss": 0.0269,
+      "step": 29677
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9157606182554263e-07,
+      "loss": 0.0362,
+      "step": 29678
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.912999290110575e-07,
+      "loss": 0.0363,
+      "step": 29679
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9102389165009635e-07,
+      "loss": 0.0692,
+      "step": 29680
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9074794974540143e-07,
+      "loss": 0.0566,
+      "step": 29681
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9047210329971274e-07,
+      "loss": 0.0397,
+      "step": 29682
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.901963523157726e-07,
+      "loss": 0.0352,
+      "step": 29683
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8992069679631984e-07,
+      "loss": 0.0248,
+      "step": 29684
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.896451367440901e-07,
+      "loss": 0.0431,
+      "step": 29685
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.893696721618234e-07,
+      "loss": 0.0394,
+      "step": 29686
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8909430305225536e-07,
+      "loss": 0.0502,
+      "step": 29687
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8881902941812266e-07,
+      "loss": 0.0419,
+      "step": 29688
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8854385126215755e-07,
+      "loss": 0.0359,
+      "step": 29689
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8826876858709675e-07,
+      "loss": 0.0336,
+      "step": 29690
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8799378139567024e-07,
+      "loss": 0.0452,
+      "step": 29691
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8771888969061145e-07,
+      "loss": 0.0404,
+      "step": 29692
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.874440934746515e-07,
+      "loss": 0.0341,
+      "step": 29693
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.871693927505171e-07,
+      "loss": 0.0561,
+      "step": 29694
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8689478752094167e-07,
+      "loss": 0.0472,
+      "step": 29695
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.866202777886485e-07,
+      "loss": 0.0331,
+      "step": 29696
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.863458635563688e-07,
+      "loss": 0.0353,
+      "step": 29697
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.86071544826826e-07,
+      "loss": 0.0636,
+      "step": 29698
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8579732160274663e-07,
+      "loss": 0.0545,
+      "step": 29699
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.855231938868531e-07,
+      "loss": 0.0659,
+      "step": 29700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.85249161681871e-07,
+      "loss": 0.0312,
+      "step": 29701
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.849752249905203e-07,
+      "loss": 0.0609,
+      "step": 29702
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8470138381552336e-07,
+      "loss": 0.0349,
+      "step": 29703
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8442763815960125e-07,
+      "loss": 0.0339,
+      "step": 29704
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.841539880254708e-07,
+      "loss": 0.0621,
+      "step": 29705
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.83880433415853e-07,
+      "loss": 0.0359,
+      "step": 29706
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.836069743334658e-07,
+      "loss": 0.0301,
+      "step": 29707
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.833336107810237e-07,
+      "loss": 0.0308,
+      "step": 29708
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.830603427612434e-07,
+      "loss": 0.048,
+      "step": 29709
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8278717027683933e-07,
+      "loss": 0.0417,
+      "step": 29710
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8251409333052493e-07,
+      "loss": 0.0469,
+      "step": 29711
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8224111192501357e-07,
+      "loss": 0.0403,
+      "step": 29712
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8196822606301755e-07,
+      "loss": 0.031,
+      "step": 29713
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.816954357472458e-07,
+      "loss": 0.03,
+      "step": 29714
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.814227409804094e-07,
+      "loss": 0.0532,
+      "step": 29715
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.811501417652175e-07,
+      "loss": 0.0412,
+      "step": 29716
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.808776381043788e-07,
+      "loss": 0.0354,
+      "step": 29717
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8060523000059803e-07,
+      "loss": 0.0592,
+      "step": 29718
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8033291745658396e-07,
+      "loss": 0.0429,
+      "step": 29719
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.800607004750401e-07,
+      "loss": 0.0364,
+      "step": 29720
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.797885790586708e-07,
+      "loss": 0.0282,
+      "step": 29721
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.795165532101819e-07,
+      "loss": 0.0554,
+      "step": 29722
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.79244622932271e-07,
+      "loss": 0.026,
+      "step": 29723
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.789727882276417e-07,
+      "loss": 0.0302,
+      "step": 29724
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.787010490989973e-07,
+      "loss": 0.0488,
+      "step": 29725
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.784294055490334e-07,
+      "loss": 0.0499,
+      "step": 29726
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7815785758044897e-07,
+      "loss": 0.0309,
+      "step": 29727
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.778864051959441e-07,
+      "loss": 0.0476,
+      "step": 29728
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7761504839821216e-07,
+      "loss": 0.0452,
+      "step": 29729
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.773437871899499e-07,
+      "loss": 0.0617,
+      "step": 29730
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7707262157385625e-07,
+      "loss": 0.0415,
+      "step": 29731
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.76801551552618e-07,
+      "loss": 0.0489,
+      "step": 29732
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7653057712893183e-07,
+      "loss": 0.0403,
+      "step": 29733
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7625969830549003e-07,
+      "loss": 0.0394,
+      "step": 29734
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.759889150849805e-07,
+      "loss": 0.0468,
+      "step": 29735
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.757182274700977e-07,
+      "loss": 0.0341,
+      "step": 29736
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7544763546352834e-07,
+      "loss": 0.0464,
+      "step": 29737
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7517713906795814e-07,
+      "loss": 0.0419,
+      "step": 29738
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7490673828607716e-07,
+      "loss": 0.0416,
+      "step": 29739
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.746364331205732e-07,
+      "loss": 0.0312,
+      "step": 29740
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7436622357412745e-07,
+      "loss": 0.0323,
+      "step": 29741
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.740961096494256e-07,
+      "loss": 0.0306,
+      "step": 29742
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7382609134915314e-07,
+      "loss": 0.0454,
+      "step": 29743
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.735561686759881e-07,
+      "loss": 0.0532,
+      "step": 29744
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.73286341632616e-07,
+      "loss": 0.0611,
+      "step": 29745
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.73016610221717e-07,
+      "loss": 0.0498,
+      "step": 29746
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7274697444596775e-07,
+      "loss": 0.0351,
+      "step": 29747
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.724774343080484e-07,
+      "loss": 0.0366,
+      "step": 29748
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7220798981063677e-07,
+      "loss": 0.0435,
+      "step": 29749
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7193864095640965e-07,
+      "loss": 0.0258,
+      "step": 29750
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7166938774804375e-07,
+      "loss": 0.0478,
+      "step": 29751
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7140023018821135e-07,
+      "loss": 0.0345,
+      "step": 29752
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.711311682795882e-07,
+      "loss": 0.0381,
+      "step": 29753
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7086220202484646e-07,
+      "loss": 0.0507,
+      "step": 29754
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7059333142665854e-07,
+      "loss": 0.0461,
+      "step": 29755
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7032455648769447e-07,
+      "loss": 0.0263,
+      "step": 29756
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.700558772106244e-07,
+      "loss": 0.0272,
+      "step": 29757
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.697872935981195e-07,
+      "loss": 0.0324,
+      "step": 29758
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6951880565284423e-07,
+      "loss": 0.0501,
+      "step": 29759
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6925041337746993e-07,
+      "loss": 0.0424,
+      "step": 29760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6898211677465987e-07,
+      "loss": 0.0429,
+      "step": 29761
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6871391584707984e-07,
+      "loss": 0.0548,
+      "step": 29762
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.684458105973954e-07,
+      "loss": 0.0339,
+      "step": 29763
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6817780102826774e-07,
+      "loss": 0.0465,
+      "step": 29764
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6790988714236144e-07,
+      "loss": 0.0616,
+      "step": 29765
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6764206894233655e-07,
+      "loss": 0.0379,
+      "step": 29766
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6737434643085546e-07,
+      "loss": 0.0521,
+      "step": 29767
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.671067196105749e-07,
+      "loss": 0.0368,
+      "step": 29768
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.66839188484156e-07,
+      "loss": 0.0564,
+      "step": 29769
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6657175305425673e-07,
+      "loss": 0.0444,
+      "step": 29770
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.663044133235305e-07,
+      "loss": 0.0421,
+      "step": 29771
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.660371692946363e-07,
+      "loss": 0.0775,
+      "step": 29772
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.657700209702264e-07,
+      "loss": 0.0377,
+      "step": 29773
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.655029683529565e-07,
+      "loss": 0.048,
+      "step": 29774
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6523601144548003e-07,
+      "loss": 0.0566,
+      "step": 29775
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.649691502504482e-07,
+      "loss": 0.0581,
+      "step": 29776
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6470238477051003e-07,
+      "loss": 0.0514,
+      "step": 29777
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.644357150083189e-07,
+      "loss": 0.0519,
+      "step": 29778
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6416914096652157e-07,
+      "loss": 0.0438,
+      "step": 29779
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6390266264776817e-07,
+      "loss": 0.0388,
+      "step": 29780
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.636362800547055e-07,
+      "loss": 0.0372,
+      "step": 29781
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6336999318997703e-07,
+      "loss": 0.0462,
+      "step": 29782
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6310380205623054e-07,
+      "loss": 0.0399,
+      "step": 29783
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6283770665611064e-07,
+      "loss": 0.0394,
+      "step": 29784
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6257170699226076e-07,
+      "loss": 0.0486,
+      "step": 29785
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.623058030673221e-07,
+      "loss": 0.0423,
+      "step": 29786
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.62039994883937e-07,
+      "loss": 0.0428,
+      "step": 29787
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6177428244474456e-07,
+      "loss": 0.0395,
+      "step": 29788
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6150866575238696e-07,
+      "loss": 0.0434,
+      "step": 29789
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.612431448095022e-07,
+      "loss": 0.0278,
+      "step": 29790
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6097771961872584e-07,
+      "loss": 0.0287,
+      "step": 29791
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.607123901826959e-07,
+      "loss": 0.0361,
+      "step": 29792
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.604471565040502e-07,
+      "loss": 0.0469,
+      "step": 29793
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.60182018585421e-07,
+      "loss": 0.0445,
+      "step": 29794
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.599169764294419e-07,
+      "loss": 0.0323,
+      "step": 29795
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.596520300387485e-07,
+      "loss": 0.0427,
+      "step": 29796
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.593871794159698e-07,
+      "loss": 0.0355,
+      "step": 29797
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5912242456373924e-07,
+      "loss": 0.0425,
+      "step": 29798
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.58857765484687e-07,
+      "loss": 0.053,
+      "step": 29799
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5859320218143976e-07,
+      "loss": 0.0392,
+      "step": 29800
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5832873465662663e-07,
+      "loss": 0.0375,
+      "step": 29801
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.580643629128766e-07,
+      "loss": 0.0394,
+      "step": 29802
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.578000869528131e-07,
+      "loss": 0.0291,
+      "step": 29803
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5753590677906515e-07,
+      "loss": 0.0384,
+      "step": 29804
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5727182239425393e-07,
+      "loss": 0.0328,
+      "step": 29805
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5700783380100414e-07,
+      "loss": 0.0328,
+      "step": 29806
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5674394100193685e-07,
+      "loss": 0.0408,
+      "step": 29807
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.56480143999679e-07,
+      "loss": 0.0316,
+      "step": 29808
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5621644279684286e-07,
+      "loss": 0.0505,
+      "step": 29809
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.55952837396053e-07,
+      "loss": 0.0436,
+      "step": 29810
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5568932779992846e-07,
+      "loss": 0.0289,
+      "step": 29811
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5542591401108496e-07,
+      "loss": 0.0441,
+      "step": 29812
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.551625960321403e-07,
+      "loss": 0.0339,
+      "step": 29813
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.548993738657114e-07,
+      "loss": 0.0527,
+      "step": 29814
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.546362475144105e-07,
+      "loss": 0.0413,
+      "step": 29815
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.543732169808523e-07,
+      "loss": 0.046,
+      "step": 29816
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.541102822676512e-07,
+      "loss": 0.0397,
+      "step": 29817
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.538474433774197e-07,
+      "loss": 0.0321,
+      "step": 29818
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.535847003127657e-07,
+      "loss": 0.0439,
+      "step": 29819
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.533220530763015e-07,
+      "loss": 0.0551,
+      "step": 29820
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5305950167063617e-07,
+      "loss": 0.0396,
+      "step": 29821
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.527970460983765e-07,
+      "loss": 0.0423,
+      "step": 29822
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5253468636213374e-07,
+      "loss": 0.0395,
+      "step": 29823
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.522724224645091e-07,
+      "loss": 0.0613,
+      "step": 29824
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5201025440810943e-07,
+      "loss": 0.0404,
+      "step": 29825
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.517481821955415e-07,
+      "loss": 0.036,
+      "step": 29826
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5148620582940665e-07,
+      "loss": 0.039,
+      "step": 29827
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.512243253123082e-07,
+      "loss": 0.0413,
+      "step": 29828
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.509625406468475e-07,
+      "loss": 0.0407,
+      "step": 29829
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.507008518356236e-07,
+      "loss": 0.0621,
+      "step": 29830
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5043925888123774e-07,
+      "loss": 0.057,
+      "step": 29831
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5017776178628894e-07,
+      "loss": 0.0304,
+      "step": 29832
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4991636055337507e-07,
+      "loss": 0.0426,
+      "step": 29833
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.496550551850908e-07,
+      "loss": 0.0286,
+      "step": 29834
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.49393845684034e-07,
+      "loss": 0.046,
+      "step": 29835
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4913273205279817e-07,
+      "loss": 0.0509,
+      "step": 29836
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4887171429397903e-07,
+      "loss": 0.0385,
+      "step": 29837
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4861079241016894e-07,
+      "loss": 0.0505,
+      "step": 29838
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.48349966403958e-07,
+      "loss": 0.0393,
+      "step": 29839
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.480892362779387e-07,
+      "loss": 0.03,
+      "step": 29840
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4782860203470217e-07,
+      "loss": 0.0586,
+      "step": 29841
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.475680636768364e-07,
+      "loss": 0.045,
+      "step": 29842
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.473076212069293e-07,
+      "loss": 0.0619,
+      "step": 29843
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.470472746275699e-07,
+      "loss": 0.0464,
+      "step": 29844
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.467870239413407e-07,
+      "loss": 0.0443,
+      "step": 29845
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.465268691508317e-07,
+      "loss": 0.0411,
+      "step": 29846
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.462668102586253e-07,
+      "loss": 0.0283,
+      "step": 29847
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.460068472673028e-07,
+      "loss": 0.0314,
+      "step": 29848
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4574698017944995e-07,
+      "loss": 0.0405,
+      "step": 29849
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.454872089976469e-07,
+      "loss": 0.0294,
+      "step": 29850
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.452275337244737e-07,
+      "loss": 0.0375,
+      "step": 29851
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.449679543625117e-07,
+      "loss": 0.0397,
+      "step": 29852
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.447084709143389e-07,
+      "loss": 0.0377,
+      "step": 29853
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4444908338253093e-07,
+      "loss": 0.0419,
+      "step": 29854
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.441897917696679e-07,
+      "loss": 0.0412,
+      "step": 29855
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4393059607832236e-07,
+      "loss": 0.0237,
+      "step": 29856
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.436714963110721e-07,
+      "loss": 0.0449,
+      "step": 29857
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.434124924704885e-07,
+      "loss": 0.0382,
+      "step": 29858
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.431535845591472e-07,
+      "loss": 0.0408,
+      "step": 29859
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.428947725796183e-07,
+      "loss": 0.0516,
+      "step": 29860
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.426360565344733e-07,
+      "loss": 0.0391,
+      "step": 29861
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.423774364262833e-07,
+      "loss": 0.0483,
+      "step": 29862
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.421189122576152e-07,
+      "loss": 0.0462,
+      "step": 29863
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.418604840310391e-07,
+      "loss": 0.0281,
+      "step": 29864
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.416021517491208e-07,
+      "loss": 0.0552,
+      "step": 29865
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4134391541442826e-07,
+      "loss": 0.0349,
+      "step": 29866
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.41085775029526e-07,
+      "loss": 0.0546,
+      "step": 29867
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.408277305969787e-07,
+      "loss": 0.0401,
+      "step": 29868
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4056978211934767e-07,
+      "loss": 0.04,
+      "step": 29869
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.403119295991997e-07,
+      "loss": 0.0463,
+      "step": 29870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.400541730390927e-07,
+      "loss": 0.0315,
+      "step": 29871
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.397965124415881e-07,
+      "loss": 0.0325,
+      "step": 29872
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.39538947809247e-07,
+      "loss": 0.0456,
+      "step": 29873
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.392814791446253e-07,
+      "loss": 0.0421,
+      "step": 29874
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3902410645028197e-07,
+      "loss": 0.0295,
+      "step": 29875
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.387668297287761e-07,
+      "loss": 0.0367,
+      "step": 29876
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3850964898266005e-07,
+      "loss": 0.048,
+      "step": 29877
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.382525642144896e-07,
+      "loss": 0.0391,
+      "step": 29878
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.379955754268205e-07,
+      "loss": 0.0347,
+      "step": 29879
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.37738682622204e-07,
+      "loss": 0.0451,
+      "step": 29880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.374818858031925e-07,
+      "loss": 0.0446,
+      "step": 29881
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3722518497233735e-07,
+      "loss": 0.0367,
+      "step": 29882
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.369685801321865e-07,
+      "loss": 0.0389,
+      "step": 29883
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.367120712852912e-07,
+      "loss": 0.079,
+      "step": 29884
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.364556584342005e-07,
+      "loss": 0.0493,
+      "step": 29885
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3619934158146016e-07,
+      "loss": 0.0424,
+      "step": 29886
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.35943120729616e-07,
+      "loss": 0.0343,
+      "step": 29887
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3568699588121477e-07,
+      "loss": 0.0313,
+      "step": 29888
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3543096703879897e-07,
+      "loss": 0.035,
+      "step": 29889
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3517503420491315e-07,
+      "loss": 0.0451,
+      "step": 29890
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.349191973821031e-07,
+      "loss": 0.0317,
+      "step": 29891
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3466345657290344e-07,
+      "loss": 0.0482,
+      "step": 29892
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.344078117798577e-07,
+      "loss": 0.0383,
+      "step": 29893
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3415226300550827e-07,
+      "loss": 0.0434,
+      "step": 29894
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3389681025239095e-07,
+      "loss": 0.0302,
+      "step": 29895
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3364145352304257e-07,
+      "loss": 0.0358,
+      "step": 29896
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.333861928200033e-07,
+      "loss": 0.0413,
+      "step": 29897
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3313102814580554e-07,
+      "loss": 0.0593,
+      "step": 29898
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.328759595029851e-07,
+      "loss": 0.0584,
+      "step": 29899
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3262098689407885e-07,
+      "loss": 0.0705,
+      "step": 29900
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.323661103216147e-07,
+      "loss": 0.0614,
+      "step": 29901
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3211132978812623e-07,
+      "loss": 0.05,
+      "step": 29902
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.318566452961469e-07,
+      "loss": 0.0231,
+      "step": 29903
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.316020568482048e-07,
+      "loss": 0.0547,
+      "step": 29904
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3134756444683e-07,
+      "loss": 0.0326,
+      "step": 29905
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.310931680945495e-07,
+      "loss": 0.0454,
+      "step": 29906
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.308388677938901e-07,
+      "loss": 0.0423,
+      "step": 29907
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.305846635473786e-07,
+      "loss": 0.0491,
+      "step": 29908
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.30330555357542e-07,
+      "loss": 0.0396,
+      "step": 29909
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.300765432269026e-07,
+      "loss": 0.0584,
+      "step": 29910
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2982262715798406e-07,
+      "loss": 0.0499,
+      "step": 29911
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2956880715330986e-07,
+      "loss": 0.0529,
+      "step": 29912
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.293150832153991e-07,
+      "loss": 0.0339,
+      "step": 29913
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2906145534677634e-07,
+      "loss": 0.0567,
+      "step": 29914
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.288079235499575e-07,
+      "loss": 0.0733,
+      "step": 29915
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.285544878274616e-07,
+      "loss": 0.0472,
+      "step": 29916
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.283011481818077e-07,
+      "loss": 0.0479,
+      "step": 29917
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2804790461551274e-07,
+      "loss": 0.0383,
+      "step": 29918
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.277947571310902e-07,
+      "loss": 0.0416,
+      "step": 29919
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2754170573105706e-07,
+      "loss": 0.03,
+      "step": 29920
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.272887504179267e-07,
+      "loss": 0.0556,
+      "step": 29921
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2703589119421065e-07,
+      "loss": 0.0586,
+      "step": 29922
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.267831280624223e-07,
+      "loss": 0.033,
+      "step": 29923
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.26530461025073e-07,
+      "loss": 0.0392,
+      "step": 29924
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.262778900846708e-07,
+      "loss": 0.0356,
+      "step": 29925
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.260254152437259e-07,
+      "loss": 0.0356,
+      "step": 29926
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2577303650474623e-07,
+      "loss": 0.0742,
+      "step": 29927
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2552075387023874e-07,
+      "loss": 0.0473,
+      "step": 29928
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2526856734271137e-07,
+      "loss": 0.0358,
+      "step": 29929
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2501647692466776e-07,
+      "loss": 0.0621,
+      "step": 29930
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2476448261861136e-07,
+      "loss": 0.0409,
+      "step": 29931
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.24512584427048e-07,
+      "loss": 0.0552,
+      "step": 29932
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.242607823524768e-07,
+      "loss": 0.0389,
+      "step": 29933
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2400907639740243e-07,
+      "loss": 0.0432,
+      "step": 29934
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2375746656432284e-07,
+      "loss": 0.0257,
+      "step": 29935
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.235059528557394e-07,
+      "loss": 0.0445,
+      "step": 29936
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2325453527414896e-07,
+      "loss": 0.0511,
+      "step": 29937
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.230032138220518e-07,
+      "loss": 0.0385,
+      "step": 29938
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2275198850194147e-07,
+      "loss": 0.0408,
+      "step": 29939
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.225008593163148e-07,
+      "loss": 0.0266,
+      "step": 29940
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2224982626766767e-07,
+      "loss": 0.0481,
+      "step": 29941
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2199888935849133e-07,
+      "loss": 0.0482,
+      "step": 29942
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.217480485912805e-07,
+      "loss": 0.0439,
+      "step": 29943
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.214973039685276e-07,
+      "loss": 0.0397,
+      "step": 29944
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.212466554927229e-07,
+      "loss": 0.0449,
+      "step": 29945
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2099610316635444e-07,
+      "loss": 0.0527,
+      "step": 29946
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.207456469919146e-07,
+      "loss": 0.0381,
+      "step": 29947
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2049528697188803e-07,
+      "loss": 0.0283,
+      "step": 29948
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2024502310876615e-07,
+      "loss": 0.0386,
+      "step": 29949
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.199948554050314e-07,
+      "loss": 0.046,
+      "step": 29950
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.197447838631684e-07,
+      "loss": 0.0379,
+      "step": 29951
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1949480848566416e-07,
+      "loss": 0.0515,
+      "step": 29952
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.192449292750011e-07,
+      "loss": 0.0483,
+      "step": 29953
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1899514623366156e-07,
+      "loss": 0.0412,
+      "step": 29954
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.187454593641248e-07,
+      "loss": 0.0438,
+      "step": 29955
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.184958686688755e-07,
+      "loss": 0.0445,
+      "step": 29956
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.182463741503883e-07,
+      "loss": 0.0592,
+      "step": 29957
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1799697581114564e-07,
+      "loss": 0.0359,
+      "step": 29958
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.177476736536234e-07,
+      "loss": 0.0362,
+      "step": 29959
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1749846768029724e-07,
+      "loss": 0.0585,
+      "step": 29960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1724935789364417e-07,
+      "loss": 0.0456,
+      "step": 29961
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.170003442961389e-07,
+      "loss": 0.0587,
+      "step": 29962
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1675142689025493e-07,
+      "loss": 0.0371,
+      "step": 29963
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.165026056784637e-07,
+      "loss": 0.037,
+      "step": 29964
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.162538806632387e-07,
+      "loss": 0.0503,
+      "step": 29965
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.160052518470502e-07,
+      "loss": 0.0524,
+      "step": 29966
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1575671923236737e-07,
+      "loss": 0.033,
+      "step": 29967
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1550828282166157e-07,
+      "loss": 0.0519,
+      "step": 29968
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.152599426173975e-07,
+      "loss": 0.0399,
+      "step": 29969
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.150116986220442e-07,
+      "loss": 0.0462,
+      "step": 29970
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1476355083806753e-07,
+      "loss": 0.0487,
+      "step": 29971
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.145154992679322e-07,
+      "loss": 0.038,
+      "step": 29972
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.14267543914103e-07,
+      "loss": 0.0393,
+      "step": 29973
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1401968477904333e-07,
+      "loss": 0.0424,
+      "step": 29974
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1377192186521357e-07,
+      "loss": 0.0609,
+      "step": 29975
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1352425517507615e-07,
+      "loss": 0.044,
+      "step": 29976
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1327668471109353e-07,
+      "loss": 0.0448,
+      "step": 29977
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1302921047572153e-07,
+      "loss": 0.043,
+      "step": 29978
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1278183247141936e-07,
+      "loss": 0.0518,
+      "step": 29979
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1253455070064717e-07,
+      "loss": 0.0817,
+      "step": 29980
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1228736516585754e-07,
+      "loss": 0.043,
+      "step": 29981
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.120402758695096e-07,
+      "loss": 0.0501,
+      "step": 29982
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1179328281405685e-07,
+      "loss": 0.0292,
+      "step": 29983
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1154638600195074e-07,
+      "loss": 0.0377,
+      "step": 29984
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.11299585435646e-07,
+      "loss": 0.0335,
+      "step": 29985
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.110528811175961e-07,
+      "loss": 0.0665,
+      "step": 29986
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.108062730502481e-07,
+      "loss": 0.034,
+      "step": 29987
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1055976123605334e-07,
+      "loss": 0.0496,
+      "step": 29988
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.103133456774621e-07,
+      "loss": 0.0551,
+      "step": 29989
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.100670263769201e-07,
+      "loss": 0.0475,
+      "step": 29990
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.098208033368744e-07,
+      "loss": 0.0445,
+      "step": 29991
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.095746765597751e-07,
+      "loss": 0.035,
+      "step": 29992
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0932864604806154e-07,
+      "loss": 0.0375,
+      "step": 29993
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.090827118041795e-07,
+      "loss": 0.0279,
+      "step": 29994
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.088368738305747e-07,
+      "loss": 0.0508,
+      "step": 29995
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0859113212968637e-07,
+      "loss": 0.0329,
+      "step": 29996
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0834548670395814e-07,
+      "loss": 0.0374,
+      "step": 29997
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.08099937555828e-07,
+      "loss": 0.0667,
+      "step": 29998
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.078544846877363e-07,
+      "loss": 0.0301,
+      "step": 29999
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0760912810212094e-07,
+      "loss": 0.0649,
+      "step": 30000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0736386780142014e-07,
+      "loss": 0.0286,
+      "step": 30001
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0711870378807074e-07,
+      "loss": 0.042,
+      "step": 30002
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.068736360645064e-07,
+      "loss": 0.044,
+      "step": 30003
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0662866463316286e-07,
+      "loss": 0.0306,
+      "step": 30004
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0638378949647275e-07,
+      "loss": 0.0427,
+      "step": 30005
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0613901065686957e-07,
+      "loss": 0.0527,
+      "step": 30006
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0589432811678586e-07,
+      "loss": 0.0445,
+      "step": 30007
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.056497418786497e-07,
+      "loss": 0.0299,
+      "step": 30008
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.054052519448925e-07,
+      "loss": 0.0362,
+      "step": 30009
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.051608583179444e-07,
+      "loss": 0.0374,
+      "step": 30010
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0491656100023137e-07,
+      "loss": 0.0563,
+      "step": 30011
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.046723599941792e-07,
+      "loss": 0.0259,
+      "step": 30012
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0442825530221596e-07,
+      "loss": 0.0562,
+      "step": 30013
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.041842469267653e-07,
+      "loss": 0.0502,
+      "step": 30014
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.03940334870253e-07,
+      "loss": 0.0304,
+      "step": 30015
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0369651913510046e-07,
+      "loss": 0.0305,
+      "step": 30016
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.034527997237291e-07,
+      "loss": 0.054,
+      "step": 30017
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.032091766385614e-07,
+      "loss": 0.0491,
+      "step": 30018
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.029656498820177e-07,
+      "loss": 0.0294,
+      "step": 30019
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0272221945651603e-07,
+      "loss": 0.0321,
+      "step": 30020
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0247888536447666e-07,
+      "loss": 0.0288,
+      "step": 30021
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.022356476083155e-07,
+      "loss": 0.0478,
+      "step": 30022
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0199250619044827e-07,
+      "loss": 0.0445,
+      "step": 30023
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0174946111329205e-07,
+      "loss": 0.0396,
+      "step": 30024
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.015065123792593e-07,
+      "loss": 0.0408,
+      "step": 30025
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.012636599907648e-07,
+      "loss": 0.0402,
+      "step": 30026
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.010209039502199e-07,
+      "loss": 0.0619,
+      "step": 30027
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.007782442600382e-07,
+      "loss": 0.0573,
+      "step": 30028
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0053568092262785e-07,
+      "loss": 0.0705,
+      "step": 30029
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0029321394040134e-07,
+      "loss": 0.0296,
+      "step": 30030
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.000508433157656e-07,
+      "loss": 0.0434,
+      "step": 30031
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9980856905112763e-07,
+      "loss": 0.0486,
+      "step": 30032
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9956639114889554e-07,
+      "loss": 0.0468,
+      "step": 30033
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9932430961147395e-07,
+      "loss": 0.0309,
+      "step": 30034
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9908232444126993e-07,
+      "loss": 0.0282,
+      "step": 30035
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.988404356406838e-07,
+      "loss": 0.0375,
+      "step": 30036
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.985986432121224e-07,
+      "loss": 0.0237,
+      "step": 30037
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.98356947157985e-07,
+      "loss": 0.0586,
+      "step": 30038
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.981153474806742e-07,
+      "loss": 0.0417,
+      "step": 30039
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9787384418259016e-07,
+      "loss": 0.0357,
+      "step": 30040
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.976324372661288e-07,
+      "loss": 0.0389,
+      "step": 30041
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9739112673369263e-07,
+      "loss": 0.0618,
+      "step": 30042
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9714991258767647e-07,
+      "loss": 0.0377,
+      "step": 30043
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.969087948304761e-07,
+      "loss": 0.045,
+      "step": 30044
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.966677734644896e-07,
+      "loss": 0.0391,
+      "step": 30045
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.964268484921085e-07,
+      "loss": 0.0498,
+      "step": 30046
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.961860199157263e-07,
+      "loss": 0.0547,
+      "step": 30047
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.959452877377367e-07,
+      "loss": 0.039,
+      "step": 30048
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9570465196053e-07,
+      "loss": 0.0541,
+      "step": 30049
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9546411258649876e-07,
+      "loss": 0.0244,
+      "step": 30050
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9522366961803107e-07,
+      "loss": 0.0397,
+      "step": 30051
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9498332305751496e-07,
+      "loss": 0.0384,
+      "step": 30052
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.947430729073375e-07,
+      "loss": 0.0418,
+      "step": 30053
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.945029191698878e-07,
+      "loss": 0.0621,
+      "step": 30054
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.942628618475507e-07,
+      "loss": 0.0495,
+      "step": 30055
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9402290094270867e-07,
+      "loss": 0.0396,
+      "step": 30056
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9378303645774987e-07,
+      "loss": 0.0427,
+      "step": 30057
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.935432683950523e-07,
+      "loss": 0.0362,
+      "step": 30058
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9330359675699973e-07,
+      "loss": 0.0472,
+      "step": 30059
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9306402154597677e-07,
+      "loss": 0.0423,
+      "step": 30060
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9282454276435725e-07,
+      "loss": 0.0472,
+      "step": 30061
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9258516041452243e-07,
+      "loss": 0.0518,
+      "step": 30062
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9234587449885276e-07,
+      "loss": 0.0298,
+      "step": 30063
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9210668501972404e-07,
+      "loss": 0.0606,
+      "step": 30064
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9186759197951e-07,
+      "loss": 0.0253,
+      "step": 30065
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9162859538058864e-07,
+      "loss": 0.0458,
+      "step": 30066
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9138969522533143e-07,
+      "loss": 0.035,
+      "step": 30067
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.911508915161143e-07,
+      "loss": 0.0492,
+      "step": 30068
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9091218425531085e-07,
+      "loss": 0.0396,
+      "step": 30069
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.906735734452881e-07,
+      "loss": 0.0359,
+      "step": 30070
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.904350590884175e-07,
+      "loss": 0.0365,
+      "step": 30071
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.901966411870716e-07,
+      "loss": 0.0378,
+      "step": 30072
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.899583197436151e-07,
+      "loss": 0.0535,
+      "step": 30073
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8972009476041843e-07,
+      "loss": 0.0449,
+      "step": 30074
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.894819662398463e-07,
+      "loss": 0.0281,
+      "step": 30075
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8924393418426343e-07,
+      "loss": 0.0402,
+      "step": 30076
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.89005998596037e-07,
+      "loss": 0.0503,
+      "step": 30077
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8876815947753044e-07,
+      "loss": 0.0326,
+      "step": 30078
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8853041683110426e-07,
+      "loss": 0.0459,
+      "step": 30079
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.882927706591221e-07,
+      "loss": 0.0453,
+      "step": 30080
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.880552209639442e-07,
+      "loss": 0.042,
+      "step": 30081
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.878177677479288e-07,
+      "loss": 0.0431,
+      "step": 30082
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.875804110134384e-07,
+      "loss": 0.0501,
+      "step": 30083
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.873431507628288e-07,
+      "loss": 0.0452,
+      "step": 30084
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8710598699845494e-07,
+      "loss": 0.0427,
+      "step": 30085
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8686891972267593e-07,
+      "loss": 0.0562,
+      "step": 30086
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.866319489378466e-07,
+      "loss": 0.0311,
+      "step": 30087
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8639507464631954e-07,
+      "loss": 0.0436,
+      "step": 30088
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8615829685044837e-07,
+      "loss": 0.0351,
+      "step": 30089
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.859216155525868e-07,
+      "loss": 0.0477,
+      "step": 30090
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.856850307550818e-07,
+      "loss": 0.0282,
+      "step": 30091
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.854485424602893e-07,
+      "loss": 0.0432,
+      "step": 30092
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.852121506705552e-07,
+      "loss": 0.047,
+      "step": 30093
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.849758553882276e-07,
+      "loss": 0.0411,
+      "step": 30094
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8473965661565353e-07,
+      "loss": 0.0376,
+      "step": 30095
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8450355435518216e-07,
+      "loss": 0.0331,
+      "step": 30096
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.842675486091573e-07,
+      "loss": 0.0302,
+      "step": 30097
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.840316393799236e-07,
+      "loss": 0.0609,
+      "step": 30098
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8379582666982485e-07,
+      "loss": 0.0419,
+      "step": 30099
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8356011048120133e-07,
+      "loss": 0.057,
+      "step": 30100
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.83324490816399e-07,
+      "loss": 0.0537,
+      "step": 30101
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.830889676777537e-07,
+      "loss": 0.0455,
+      "step": 30102
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8285354106761033e-07,
+      "loss": 0.0341,
+      "step": 30103
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8261821098830247e-07,
+      "loss": 0.0483,
+      "step": 30104
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.823829774421716e-07,
+      "loss": 0.0435,
+      "step": 30105
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.821478404315525e-07,
+      "loss": 0.0381,
+      "step": 30106
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8191279995878227e-07,
+      "loss": 0.0274,
+      "step": 30107
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.816778560261957e-07,
+      "loss": 0.0403,
+      "step": 30108
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.814430086361253e-07,
+      "loss": 0.039,
+      "step": 30109
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.81208257790907e-07,
+      "loss": 0.0479,
+      "step": 30110
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8097360349286894e-07,
+      "loss": 0.0435,
+      "step": 30111
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8073904574434374e-07,
+      "loss": 0.0405,
+      "step": 30112
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8050458454766393e-07,
+      "loss": 0.0384,
+      "step": 30113
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8027021990515657e-07,
+      "loss": 0.0585,
+      "step": 30114
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8003595181914866e-07,
+      "loss": 0.0363,
+      "step": 30115
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7980178029197057e-07,
+      "loss": 0.0519,
+      "step": 30116
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.79567705325946e-07,
+      "loss": 0.0595,
+      "step": 30117
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7933372692340087e-07,
+      "loss": 0.0345,
+      "step": 30118
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7909984508666e-07,
+      "loss": 0.0389,
+      "step": 30119
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7886605981804595e-07,
+      "loss": 0.0269,
+      "step": 30120
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.786323711198813e-07,
+      "loss": 0.0459,
+      "step": 30121
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7839877899448864e-07,
+      "loss": 0.0594,
+      "step": 30122
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7816528344418836e-07,
+      "loss": 0.051,
+      "step": 30123
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.779318844712986e-07,
+      "loss": 0.0522,
+      "step": 30124
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.776985820781397e-07,
+      "loss": 0.0427,
+      "step": 30125
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7746537626702763e-07,
+      "loss": 0.0417,
+      "step": 30126
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.772322670402805e-07,
+      "loss": 0.0502,
+      "step": 30127
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.769992544002142e-07,
+      "loss": 0.0274,
+      "step": 30128
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.767663383491404e-07,
+      "loss": 0.0486,
+      "step": 30129
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7653351888937586e-07,
+      "loss": 0.0355,
+      "step": 30130
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7630079602323447e-07,
+      "loss": 0.0889,
+      "step": 30131
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.760681697530254e-07,
+      "loss": 0.0437,
+      "step": 30132
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7583564008106023e-07,
+      "loss": 0.033,
+      "step": 30133
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.756032070096504e-07,
+      "loss": 0.0685,
+      "step": 30134
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.753708705411018e-07,
+      "loss": 0.0405,
+      "step": 30135
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7513863067772596e-07,
+      "loss": 0.0631,
+      "step": 30136
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.749064874218299e-07,
+      "loss": 0.066,
+      "step": 30137
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7467444077571624e-07,
+      "loss": 0.0545,
+      "step": 30138
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7444249074169206e-07,
+      "loss": 0.0505,
+      "step": 30139
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.742106373220621e-07,
+      "loss": 0.0631,
+      "step": 30140
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.739788805191301e-07,
+      "loss": 0.0465,
+      "step": 30141
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.737472203351965e-07,
+      "loss": 0.0397,
+      "step": 30142
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.73515656772565e-07,
+      "loss": 0.0385,
+      "step": 30143
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7328418983353365e-07,
+      "loss": 0.0376,
+      "step": 30144
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7305281952040183e-07,
+      "loss": 0.0492,
+      "step": 30145
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7282154583547214e-07,
+      "loss": 0.0414,
+      "step": 30146
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7259036878103605e-07,
+      "loss": 0.0431,
+      "step": 30147
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7235928835939396e-07,
+      "loss": 0.0499,
+      "step": 30148
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.721283045728407e-07,
+      "loss": 0.0309,
+      "step": 30149
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7189741742366994e-07,
+      "loss": 0.0371,
+      "step": 30150
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7166662691417655e-07,
+      "loss": 0.0285,
+      "step": 30151
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7143593304665316e-07,
+      "loss": 0.043,
+      "step": 30152
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7120533582339016e-07,
+      "loss": 0.0526,
+      "step": 30153
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7097483524667787e-07,
+      "loss": 0.0354,
+      "step": 30154
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.707444313188101e-07,
+      "loss": 0.0235,
+      "step": 30155
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.705141240420717e-07,
+      "loss": 0.0496,
+      "step": 30156
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.702839134187518e-07,
+      "loss": 0.0538,
+      "step": 30157
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.700537994511376e-07,
+      "loss": 0.0407,
+      "step": 30158
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6982378214151504e-07,
+      "loss": 0.0376,
+      "step": 30159
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.695938614921678e-07,
+      "loss": 0.0372,
+      "step": 30160
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6936403750538407e-07,
+      "loss": 0.0312,
+      "step": 30161
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6913431018344097e-07,
+      "loss": 0.0374,
+      "step": 30162
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6890467952862433e-07,
+      "loss": 0.0327,
+      "step": 30163
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6867514554321463e-07,
+      "loss": 0.0448,
+      "step": 30164
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6844570822949224e-07,
+      "loss": 0.0512,
+      "step": 30165
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682163675897365e-07,
+      "loss": 0.0402,
+      "step": 30166
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.679871236262255e-07,
+      "loss": 0.04,
+      "step": 30167
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6775797634123524e-07,
+      "loss": 0.0385,
+      "step": 30168
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.675289257370439e-07,
+      "loss": 0.0436,
+      "step": 30169
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.672999718159275e-07,
+      "loss": 0.0308,
+      "step": 30170
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6707111458015853e-07,
+      "loss": 0.0599,
+      "step": 30171
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6684235403201085e-07,
+      "loss": 0.0481,
+      "step": 30172
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.66613690173757e-07,
+      "loss": 0.0377,
+      "step": 30173
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.663851230076697e-07,
+      "loss": 0.0461,
+      "step": 30174
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.661566525360193e-07,
+      "loss": 0.0412,
+      "step": 30175
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.65928278761074e-07,
+      "loss": 0.0435,
+      "step": 30176
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.65700001685103e-07,
+      "loss": 0.0428,
+      "step": 30177
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.654718213103746e-07,
+      "loss": 0.0566,
+      "step": 30178
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.652437376391559e-07,
+      "loss": 0.0441,
+      "step": 30179
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6501575067371276e-07,
+      "loss": 0.0514,
+      "step": 30180
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.647878604163079e-07,
+      "loss": 0.0389,
+      "step": 30181
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6456006686920833e-07,
+      "loss": 0.0376,
+      "step": 30182
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.643323700346734e-07,
+      "loss": 0.0278,
+      "step": 30183
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6410476991496904e-07,
+      "loss": 0.034,
+      "step": 30184
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6387726651235343e-07,
+      "loss": 0.0255,
+      "step": 30185
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.636498598290871e-07,
+      "loss": 0.0446,
+      "step": 30186
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6342254986742923e-07,
+      "loss": 0.0493,
+      "step": 30187
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6319533662963913e-07,
+      "loss": 0.0412,
+      "step": 30188
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6296822011797176e-07,
+      "loss": 0.0267,
+      "step": 30189
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.627412003346852e-07,
+      "loss": 0.0432,
+      "step": 30190
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.625142772820344e-07,
+      "loss": 0.0457,
+      "step": 30191
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.62287450962273e-07,
+      "loss": 0.0457,
+      "step": 30192
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.620607213776538e-07,
+      "loss": 0.0433,
+      "step": 30193
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6183408853043047e-07,
+      "loss": 0.0449,
+      "step": 30194
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.616075524228545e-07,
+      "loss": 0.0347,
+      "step": 30195
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.613811130571742e-07,
+      "loss": 0.0529,
+      "step": 30196
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.61154770435641e-07,
+      "loss": 0.0349,
+      "step": 30197
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6092852456050313e-07,
+      "loss": 0.0289,
+      "step": 30198
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6070237543400765e-07,
+      "loss": 0.0404,
+      "step": 30199
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6047632305840177e-07,
+      "loss": 0.0444,
+      "step": 30200
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6025036743593024e-07,
+      "loss": 0.0367,
+      "step": 30201
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.600245085688391e-07,
+      "loss": 0.0533,
+      "step": 30202
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5979874645936877e-07,
+      "loss": 0.0416,
+      "step": 30203
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.595730811097674e-07,
+      "loss": 0.0389,
+      "step": 30204
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.593475125222722e-07,
+      "loss": 0.0353,
+      "step": 30205
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5912204069912463e-07,
+      "loss": 0.0403,
+      "step": 30206
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5889666564256623e-07,
+      "loss": 0.0462,
+      "step": 30207
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.586713873548352e-07,
+      "loss": 0.0455,
+      "step": 30208
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.584462058381698e-07,
+      "loss": 0.0592,
+      "step": 30209
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.58221121094806e-07,
+      "loss": 0.0308,
+      "step": 30210
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.579961331269809e-07,
+      "loss": 0.0547,
+      "step": 30211
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5777124193692714e-07,
+      "loss": 0.032,
+      "step": 30212
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5754644752688187e-07,
+      "loss": 0.0363,
+      "step": 30213
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5732174989907875e-07,
+      "loss": 0.0491,
+      "step": 30214
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.570971490557461e-07,
+      "loss": 0.0342,
+      "step": 30215
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5687264499911655e-07,
+      "loss": 0.0275,
+      "step": 30216
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5664823773142277e-07,
+      "loss": 0.0638,
+      "step": 30217
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.564239272548918e-07,
+      "loss": 0.0629,
+      "step": 30218
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.56199713571752e-07,
+      "loss": 0.0523,
+      "step": 30219
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5597559668423253e-07,
+      "loss": 0.0493,
+      "step": 30220
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.557515765945562e-07,
+      "loss": 0.03,
+      "step": 30221
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.555276533049511e-07,
+      "loss": 0.0533,
+      "step": 30222
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.553038268176433e-07,
+      "loss": 0.0436,
+      "step": 30223
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5508009713485327e-07,
+      "loss": 0.0558,
+      "step": 30224
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.548564642588036e-07,
+      "loss": 0.0518,
+      "step": 30225
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5463292819171816e-07,
+      "loss": 0.0508,
+      "step": 30226
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5440948893581397e-07,
+      "loss": 0.0475,
+      "step": 30227
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5418614649331487e-07,
+      "loss": 0.0384,
+      "step": 30228
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5396290086643796e-07,
+      "loss": 0.0391,
+      "step": 30229
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.537397520574003e-07,
+      "loss": 0.0675,
+      "step": 30230
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.53516700068418e-07,
+      "loss": 0.0538,
+      "step": 30231
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5329374490170923e-07,
+      "loss": 0.0436,
+      "step": 30232
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5307088655948666e-07,
+      "loss": 0.0326,
+      "step": 30233
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5284812504396516e-07,
+      "loss": 0.0502,
+      "step": 30234
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.526254603573586e-07,
+      "loss": 0.0297,
+      "step": 30235
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.524028925018762e-07,
+      "loss": 0.0377,
+      "step": 30236
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.521804214797308e-07,
+      "loss": 0.0463,
+      "step": 30237
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.519580472931338e-07,
+      "loss": 0.0404,
+      "step": 30238
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.517357699442913e-07,
+      "loss": 0.0343,
+      "step": 30239
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.515135894354137e-07,
+      "loss": 0.0429,
+      "step": 30240
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.512915057687071e-07,
+      "loss": 0.034,
+      "step": 30241
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5106951894637745e-07,
+      "loss": 0.0408,
+      "step": 30242
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5084762897063186e-07,
+      "loss": 0.0379,
+      "step": 30243
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.506258358436731e-07,
+      "loss": 0.043,
+      "step": 30244
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.504041395677026e-07,
+      "loss": 0.0337,
+      "step": 30245
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.501825401449265e-07,
+      "loss": 0.0309,
+      "step": 30246
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4996103757754407e-07,
+      "loss": 0.0522,
+      "step": 30247
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4973963186775587e-07,
+      "loss": 0.0493,
+      "step": 30248
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4951832301776115e-07,
+      "loss": 0.0317,
+      "step": 30249
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.492971110297604e-07,
+      "loss": 0.0661,
+      "step": 30250
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4907599590594746e-07,
+      "loss": 0.0312,
+      "step": 30251
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4885497764852274e-07,
+      "loss": 0.0422,
+      "step": 30252
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4863405625968005e-07,
+      "loss": 0.0398,
+      "step": 30253
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4841323174161213e-07,
+      "loss": 0.0437,
+      "step": 30254
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4819250409651605e-07,
+      "loss": 0.0517,
+      "step": 30255
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4797187332658343e-07,
+      "loss": 0.0479,
+      "step": 30256
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4775133943400477e-07,
+      "loss": 0.0356,
+      "step": 30257
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4753090242097266e-07,
+      "loss": 0.0304,
+      "step": 30258
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.473105622896765e-07,
+      "loss": 0.0401,
+      "step": 30259
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4709031904230354e-07,
+      "loss": 0.047,
+      "step": 30260
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4687017268104406e-07,
+      "loss": 0.0564,
+      "step": 30261
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4665012320808314e-07,
+      "loss": 0.0227,
+      "step": 30262
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4643017062560673e-07,
+      "loss": 0.0633,
+      "step": 30263
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4621031493580195e-07,
+      "loss": 0.0557,
+      "step": 30264
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.459905561408515e-07,
+      "loss": 0.0354,
+      "step": 30265
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.457708942429371e-07,
+      "loss": 0.045,
+      "step": 30266
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.455513292442435e-07,
+      "loss": 0.0776,
+      "step": 30267
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.453318611469513e-07,
+      "loss": 0.0338,
+      "step": 30268
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.451124899532387e-07,
+      "loss": 0.0308,
+      "step": 30269
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4489321566528724e-07,
+      "loss": 0.0408,
+      "step": 30270
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4467403828527415e-07,
+      "loss": 0.0475,
+      "step": 30271
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4445495781537766e-07,
+      "loss": 0.0349,
+      "step": 30272
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.442359742577716e-07,
+      "loss": 0.0287,
+      "step": 30273
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4401708761463416e-07,
+      "loss": 0.0484,
+      "step": 30274
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4379829788813815e-07,
+      "loss": 0.0308,
+      "step": 30275
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.435796050804595e-07,
+      "loss": 0.0577,
+      "step": 30276
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.433610091937677e-07,
+      "loss": 0.04,
+      "step": 30277
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.431425102302343e-07,
+      "loss": 0.0398,
+      "step": 30278
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4292410819203304e-07,
+      "loss": 0.0418,
+      "step": 30279
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.427058030813301e-07,
+      "loss": 0.0376,
+      "step": 30280
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.424875949002958e-07,
+      "loss": 0.0473,
+      "step": 30281
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.422694836510986e-07,
+      "loss": 0.0394,
+      "step": 30282
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.420514693359044e-07,
+      "loss": 0.0547,
+      "step": 30283
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.418335519568771e-07,
+      "loss": 0.0408,
+      "step": 30284
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.416157315161849e-07,
+      "loss": 0.0313,
+      "step": 30285
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4139800801598945e-07,
+      "loss": 0.035,
+      "step": 30286
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4118038145845346e-07,
+      "loss": 0.0452,
+      "step": 30287
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4096285184574077e-07,
+      "loss": 0.0451,
+      "step": 30288
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.407454191800107e-07,
+      "loss": 0.0253,
+      "step": 30289
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.405280834634227e-07,
+      "loss": 0.0425,
+      "step": 30290
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.403108446981384e-07,
+      "loss": 0.0228,
+      "step": 30291
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4009370288631374e-07,
+      "loss": 0.0415,
+      "step": 30292
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.398766580301048e-07,
+      "loss": 0.0512,
+      "step": 30293
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.396597101316711e-07,
+      "loss": 0.0312,
+      "step": 30294
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3944285919316525e-07,
+      "loss": 0.043,
+      "step": 30295
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.392261052167422e-07,
+      "loss": 0.0386,
+      "step": 30296
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3900944820455573e-07,
+      "loss": 0.0605,
+      "step": 30297
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3879288815875645e-07,
+      "loss": 0.0226,
+      "step": 30298
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.38576425081497e-07,
+      "loss": 0.0272,
+      "step": 30299
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.383600589749291e-07,
+      "loss": 0.0534,
+      "step": 30300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.381437898411998e-07,
+      "loss": 0.0743,
+      "step": 30301
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3792761768245744e-07,
+      "loss": 0.0432,
+      "step": 30302
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3771154250085139e-07,
+      "loss": 0.0415,
+      "step": 30303
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.374955642985255e-07,
+      "loss": 0.076,
+      "step": 30304
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3727968307762806e-07,
+      "loss": 0.0515,
+      "step": 30305
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3706389884030289e-07,
+      "loss": 0.0368,
+      "step": 30306
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3684821158869274e-07,
+      "loss": 0.0517,
+      "step": 30307
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3663262132494037e-07,
+      "loss": 0.0675,
+      "step": 30308
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3641712805118954e-07,
+      "loss": 0.0304,
+      "step": 30309
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.362017317695775e-07,
+      "loss": 0.038,
+      "step": 30310
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3598643248224694e-07,
+      "loss": 0.0418,
+      "step": 30311
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3577123019133507e-07,
+      "loss": 0.0598,
+      "step": 30312
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3555612489898018e-07,
+      "loss": 0.047,
+      "step": 30313
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.353411166073183e-07,
+      "loss": 0.0451,
+      "step": 30314
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3512620531848885e-07,
+      "loss": 0.0578,
+      "step": 30315
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3491139103462125e-07,
+      "loss": 0.0559,
+      "step": 30316
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3469667375785378e-07,
+      "loss": 0.0368,
+      "step": 30317
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3448205349031805e-07,
+      "loss": 0.0742,
+      "step": 30318
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.342675302341446e-07,
+      "loss": 0.0501,
+      "step": 30319
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.340531039914673e-07,
+      "loss": 0.059,
+      "step": 30320
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3383877476441553e-07,
+      "loss": 0.0479,
+      "step": 30321
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3362454255511647e-07,
+      "loss": 0.0287,
+      "step": 30322
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3341040736570064e-07,
+      "loss": 0.0576,
+      "step": 30323
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3319636919829413e-07,
+      "loss": 0.0544,
+      "step": 30324
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3298242805502414e-07,
+      "loss": 0.037,
+      "step": 30325
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3276858393801448e-07,
+      "loss": 0.0332,
+      "step": 30326
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.325548368493913e-07,
+      "loss": 0.0556,
+      "step": 30327
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3234118679127615e-07,
+      "loss": 0.0384,
+      "step": 30328
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3212763376579184e-07,
+      "loss": 0.0541,
+      "step": 30329
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.319141777750622e-07,
+      "loss": 0.0497,
+      "step": 30330
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3170081882120333e-07,
+      "loss": 0.0524,
+      "step": 30331
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3148755690633795e-07,
+      "loss": 0.0529,
+      "step": 30332
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.312743920325855e-07,
+      "loss": 0.0395,
+      "step": 30333
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3106132420205983e-07,
+      "loss": 0.0299,
+      "step": 30334
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.308483534168815e-07,
+      "loss": 0.0386,
+      "step": 30335
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.306354796791632e-07,
+      "loss": 0.0354,
+      "step": 30336
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3042270299102109e-07,
+      "loss": 0.0421,
+      "step": 30337
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3021002335456788e-07,
+      "loss": 0.0604,
+      "step": 30338
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2999744077191854e-07,
+      "loss": 0.0461,
+      "step": 30339
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2978495524518362e-07,
+      "loss": 0.0476,
+      "step": 30340
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2957256677647256e-07,
+      "loss": 0.0658,
+      "step": 30341
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2936027536789696e-07,
+      "loss": 0.0304,
+      "step": 30342
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2914808102156404e-07,
+      "loss": 0.0303,
+      "step": 30343
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2893598373958547e-07,
+      "loss": 0.0296,
+      "step": 30344
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.28723983524064e-07,
+      "loss": 0.0523,
+      "step": 30345
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2851208037710794e-07,
+      "loss": 0.0407,
+      "step": 30346
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2830027430082114e-07,
+      "loss": 0.0443,
+      "step": 30347
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.280885652973086e-07,
+      "loss": 0.0343,
+      "step": 30348
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2787695336867422e-07,
+      "loss": 0.0553,
+      "step": 30349
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.276654385170174e-07,
+      "loss": 0.0395,
+      "step": 30350
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.274540207444431e-07,
+      "loss": 0.0308,
+      "step": 30351
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.272427000530475e-07,
+      "loss": 0.037,
+      "step": 30352
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2703147644493327e-07,
+      "loss": 0.0345,
+      "step": 30353
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.268203499221977e-07,
+      "loss": 0.04,
+      "step": 30354
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2660932048693686e-07,
+      "loss": 0.0493,
+      "step": 30355
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2639838814124792e-07,
+      "loss": 0.0393,
+      "step": 30356
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2618755288722705e-07,
+      "loss": 0.0441,
+      "step": 30357
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2597681472696698e-07,
+      "loss": 0.0422,
+      "step": 30358
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.257661736625638e-07,
+      "loss": 0.039,
+      "step": 30359
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2555562969610921e-07,
+      "loss": 0.0396,
+      "step": 30360
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2534518282969266e-07,
+      "loss": 0.0529,
+      "step": 30361
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2513483306540684e-07,
+      "loss": 0.0604,
+      "step": 30362
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2492458040534015e-07,
+      "loss": 0.0392,
+      "step": 30363
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.24714424851582e-07,
+      "loss": 0.0445,
+      "step": 30364
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2450436640622076e-07,
+      "loss": 0.049,
+      "step": 30365
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2429440507134137e-07,
+      "loss": 0.0358,
+      "step": 30366
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2408454084903108e-07,
+      "loss": 0.0444,
+      "step": 30367
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2387477374137377e-07,
+      "loss": 0.0632,
+      "step": 30368
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.236651037504556e-07,
+      "loss": 0.0431,
+      "step": 30369
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.234555308783548e-07,
+      "loss": 0.043,
+      "step": 30370
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.232460551271576e-07,
+      "loss": 0.0396,
+      "step": 30371
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2303667649894333e-07,
+      "loss": 0.0568,
+      "step": 30372
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2282739499579153e-07,
+      "loss": 0.0508,
+      "step": 30373
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2261821061978273e-07,
+      "loss": 0.0419,
+      "step": 30374
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2240912337299304e-07,
+      "loss": 0.058,
+      "step": 30375
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2220013325749968e-07,
+      "loss": 0.0387,
+      "step": 30376
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.21991240275381e-07,
+      "loss": 0.0424,
+      "step": 30377
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.217824444287109e-07,
+      "loss": 0.0505,
+      "step": 30378
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2157374571956214e-07,
+      "loss": 0.0361,
+      "step": 30379
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2136514415001088e-07,
+      "loss": 0.0374,
+      "step": 30380
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2115663972212543e-07,
+      "loss": 0.0285,
+      "step": 30381
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.209482324379808e-07,
+      "loss": 0.0337,
+      "step": 30382
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2073992229964647e-07,
+      "loss": 0.0374,
+      "step": 30383
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2053170930919078e-07,
+      "loss": 0.0433,
+      "step": 30384
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2032359346868203e-07,
+      "loss": 0.0592,
+      "step": 30385
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2011557478018863e-07,
+      "loss": 0.0354,
+      "step": 30386
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1990765324577555e-07,
+      "loss": 0.049,
+      "step": 30387
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1969982886751119e-07,
+      "loss": 0.0387,
+      "step": 30388
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1949210164745826e-07,
+      "loss": 0.0369,
+      "step": 30389
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.192844715876785e-07,
+      "loss": 0.0448,
+      "step": 30390
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1907693869023694e-07,
+      "loss": 0.0445,
+      "step": 30391
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1886950295719633e-07,
+      "loss": 0.049,
+      "step": 30392
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1866216439061506e-07,
+      "loss": 0.0483,
+      "step": 30393
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1845492299255256e-07,
+      "loss": 0.0359,
+      "step": 30394
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.182477787650694e-07,
+      "loss": 0.0697,
+      "step": 30395
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1804073171022176e-07,
+      "loss": 0.0523,
+      "step": 30396
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1783378183006687e-07,
+      "loss": 0.0504,
+      "step": 30397
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1762692912666306e-07,
+      "loss": 0.0415,
+      "step": 30398
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1742017360206092e-07,
+      "loss": 0.0359,
+      "step": 30399
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1721351525831658e-07,
+      "loss": 0.0461,
+      "step": 30400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1700695409748285e-07,
+      "loss": 0.0368,
+      "step": 30401
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1680049012161364e-07,
+      "loss": 0.0444,
+      "step": 30402
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.165941233327562e-07,
+      "loss": 0.0311,
+      "step": 30403
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1638785373296333e-07,
+      "loss": 0.0776,
+      "step": 30404
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.161816813242823e-07,
+      "loss": 0.0412,
+      "step": 30405
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1597560610876255e-07,
+      "loss": 0.0415,
+      "step": 30406
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.157696280884536e-07,
+      "loss": 0.0429,
+      "step": 30407
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1556374726539598e-07,
+      "loss": 0.0306,
+      "step": 30408
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.153579636416392e-07,
+      "loss": 0.0363,
+      "step": 30409
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1515227721922604e-07,
+      "loss": 0.0527,
+      "step": 30410
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1494668800019935e-07,
+      "loss": 0.0473,
+      "step": 30411
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1474119598660304e-07,
+      "loss": 0.0399,
+      "step": 30412
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1453580118047768e-07,
+      "loss": 0.0664,
+      "step": 30413
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1433050358386387e-07,
+      "loss": 0.0635,
+      "step": 30414
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1412530319879888e-07,
+      "loss": 0.0491,
+      "step": 30415
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.139202000273255e-07,
+      "loss": 0.0311,
+      "step": 30416
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1371519407147766e-07,
+      "loss": 0.0573,
+      "step": 30417
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1351028533329265e-07,
+      "loss": 0.0395,
+      "step": 30418
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.133054738148077e-07,
+      "loss": 0.0573,
+      "step": 30419
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1310075951805564e-07,
+      "loss": 0.0298,
+      "step": 30420
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.128961424450704e-07,
+      "loss": 0.0377,
+      "step": 30421
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.126916225978859e-07,
+      "loss": 0.0535,
+      "step": 30422
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1248719997853163e-07,
+      "loss": 0.0464,
+      "step": 30423
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.122828745890393e-07,
+      "loss": 0.0354,
+      "step": 30424
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1207864643144061e-07,
+      "loss": 0.0254,
+      "step": 30425
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1187451550776173e-07,
+      "loss": 0.0323,
+      "step": 30426
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1167048182003214e-07,
+      "loss": 0.0215,
+      "step": 30427
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.11466545370278e-07,
+      "loss": 0.0437,
+      "step": 30428
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1126270616052436e-07,
+      "loss": 0.0373,
+      "step": 30429
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1105896419279848e-07,
+      "loss": 0.0372,
+      "step": 30430
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.108553194691232e-07,
+      "loss": 0.0287,
+      "step": 30431
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1065177199152021e-07,
+      "loss": 0.0217,
+      "step": 30432
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1044832176201237e-07,
+      "loss": 0.0292,
+      "step": 30433
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1024496878262245e-07,
+      "loss": 0.0716,
+      "step": 30434
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.100417130553678e-07,
+      "loss": 0.0378,
+      "step": 30435
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.098385545822701e-07,
+      "loss": 0.0308,
+      "step": 30436
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0963549336534773e-07,
+      "loss": 0.0398,
+      "step": 30437
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0943252940661462e-07,
+      "loss": 0.0324,
+      "step": 30438
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0922966270809031e-07,
+      "loss": 0.0447,
+      "step": 30439
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.090268932717876e-07,
+      "loss": 0.0396,
+      "step": 30440
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0882422109972267e-07,
+      "loss": 0.0355,
+      "step": 30441
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0862164619390835e-07,
+      "loss": 0.0534,
+      "step": 30442
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0841916855635745e-07,
+      "loss": 0.0167,
+      "step": 30443
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0821678818907953e-07,
+      "loss": 0.0427,
+      "step": 30444
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.080145050940874e-07,
+      "loss": 0.0477,
+      "step": 30445
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0781231927339053e-07,
+      "loss": 0.0555,
+      "step": 30446
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.076102307289951e-07,
+      "loss": 0.0359,
+      "step": 30447
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0740823946291067e-07,
+      "loss": 0.0637,
+      "step": 30448
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0720634547714225e-07,
+      "loss": 0.0329,
+      "step": 30449
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0700454877369715e-07,
+      "loss": 0.0442,
+      "step": 30450
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0680284935457928e-07,
+      "loss": 0.0707,
+      "step": 30451
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0660124722179376e-07,
+      "loss": 0.0469,
+      "step": 30452
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0639974237734006e-07,
+      "loss": 0.0365,
+      "step": 30453
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0619833482322327e-07,
+      "loss": 0.0452,
+      "step": 30454
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0599702456144178e-07,
+      "loss": 0.0392,
+      "step": 30455
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.057958115939962e-07,
+      "loss": 0.0387,
+      "step": 30456
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0559469592288717e-07,
+      "loss": 0.0325,
+      "step": 30457
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0539367755010863e-07,
+      "loss": 0.0584,
+      "step": 30458
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0519275647766125e-07,
+      "loss": 0.0387,
+      "step": 30459
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0499193270754002e-07,
+      "loss": 0.0668,
+      "step": 30460
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0479120624173897e-07,
+      "loss": 0.0483,
+      "step": 30461
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0459057708225316e-07,
+      "loss": 0.0585,
+      "step": 30462
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0439004523107543e-07,
+      "loss": 0.0579,
+      "step": 30463
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0418961069019637e-07,
+      "loss": 0.0435,
+      "step": 30464
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0398927346161e-07,
+      "loss": 0.029,
+      "step": 30465
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0378903354730363e-07,
+      "loss": 0.0236,
+      "step": 30466
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0358889094926782e-07,
+      "loss": 0.045,
+      "step": 30467
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0338884566949102e-07,
+      "loss": 0.0348,
+      "step": 30468
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0318889770996052e-07,
+      "loss": 0.0479,
+      "step": 30469
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0298904707266254e-07,
+      "loss": 0.0524,
+      "step": 30470
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.02789293759581e-07,
+      "loss": 0.0569,
+      "step": 30471
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0258963777270212e-07,
+      "loss": 0.0421,
+      "step": 30472
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0239007911400767e-07,
+      "loss": 0.0577,
+      "step": 30473
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0219061778548155e-07,
+      "loss": 0.0453,
+      "step": 30474
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0199125378910666e-07,
+      "loss": 0.0358,
+      "step": 30475
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0179198712685922e-07,
+      "loss": 0.0415,
+      "step": 30476
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0159281780072094e-07,
+      "loss": 0.0452,
+      "step": 30477
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0139374581267135e-07,
+      "loss": 0.0432,
+      "step": 30478
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0119477116468667e-07,
+      "loss": 0.0587,
+      "step": 30479
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0099589385874417e-07,
+      "loss": 0.0537,
+      "step": 30480
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.007971138968201e-07,
+      "loss": 0.0464,
+      "step": 30481
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.005984312808873e-07,
+      "loss": 0.0271,
+      "step": 30482
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0039984601292084e-07,
+      "loss": 0.0331,
+      "step": 30483
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0020135809489472e-07,
+      "loss": 0.0568,
+      "step": 30484
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0000296752877734e-07,
+      "loss": 0.0396,
+      "step": 30485
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.998046743165416e-07,
+      "loss": 0.0614,
+      "step": 30486
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9960647846015923e-07,
+      "loss": 0.0439,
+      "step": 30487
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9940837996159535e-07,
+      "loss": 0.0456,
+      "step": 30488
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9921037882282058e-07,
+      "loss": 0.037,
+      "step": 30489
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.990124750458e-07,
+      "loss": 0.0259,
+      "step": 30490
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9881466863250098e-07,
+      "loss": 0.055,
+      "step": 30491
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9861695958488747e-07,
+      "loss": 0.0417,
+      "step": 30492
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9841934790492568e-07,
+      "loss": 0.0437,
+      "step": 30493
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9822183359457625e-07,
+      "loss": 0.0421,
+      "step": 30494
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9802441665580208e-07,
+      "loss": 0.0458,
+      "step": 30495
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9782709709056492e-07,
+      "loss": 0.0472,
+      "step": 30496
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.976298749008243e-07,
+      "loss": 0.0273,
+      "step": 30497
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9743275008853978e-07,
+      "loss": 0.0388,
+      "step": 30498
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.972357226556698e-07,
+      "loss": 0.0374,
+      "step": 30499
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9703879260417058e-07,
+      "loss": 0.0423,
+      "step": 30500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.968419599359994e-07,
+      "loss": 0.0521,
+      "step": 30501
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9664522465311253e-07,
+      "loss": 0.0351,
+      "step": 30502
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.964485867574617e-07,
+      "loss": 0.0415,
+      "step": 30503
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9625204625100314e-07,
+      "loss": 0.0332,
+      "step": 30504
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.960556031356875e-07,
+      "loss": 0.0374,
+      "step": 30505
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9585925741346657e-07,
+      "loss": 0.0349,
+      "step": 30506
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9566300908629098e-07,
+      "loss": 0.0617,
+      "step": 30507
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9546685815611032e-07,
+      "loss": 0.0519,
+      "step": 30508
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9527080462487413e-07,
+      "loss": 0.0403,
+      "step": 30509
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.950748484945275e-07,
+      "loss": 0.0515,
+      "step": 30510
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9487898976702e-07,
+      "loss": 0.033,
+      "step": 30511
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9468322844429565e-07,
+      "loss": 0.0638,
+      "step": 30512
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9448756452829953e-07,
+      "loss": 0.0415,
+      "step": 30513
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9429199802097566e-07,
+      "loss": 0.0395,
+      "step": 30514
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9409652892426578e-07,
+      "loss": 0.0481,
+      "step": 30515
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.939011572401117e-07,
+      "loss": 0.0322,
+      "step": 30516
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9370588297045634e-07,
+      "loss": 0.0393,
+      "step": 30517
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.935107061172381e-07,
+      "loss": 0.0511,
+      "step": 30518
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9331562668239545e-07,
+      "loss": 0.0273,
+      "step": 30519
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9312064466786796e-07,
+      "loss": 0.0464,
+      "step": 30520
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.929257600755896e-07,
+      "loss": 0.0551,
+      "step": 30521
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9273097290749997e-07,
+      "loss": 0.0358,
+      "step": 30522
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9253628316553196e-07,
+      "loss": 0.0494,
+      "step": 30523
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9234169085161847e-07,
+      "loss": 0.0519,
+      "step": 30524
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.041,
+      "step": 30525
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9195279851569436e-07,
+      "loss": 0.0353,
+      "step": 30526
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9175849849754403e-07,
+      "loss": 0.0358,
+      "step": 30527
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9156429591517756e-07,
+      "loss": 0.0557,
+      "step": 30528
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9137019077052343e-07,
+      "loss": 0.0451,
+      "step": 30529
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.91176183065509e-07,
+      "loss": 0.0295,
+      "step": 30530
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.909822728020616e-07,
+      "loss": 0.0389,
+      "step": 30531
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.907884599821075e-07,
+      "loss": 0.0511,
+      "step": 30532
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.90594744607574e-07,
+      "loss": 0.0334,
+      "step": 30533
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9040112668038291e-07,
+      "loss": 0.0416,
+      "step": 30534
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9020760620245938e-07,
+      "loss": 0.0457,
+      "step": 30535
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9001418317572408e-07,
+      "loss": 0.0603,
+      "step": 30536
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.89820857602101e-07,
+      "loss": 0.0491,
+      "step": 30537
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8962762948350865e-07,
+      "loss": 0.0526,
+      "step": 30538
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.894344988218666e-07,
+      "loss": 0.044,
+      "step": 30539
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.892414656190944e-07,
+      "loss": 0.0426,
+      "step": 30540
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8904852987710942e-07,
+      "loss": 0.0382,
+      "step": 30541
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.888556915978279e-07,
+      "loss": 0.0298,
+      "step": 30542
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.886629507831661e-07,
+      "loss": 0.0393,
+      "step": 30543
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8847030743503802e-07,
+      "loss": 0.0457,
+      "step": 30544
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.882777615553577e-07,
+      "loss": 0.0381,
+      "step": 30545
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8808531314603807e-07,
+      "loss": 0.0455,
+      "step": 30546
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.878929622089909e-07,
+      "loss": 0.0362,
+      "step": 30547
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8770070874612689e-07,
+      "loss": 0.0421,
+      "step": 30548
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8750855275935564e-07,
+      "loss": 0.0435,
+      "step": 30549
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8731649425058673e-07,
+      "loss": 0.0421,
+      "step": 30550
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8712453322172641e-07,
+      "loss": 0.0403,
+      "step": 30551
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8693266967468537e-07,
+      "loss": 0.0528,
+      "step": 30552
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8674090361136544e-07,
+      "loss": 0.0576,
+      "step": 30553
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8654923503367283e-07,
+      "loss": 0.0243,
+      "step": 30554
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8635766394351384e-07,
+      "loss": 0.0557,
+      "step": 30555
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8616619034278804e-07,
+      "loss": 0.0463,
+      "step": 30556
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8597481423340057e-07,
+      "loss": 0.0399,
+      "step": 30557
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8578353561725104e-07,
+      "loss": 0.0352,
+      "step": 30558
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8559235449623902e-07,
+      "loss": 0.0343,
+      "step": 30559
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8540127087226522e-07,
+      "loss": 0.049,
+      "step": 30560
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8521028474722814e-07,
+      "loss": 0.0343,
+      "step": 30561
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8501939612302288e-07,
+      "loss": 0.041,
+      "step": 30562
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8482860500154686e-07,
+      "loss": 0.0443,
+      "step": 30563
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8463791138469633e-07,
+      "loss": 0.0387,
+      "step": 30564
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8444731527436532e-07,
+      "loss": 0.0377,
+      "step": 30565
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8425681667244567e-07,
+      "loss": 0.0286,
+      "step": 30566
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.840664155808325e-07,
+      "loss": 0.0465,
+      "step": 30567
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8387611200141431e-07,
+      "loss": 0.0412,
+      "step": 30568
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8368590593608292e-07,
+      "loss": 0.0495,
+      "step": 30569
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8349579738672795e-07,
+      "loss": 0.0282,
+      "step": 30570
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8330578635523898e-07,
+      "loss": 0.0552,
+      "step": 30571
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8311587284350119e-07,
+      "loss": 0.0362,
+      "step": 30572
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8292605685340414e-07,
+      "loss": 0.0388,
+      "step": 30573
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.827363383868297e-07,
+      "loss": 0.056,
+      "step": 30574
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.825467174456652e-07,
+      "loss": 0.0443,
+      "step": 30575
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8235719403179586e-07,
+      "loss": 0.0388,
+      "step": 30576
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8216776814710125e-07,
+      "loss": 0.045,
+      "step": 30577
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.819784397934643e-07,
+      "loss": 0.0391,
+      "step": 30578
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8178920897276575e-07,
+      "loss": 0.0263,
+      "step": 30579
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8160007568688521e-07,
+      "loss": 0.0422,
+      "step": 30580
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8141103993770226e-07,
+      "loss": 0.0331,
+      "step": 30581
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.812221017270943e-07,
+      "loss": 0.0718,
+      "step": 30582
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8103326105693763e-07,
+      "loss": 0.0331,
+      "step": 30583
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8084451792910962e-07,
+      "loss": 0.0413,
+      "step": 30584
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8065587234548543e-07,
+      "loss": 0.042,
+      "step": 30585
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.804673243079369e-07,
+      "loss": 0.0403,
+      "step": 30586
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8027887381833808e-07,
+      "loss": 0.0456,
+      "step": 30587
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8009052087856303e-07,
+      "loss": 0.0611,
+      "step": 30588
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7990226549047917e-07,
+      "loss": 0.0508,
+      "step": 30589
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7971410765596053e-07,
+      "loss": 0.037,
+      "step": 30590
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7952604737687341e-07,
+      "loss": 0.0469,
+      "step": 30591
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.793380846550874e-07,
+      "loss": 0.058,
+      "step": 30592
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.791502194924688e-07,
+      "loss": 0.0297,
+      "step": 30593
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7896245189088501e-07,
+      "loss": 0.0344,
+      "step": 30594
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.787747818522001e-07,
+      "loss": 0.032,
+      "step": 30595
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7858720937827922e-07,
+      "loss": 0.0515,
+      "step": 30596
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7839973447098647e-07,
+      "loss": 0.0497,
+      "step": 30597
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7821235713218143e-07,
+      "loss": 0.035,
+      "step": 30598
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.780250773637293e-07,
+      "loss": 0.0475,
+      "step": 30599
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.778378951674886e-07,
+      "loss": 0.0428,
+      "step": 30600
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7765081054531786e-07,
+      "loss": 0.0528,
+      "step": 30601
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7746382349907554e-07,
+      "loss": 0.0467,
+      "step": 30602
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7727693403062242e-07,
+      "loss": 0.0578,
+      "step": 30603
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.770901421418114e-07,
+      "loss": 0.0627,
+      "step": 30604
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7690344783450108e-07,
+      "loss": 0.0297,
+      "step": 30605
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7671685111054327e-07,
+      "loss": 0.0257,
+      "step": 30606
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7653035197179313e-07,
+      "loss": 0.0315,
+      "step": 30607
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7634395042010366e-07,
+      "loss": 0.0504,
+      "step": 30608
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7615764645732558e-07,
+      "loss": 0.0356,
+      "step": 30609
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7597144008531074e-07,
+      "loss": 0.0535,
+      "step": 30610
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.757853313059077e-07,
+      "loss": 0.0512,
+      "step": 30611
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7559932012096602e-07,
+      "loss": 0.0322,
+      "step": 30612
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7541340653233318e-07,
+      "loss": 0.0382,
+      "step": 30613
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7522759054185768e-07,
+      "loss": 0.042,
+      "step": 30614
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7504187215138358e-07,
+      "loss": 0.0444,
+      "step": 30615
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7485625136275496e-07,
+      "loss": 0.0608,
+      "step": 30616
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7467072817781928e-07,
+      "loss": 0.0359,
+      "step": 30617
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7448530259841501e-07,
+      "loss": 0.0463,
+      "step": 30618
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7429997462638738e-07,
+      "loss": 0.0678,
+      "step": 30619
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7411474426357822e-07,
+      "loss": 0.0393,
+      "step": 30620
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7392961151182497e-07,
+      "loss": 0.0454,
+      "step": 30621
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7374457637296726e-07,
+      "loss": 0.068,
+      "step": 30622
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7355963884884473e-07,
+      "loss": 0.0672,
+      "step": 30623
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7337479894129372e-07,
+      "loss": 0.0414,
+      "step": 30624
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7319005665214938e-07,
+      "loss": 0.0351,
+      "step": 30625
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7300541198324915e-07,
+      "loss": 0.0567,
+      "step": 30626
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.728208649364249e-07,
+      "loss": 0.0442,
+      "step": 30627
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7263641551351185e-07,
+      "loss": 0.0362,
+      "step": 30628
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7245206371634183e-07,
+      "loss": 0.0457,
+      "step": 30629
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.722678095467456e-07,
+      "loss": 0.0447,
+      "step": 30630
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7208365300655395e-07,
+      "loss": 0.049,
+      "step": 30631
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.718995940975965e-07,
+      "loss": 0.0451,
+      "step": 30632
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7171563282170178e-07,
+      "loss": 0.0434,
+      "step": 30633
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7153176918069615e-07,
+      "loss": 0.0376,
+      "step": 30634
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.713480031764081e-07,
+      "loss": 0.036,
+      "step": 30635
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7116433481066064e-07,
+      "loss": 0.0439,
+      "step": 30636
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.709807640852812e-07,
+      "loss": 0.0536,
+      "step": 30637
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7079729100209163e-07,
+      "loss": 0.0352,
+      "step": 30638
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7061391556291497e-07,
+      "loss": 0.0275,
+      "step": 30639
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7043063776957192e-07,
+      "loss": 0.0623,
+      "step": 30640
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7024745762388439e-07,
+      "loss": 0.0493,
+      "step": 30641
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7006437512767094e-07,
+      "loss": 0.0289,
+      "step": 30642
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.698813902827523e-07,
+      "loss": 0.048,
+      "step": 30643
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6969850309094594e-07,
+      "loss": 0.0636,
+      "step": 30644
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6951571355406594e-07,
+      "loss": 0.0467,
+      "step": 30645
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6933302167393085e-07,
+      "loss": 0.0347,
+      "step": 30646
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6915042745235476e-07,
+      "loss": 0.0361,
+      "step": 30647
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6896793089115182e-07,
+      "loss": 0.03,
+      "step": 30648
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.68785531992135e-07,
+      "loss": 0.0448,
+      "step": 30649
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6860323075711503e-07,
+      "loss": 0.0318,
+      "step": 30650
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6842102718790387e-07,
+      "loss": 0.0565,
+      "step": 30651
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6823892128631226e-07,
+      "loss": 0.0496,
+      "step": 30652
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6805691305414873e-07,
+      "loss": 0.0295,
+      "step": 30653
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6787500249322074e-07,
+      "loss": 0.0485,
+      "step": 30654
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6769318960533465e-07,
+      "loss": 0.0402,
+      "step": 30655
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.67511474392299e-07,
+      "loss": 0.0388,
+      "step": 30656
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6732985685591675e-07,
+      "loss": 0.0622,
+      "step": 30657
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6714833699799427e-07,
+      "loss": 0.0347,
+      "step": 30658
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6696691482033346e-07,
+      "loss": 0.0458,
+      "step": 30659
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6678559032473617e-07,
+      "loss": 0.0398,
+      "step": 30660
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6660436351300325e-07,
+      "loss": 0.0543,
+      "step": 30661
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6642323438693763e-07,
+      "loss": 0.0425,
+      "step": 30662
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6624220294833682e-07,
+      "loss": 0.0442,
+      "step": 30663
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6606126919899934e-07,
+      "loss": 0.0375,
+      "step": 30664
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6588043314072268e-07,
+      "loss": 0.048,
+      "step": 30665
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.65699694775302e-07,
+      "loss": 0.0395,
+      "step": 30666
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6551905410453485e-07,
+      "loss": 0.0462,
+      "step": 30667
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6533851113021638e-07,
+      "loss": 0.0462,
+      "step": 30668
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.651580658541363e-07,
+      "loss": 0.0532,
+      "step": 30669
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6497771827808983e-07,
+      "loss": 0.0353,
+      "step": 30670
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6479746840387001e-07,
+      "loss": 0.0581,
+      "step": 30671
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6461731623326428e-07,
+      "loss": 0.0491,
+      "step": 30672
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6443726176806451e-07,
+      "loss": 0.0505,
+      "step": 30673
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.642573050100582e-07,
+      "loss": 0.0462,
+      "step": 30674
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.640774459610328e-07,
+      "loss": 0.0226,
+      "step": 30675
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6389768462277578e-07,
+      "loss": 0.0384,
+      "step": 30676
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6371802099707346e-07,
+      "loss": 0.0437,
+      "step": 30677
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6353845508570997e-07,
+      "loss": 0.0384,
+      "step": 30678
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6335898689046838e-07,
+      "loss": 0.0465,
+      "step": 30679
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6317961641313274e-07,
+      "loss": 0.0338,
+      "step": 30680
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.630003436554839e-07,
+      "loss": 0.0429,
+      "step": 30681
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6282116861930375e-07,
+      "loss": 0.0498,
+      "step": 30682
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.626420913063731e-07,
+      "loss": 0.0469,
+      "step": 30683
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6246311171846719e-07,
+      "loss": 0.0344,
+      "step": 30684
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.622842298573668e-07,
+      "loss": 0.0431,
+      "step": 30685
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6210544572485054e-07,
+      "loss": 0.0384,
+      "step": 30686
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.619267593226914e-07,
+      "loss": 0.0228,
+      "step": 30687
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6174817065266468e-07,
+      "loss": 0.0437,
+      "step": 30688
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6156967971654669e-07,
+      "loss": 0.0526,
+      "step": 30689
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6139128651610935e-07,
+      "loss": 0.0386,
+      "step": 30690
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.612129910531246e-07,
+      "loss": 0.0495,
+      "step": 30691
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6103479332936323e-07,
+      "loss": 0.0324,
+      "step": 30692
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6085669334659603e-07,
+      "loss": 0.0214,
+      "step": 30693
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6067869110659274e-07,
+      "loss": 0.0369,
+      "step": 30694
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6050078661112078e-07,
+      "loss": 0.0522,
+      "step": 30695
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6032297986194766e-07,
+      "loss": 0.0361,
+      "step": 30696
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6014527086084086e-07,
+      "loss": 0.045,
+      "step": 30697
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.599676596095645e-07,
+      "loss": 0.037,
+      "step": 30698
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5979014610988276e-07,
+      "loss": 0.0419,
+      "step": 30699
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5961273036355974e-07,
+      "loss": 0.0431,
+      "step": 30700
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.594354123723574e-07,
+      "loss": 0.0388,
+      "step": 30701
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5925819213803873e-07,
+      "loss": 0.0348,
+      "step": 30702
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5908106966236347e-07,
+      "loss": 0.054,
+      "step": 30703
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5890404494708912e-07,
+      "loss": 0.0408,
+      "step": 30704
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5872711799397645e-07,
+      "loss": 0.0748,
+      "step": 30705
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5855028880478296e-07,
+      "loss": 0.0753,
+      "step": 30706
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5837355738126503e-07,
+      "loss": 0.0567,
+      "step": 30707
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5819692372517792e-07,
+      "loss": 0.038,
+      "step": 30708
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5802038783827689e-07,
+      "loss": 0.0366,
+      "step": 30709
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.57843949722315e-07,
+      "loss": 0.0291,
+      "step": 30710
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5766760937904525e-07,
+      "loss": 0.045,
+      "step": 30711
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.574913668102207e-07,
+      "loss": 0.0371,
+      "step": 30712
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5731522201758998e-07,
+      "loss": 0.0497,
+      "step": 30713
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5713917500290277e-07,
+      "loss": 0.0385,
+      "step": 30714
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5696322576791101e-07,
+      "loss": 0.0396,
+      "step": 30715
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5678737431435997e-07,
+      "loss": 0.0327,
+      "step": 30716
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5661162064399716e-07,
+      "loss": 0.0326,
+      "step": 30717
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5643596475856892e-07,
+      "loss": 0.0379,
+      "step": 30718
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5626040665981946e-07,
+      "loss": 0.0601,
+      "step": 30719
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.56084946349494e-07,
+      "loss": 0.0382,
+      "step": 30720
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5590958382933564e-07,
+      "loss": 0.0395,
+      "step": 30721
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5573431910108404e-07,
+      "loss": 0.0342,
+      "step": 30722
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.555591521664823e-07,
+      "loss": 0.0455,
+      "step": 30723
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5538408302727014e-07,
+      "loss": 0.0352,
+      "step": 30724
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5520911168518616e-07,
+      "loss": 0.0389,
+      "step": 30725
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5503423814197117e-07,
+      "loss": 0.0471,
+      "step": 30726
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5485946239935934e-07,
+      "loss": 0.0344,
+      "step": 30727
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.546847844590871e-07,
+      "loss": 0.0419,
+      "step": 30728
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5451020432288965e-07,
+      "loss": 0.0555,
+      "step": 30729
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5433572199250456e-07,
+      "loss": 0.0339,
+      "step": 30730
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5416133746966156e-07,
+      "loss": 0.0341,
+      "step": 30731
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5398705075609365e-07,
+      "loss": 0.0356,
+      "step": 30732
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5381286185353395e-07,
+      "loss": 0.0516,
+      "step": 30733
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5363877076370992e-07,
+      "loss": 0.0566,
+      "step": 30734
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5346477748835353e-07,
+      "loss": 0.0534,
+      "step": 30735
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.532908820291934e-07,
+      "loss": 0.0272,
+      "step": 30736
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.531170843879548e-07,
+      "loss": 0.0428,
+      "step": 30737
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5294338456636526e-07,
+      "loss": 0.0415,
+      "step": 30738
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5276978256615117e-07,
+      "loss": 0.0432,
+      "step": 30739
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5259627838903668e-07,
+      "loss": 0.0342,
+      "step": 30740
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5242287203674487e-07,
+      "loss": 0.0474,
+      "step": 30741
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5224956351099884e-07,
+      "loss": 0.0464,
+      "step": 30742
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.520763528135194e-07,
+      "loss": 0.0374,
+      "step": 30743
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5190323994602852e-07,
+      "loss": 0.0595,
+      "step": 30744
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5173022491024703e-07,
+      "loss": 0.0329,
+      "step": 30745
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.515573077078891e-07,
+      "loss": 0.0305,
+      "step": 30746
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5138448834067675e-07,
+      "loss": 0.0353,
+      "step": 30747
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5121176681032523e-07,
+      "loss": 0.0473,
+      "step": 30748
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5103914311854983e-07,
+      "loss": 0.0301,
+      "step": 30749
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.508666172670681e-07,
+      "loss": 0.0326,
+      "step": 30750
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5069418925759083e-07,
+      "loss": 0.0329,
+      "step": 30751
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5052185909183115e-07,
+      "loss": 0.0431,
+      "step": 30752
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.503496267715021e-07,
+      "loss": 0.0509,
+      "step": 30753
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5017749229831568e-07,
+      "loss": 0.0426,
+      "step": 30754
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.500054556739805e-07,
+      "loss": 0.0508,
+      "step": 30755
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.498335169002041e-07,
+      "loss": 0.0507,
+      "step": 30756
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4966167597869841e-07,
+      "loss": 0.0283,
+      "step": 30757
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4948993291116654e-07,
+      "loss": 0.0446,
+      "step": 30758
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4931828769931712e-07,
+      "loss": 0.0328,
+      "step": 30759
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4914674034485542e-07,
+      "loss": 0.0452,
+      "step": 30760
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4897529084948237e-07,
+      "loss": 0.0359,
+      "step": 30761
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4880393921490433e-07,
+      "loss": 0.0295,
+      "step": 30762
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4863268544282327e-07,
+      "loss": 0.0512,
+      "step": 30763
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.484615295349401e-07,
+      "loss": 0.0425,
+      "step": 30764
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4829047149295452e-07,
+      "loss": 0.0449,
+      "step": 30765
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.481195113185663e-07,
+      "loss": 0.0522,
+      "step": 30766
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4794864901347295e-07,
+      "loss": 0.0395,
+      "step": 30767
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4777788457937315e-07,
+      "loss": 0.0417,
+      "step": 30768
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4760721801796218e-07,
+      "loss": 0.0254,
+      "step": 30769
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4743664933093537e-07,
+      "loss": 0.0543,
+      "step": 30770
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4726617851998805e-07,
+      "loss": 0.0396,
+      "step": 30771
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4709580558681436e-07,
+      "loss": 0.0466,
+      "step": 30772
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.469255305331041e-07,
+      "loss": 0.0428,
+      "step": 30773
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4675535336055257e-07,
+      "loss": 0.0645,
+      "step": 30774
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4658527407084732e-07,
+      "loss": 0.0381,
+      "step": 30775
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4641529266567812e-07,
+      "loss": 0.0506,
+      "step": 30776
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.462454091467347e-07,
+      "loss": 0.027,
+      "step": 30777
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.460756235157035e-07,
+      "loss": 0.0488,
+      "step": 30778
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4590593577427315e-07,
+      "loss": 0.0392,
+      "step": 30779
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4573634592412788e-07,
+      "loss": 0.0386,
+      "step": 30780
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4556685396695303e-07,
+      "loss": 0.052,
+      "step": 30781
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4539745990443056e-07,
+      "loss": 0.0483,
+      "step": 30782
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4522816373824577e-07,
+      "loss": 0.049,
+      "step": 30783
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4505896547007958e-07,
+      "loss": 0.0338,
+      "step": 30784
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4488986510161175e-07,
+      "loss": 0.0412,
+      "step": 30785
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4472086263452423e-07,
+      "loss": 0.0439,
+      "step": 30786
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4455195807049349e-07,
+      "loss": 0.0557,
+      "step": 30787
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4438315141119817e-07,
+      "loss": 0.0292,
+      "step": 30788
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4421444265831696e-07,
+      "loss": 0.0464,
+      "step": 30789
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4404583181352405e-07,
+      "loss": 0.0383,
+      "step": 30790
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4387731887849477e-07,
+      "loss": 0.0344,
+      "step": 30791
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4370890385490442e-07,
+      "loss": 0.0586,
+      "step": 30792
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4354058674442394e-07,
+      "loss": 0.0607,
+      "step": 30793
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.433723675487253e-07,
+      "loss": 0.0384,
+      "step": 30794
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.432042462694827e-07,
+      "loss": 0.0405,
+      "step": 30795
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.430362229083626e-07,
+      "loss": 0.0308,
+      "step": 30796
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4286829746703589e-07,
+      "loss": 0.0481,
+      "step": 30797
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4270046994717123e-07,
+      "loss": 0.0564,
+      "step": 30798
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4253274035043617e-07,
+      "loss": 0.045,
+      "step": 30799
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4236510867849384e-07,
+      "loss": 0.0407,
+      "step": 30800
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.42197574933014e-07,
+      "loss": 0.0269,
+      "step": 30801
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4203013911565643e-07,
+      "loss": 0.0388,
+      "step": 30802
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.418628012280887e-07,
+      "loss": 0.0396,
+      "step": 30803
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.416955612719695e-07,
+      "loss": 0.0424,
+      "step": 30804
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.41528419248963e-07,
+      "loss": 0.0568,
+      "step": 30805
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4136137516072678e-07,
+      "loss": 0.0364,
+      "step": 30806
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.41194429008924e-07,
+      "loss": 0.0557,
+      "step": 30807
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4102758079520996e-07,
+      "loss": 0.0428,
+      "step": 30808
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.408608305212422e-07,
+      "loss": 0.0495,
+      "step": 30809
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.406941781886806e-07,
+      "loss": 0.0403,
+      "step": 30810
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4052762379917594e-07,
+      "loss": 0.0678,
+      "step": 30811
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4036116735438587e-07,
+      "loss": 0.0524,
+      "step": 30812
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.401948088559646e-07,
+      "loss": 0.035,
+      "step": 30813
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4002854830556189e-07,
+      "loss": 0.0367,
+      "step": 30814
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.39862385704832e-07,
+      "loss": 0.0408,
+      "step": 30815
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3969632105542362e-07,
+      "loss": 0.0296,
+      "step": 30816
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3953035435898764e-07,
+      "loss": 0.0338,
+      "step": 30817
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3936448561717276e-07,
+      "loss": 0.0504,
+      "step": 30818
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.391987148316265e-07,
+      "loss": 0.0484,
+      "step": 30819
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3903304200399537e-07,
+      "loss": 0.0304,
+      "step": 30820
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.388674671359258e-07,
+      "loss": 0.0282,
+      "step": 30821
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3870199022906204e-07,
+      "loss": 0.0369,
+      "step": 30822
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3853661128504835e-07,
+      "loss": 0.0298,
+      "step": 30823
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3837133030552674e-07,
+      "loss": 0.0547,
+      "step": 30824
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3820614729214033e-07,
+      "loss": 0.0455,
+      "step": 30825
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.380410622465289e-07,
+      "loss": 0.0529,
+      "step": 30826
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3787607517033342e-07,
+      "loss": 0.0532,
+      "step": 30827
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3771118606519363e-07,
+      "loss": 0.0335,
+      "step": 30828
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.375463949327449e-07,
+      "loss": 0.0311,
+      "step": 30829
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3738170177462596e-07,
+      "loss": 0.0522,
+      "step": 30830
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3721710659247323e-07,
+      "loss": 0.0475,
+      "step": 30831
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.37052609387921e-07,
+      "loss": 0.0462,
+      "step": 30832
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3688821016260345e-07,
+      "loss": 0.0434,
+      "step": 30833
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.367239089181549e-07,
+      "loss": 0.0576,
+      "step": 30834
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3655970565620623e-07,
+      "loss": 0.0461,
+      "step": 30835
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.363956003783884e-07,
+      "loss": 0.0434,
+      "step": 30836
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.362315930863334e-07,
+      "loss": 0.0441,
+      "step": 30837
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.360676837816688e-07,
+      "loss": 0.0422,
+      "step": 30838
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3590387246602332e-07,
+      "loss": 0.0389,
+      "step": 30839
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3574015914102457e-07,
+      "loss": 0.0371,
+      "step": 30840
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3557654380829787e-07,
+      "loss": 0.0525,
+      "step": 30841
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3541302646947085e-07,
+      "loss": 0.0428,
+      "step": 30842
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3524960712616664e-07,
+      "loss": 0.0389,
+      "step": 30843
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3508628578000725e-07,
+      "loss": 0.0438,
+      "step": 30844
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.349230624326159e-07,
+      "loss": 0.0338,
+      "step": 30845
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3475993708561675e-07,
+      "loss": 0.0216,
+      "step": 30846
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3459690974062744e-07,
+      "loss": 0.0522,
+      "step": 30847
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.344339803992667e-07,
+      "loss": 0.0516,
+      "step": 30848
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3427114906315653e-07,
+      "loss": 0.0506,
+      "step": 30849
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3410841573391008e-07,
+      "loss": 0.0483,
+      "step": 30850
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.339457804131483e-07,
+      "loss": 0.0626,
+      "step": 30851
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3378324310248546e-07,
+      "loss": 0.04,
+      "step": 30852
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.336208038035336e-07,
+      "loss": 0.0388,
+      "step": 30853
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3345846251791029e-07,
+      "loss": 0.032,
+      "step": 30854
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3329621924722536e-07,
+      "loss": 0.0468,
+      "step": 30855
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3313407399309198e-07,
+      "loss": 0.0493,
+      "step": 30856
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3297202675712106e-07,
+      "loss": 0.0329,
+      "step": 30857
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3281007754092135e-07,
+      "loss": 0.045,
+      "step": 30858
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3264822634610154e-07,
+      "loss": 0.0515,
+      "step": 30859
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3248647317427143e-07,
+      "loss": 0.0395,
+      "step": 30860
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3232481802703646e-07,
+      "loss": 0.0365,
+      "step": 30861
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3216326090600196e-07,
+      "loss": 0.0449,
+      "step": 30862
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3200180181277334e-07,
+      "loss": 0.0356,
+      "step": 30863
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3184044074895485e-07,
+      "loss": 0.0283,
+      "step": 30864
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3167917771614969e-07,
+      "loss": 0.0359,
+      "step": 30865
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3151801271595988e-07,
+      "loss": 0.0424,
+      "step": 30866
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3135694574998525e-07,
+      "loss": 0.0446,
+      "step": 30867
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3119597681982678e-07,
+      "loss": 0.0366,
+      "step": 30868
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3103510592708425e-07,
+      "loss": 0.0431,
+      "step": 30869
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3087433307335417e-07,
+      "loss": 0.0447,
+      "step": 30870
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.307136582602353e-07,
+      "loss": 0.0644,
+      "step": 30871
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3055308148932299e-07,
+      "loss": 0.0664,
+      "step": 30872
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3039260276221155e-07,
+      "loss": 0.0528,
+      "step": 30873
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3023222208049636e-07,
+      "loss": 0.0443,
+      "step": 30874
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3007193944577056e-07,
+      "loss": 0.0396,
+      "step": 30875
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2991175485962625e-07,
+      "loss": 0.0216,
+      "step": 30876
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2975166832365327e-07,
+      "loss": 0.0455,
+      "step": 30877
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2959167983944476e-07,
+      "loss": 0.0441,
+      "step": 30878
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2943178940858836e-07,
+      "loss": 0.036,
+      "step": 30879
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.292719970326728e-07,
+      "loss": 0.0548,
+      "step": 30880
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2911230271328569e-07,
+      "loss": 0.0445,
+      "step": 30881
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.289527064520113e-07,
+      "loss": 0.0465,
+      "step": 30882
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.287932082504384e-07,
+      "loss": 0.0431,
+      "step": 30883
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2863380811015013e-07,
+      "loss": 0.0396,
+      "step": 30884
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.284745060327297e-07,
+      "loss": 0.0363,
+      "step": 30885
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2831530201975805e-07,
+      "loss": 0.0558,
+      "step": 30886
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2815619607282058e-07,
+      "loss": 0.0475,
+      "step": 30887
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2799718819349382e-07,
+      "loss": 0.0449,
+      "step": 30888
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2783827838335982e-07,
+      "loss": 0.0426,
+      "step": 30889
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2767946664399845e-07,
+      "loss": 0.0406,
+      "step": 30890
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2752075297698286e-07,
+      "loss": 0.0353,
+      "step": 30891
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2736213738389293e-07,
+      "loss": 0.0412,
+      "step": 30892
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2720361986630514e-07,
+      "loss": 0.0501,
+      "step": 30893
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.270452004257916e-07,
+      "loss": 0.0413,
+      "step": 30894
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2688687906392772e-07,
+      "loss": 0.0357,
+      "step": 30895
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2672865578228666e-07,
+      "loss": 0.0544,
+      "step": 30896
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.265705305824383e-07,
+      "loss": 0.0414,
+      "step": 30897
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2641250346595357e-07,
+      "loss": 0.0681,
+      "step": 30898
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.262545744344057e-07,
+      "loss": 0.044,
+      "step": 30899
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2609674348936007e-07,
+      "loss": 0.0385,
+      "step": 30900
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2593901063238544e-07,
+      "loss": 0.0263,
+      "step": 30901
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2578137586505057e-07,
+      "loss": 0.0367,
+      "step": 30902
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2562383918891864e-07,
+      "loss": 0.045,
+      "step": 30903
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2546640060555504e-07,
+      "loss": 0.0267,
+      "step": 30904
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2530906011652744e-07,
+      "loss": 0.0565,
+      "step": 30905
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2515181772339346e-07,
+      "loss": 0.0399,
+      "step": 30906
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2499467342771854e-07,
+      "loss": 0.044,
+      "step": 30907
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.248376272310625e-07,
+      "loss": 0.0536,
+      "step": 30908
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.246806791349875e-07,
+      "loss": 0.0456,
+      "step": 30909
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2452382914104888e-07,
+      "loss": 0.0365,
+      "step": 30910
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2436707725080876e-07,
+      "loss": 0.0454,
+      "step": 30911
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2421042346582145e-07,
+      "loss": 0.0421,
+      "step": 30912
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.240538677876446e-07,
+      "loss": 0.0363,
+      "step": 30913
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2389741021783474e-07,
+      "loss": 0.0573,
+      "step": 30914
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2374105075794175e-07,
+      "loss": 0.0424,
+      "step": 30915
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2358478940952323e-07,
+      "loss": 0.0397,
+      "step": 30916
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.234286261741302e-07,
+      "loss": 0.0455,
+      "step": 30917
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2327256105331254e-07,
+      "loss": 0.0349,
+      "step": 30918
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.0597,
+      "step": 30919
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.229607251616094e-07,
+      "loss": 0.0422,
+      "step": 30920
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2280495439382033e-07,
+      "loss": 0.046,
+      "step": 30921
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2264928174680279e-07,
+      "loss": 0.0393,
+      "step": 30922
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2249370722210552e-07,
+      "loss": 0.0441,
+      "step": 30923
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2233823082127172e-07,
+      "loss": 0.0373,
+      "step": 30924
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.221828525458457e-07,
+      "loss": 0.0526,
+      "step": 30925
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2202757239737296e-07,
+      "loss": 0.0471,
+      "step": 30926
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.218723903773944e-07,
+      "loss": 0.0609,
+      "step": 30927
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2171730648745216e-07,
+      "loss": 0.0411,
+      "step": 30928
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2156232072908725e-07,
+      "loss": 0.0319,
+      "step": 30929
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.214074331038384e-07,
+      "loss": 0.0618,
+      "step": 30930
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.212526436132455e-07,
+      "loss": 0.0456,
+      "step": 30931
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2109795225884512e-07,
+      "loss": 0.0605,
+      "step": 30932
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.209433590421738e-07,
+      "loss": 0.0272,
+      "step": 30933
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.207888639647692e-07,
+      "loss": 0.0424,
+      "step": 30934
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.206344670281645e-07,
+      "loss": 0.0322,
+      "step": 30935
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2048016823389296e-07,
+      "loss": 0.0758,
+      "step": 30936
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2032596758348892e-07,
+      "loss": 0.054,
+      "step": 30937
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2017186507848333e-07,
+      "loss": 0.0484,
+      "step": 30938
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.200178607204061e-07,
+      "loss": 0.0439,
+      "step": 30939
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1986395451078936e-07,
+      "loss": 0.0532,
+      "step": 30940
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1971014645116075e-07,
+      "loss": 0.044,
+      "step": 30941
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1955643654304795e-07,
+      "loss": 0.0505,
+      "step": 30942
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.194028247879786e-07,
+      "loss": 0.0443,
+      "step": 30943
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1924931118747928e-07,
+      "loss": 0.0382,
+      "step": 30944
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1909589574307323e-07,
+      "loss": 0.0456,
+      "step": 30945
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1894257845628588e-07,
+      "loss": 0.0378,
+      "step": 30946
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1878935932863822e-07,
+      "loss": 0.0447,
+      "step": 30947
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1863623836165572e-07,
+      "loss": 0.0403,
+      "step": 30948
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1848321555685716e-07,
+      "loss": 0.0553,
+      "step": 30949
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1833029091576354e-07,
+      "loss": 0.0378,
+      "step": 30950
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1817746443989365e-07,
+      "loss": 0.0298,
+      "step": 30951
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1802473613076515e-07,
+      "loss": 0.0387,
+      "step": 30952
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1787210598989684e-07,
+      "loss": 0.0658,
+      "step": 30953
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1771957401880307e-07,
+      "loss": 0.0298,
+      "step": 30954
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.175671402190004e-07,
+      "loss": 0.0337,
+      "step": 30955
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1741480459200316e-07,
+      "loss": 0.0459,
+      "step": 30956
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1726256713932349e-07,
+      "loss": 0.0362,
+      "step": 30957
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1711042786247462e-07,
+      "loss": 0.021,
+      "step": 30958
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1695838676296866e-07,
+      "loss": 0.0436,
+      "step": 30959
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1680644384231332e-07,
+      "loss": 0.0296,
+      "step": 30960
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1665459910202182e-07,
+      "loss": 0.0326,
+      "step": 30961
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1650285254359961e-07,
+      "loss": 0.0423,
+      "step": 30962
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1635120416855439e-07,
+      "loss": 0.0551,
+      "step": 30963
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1619965397839494e-07,
+      "loss": 0.0566,
+      "step": 30964
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1604820197462341e-07,
+      "loss": 0.0313,
+      "step": 30965
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1589684815874746e-07,
+      "loss": 0.0521,
+      "step": 30966
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1574559253226924e-07,
+      "loss": 0.0286,
+      "step": 30967
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1559443509669089e-07,
+      "loss": 0.0463,
+      "step": 30968
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.154433758535145e-07,
+      "loss": 0.0367,
+      "step": 30969
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1529241480424114e-07,
+      "loss": 0.0442,
+      "step": 30970
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1514155195036958e-07,
+      "loss": 0.0349,
+      "step": 30971
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1499078729339974e-07,
+      "loss": 0.0233,
+      "step": 30972
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1484012083482821e-07,
+      "loss": 0.0554,
+      "step": 30973
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1468955257615155e-07,
+      "loss": 0.044,
+      "step": 30974
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1453908251886636e-07,
+      "loss": 0.04,
+      "step": 30975
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1438871066446811e-07,
+      "loss": 0.0629,
+      "step": 30976
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1423843701444892e-07,
+      "loss": 0.0446,
+      "step": 30977
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1408826157030095e-07,
+      "loss": 0.0515,
+      "step": 30978
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1393818433351966e-07,
+      "loss": 0.0396,
+      "step": 30979
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1378820530559165e-07,
+      "loss": 0.0505,
+      "step": 30980
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1363832448801016e-07,
+      "loss": 0.0537,
+      "step": 30981
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1348854188226288e-07,
+      "loss": 0.0292,
+      "step": 30982
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1333885748983753e-07,
+      "loss": 0.0327,
+      "step": 30983
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1318927131222068e-07,
+      "loss": 0.0682,
+      "step": 30984
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1303978335089894e-07,
+      "loss": 0.0428,
+      "step": 30985
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1289039360735776e-07,
+      "loss": 0.0399,
+      "step": 30986
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1274110208308042e-07,
+      "loss": 0.0401,
+      "step": 30987
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1259190877955128e-07,
+      "loss": 0.0496,
+      "step": 30988
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1244281369825028e-07,
+      "loss": 0.0592,
+      "step": 30989
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1229381684065955e-07,
+      "loss": 0.0717,
+      "step": 30990
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1214491820826123e-07,
+      "loss": 0.0479,
+      "step": 30991
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1199611780253084e-07,
+      "loss": 0.0457,
+      "step": 30992
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1184741562494827e-07,
+      "loss": 0.0262,
+      "step": 30993
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1169881167699126e-07,
+      "loss": 0.0704,
+      "step": 30994
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1155030596013638e-07,
+      "loss": 0.0448,
+      "step": 30995
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1140189847585802e-07,
+      "loss": 0.0448,
+      "step": 30996
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1125358922563056e-07,
+      "loss": 0.047,
+      "step": 30997
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1110537821092616e-07,
+      "loss": 0.0533,
+      "step": 30998
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1095726543321916e-07,
+      "loss": 0.0632,
+      "step": 30999
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1080925089398065e-07,
+      "loss": 0.0367,
+      "step": 31000
+    }
+  ],
+  "max_steps": 32495,
+  "num_train_epochs": 5,
+  "total_flos": 9.86796653543424e+18,
+  "trial_name": null,
+  "trial_params": null
+}