diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,37951 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 6321,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473685e-07,
+      "loss": 1.6338,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-07,
+      "loss": 1.6562,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1578947368421055e-07,
+      "loss": 1.6938,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.210526315789474e-07,
+      "loss": 1.5347,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 1.5889,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.315789473684211e-07,
+      "loss": 1.5151,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.368421052631579e-07,
+      "loss": 1.6006,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.421052631578948e-07,
+      "loss": 1.4551,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.473684210526317e-07,
+      "loss": 1.499,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.6455,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1578947368421053e-06,
+      "loss": 1.5332,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2631578947368422e-06,
+      "loss": 1.501,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3684210526315791e-06,
+      "loss": 1.5273,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4736842105263159e-06,
+      "loss": 1.4268,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 1.4648,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6842105263157895e-06,
+      "loss": 1.373,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7894736842105265e-06,
+      "loss": 1.3176,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8947368421052634e-06,
+      "loss": 1.2231,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.3201,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.1602,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2105263157894738e-06,
+      "loss": 1.1265,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3157894736842105e-06,
+      "loss": 1.1633,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4210526315789477e-06,
+      "loss": 1.136,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5263157894736844e-06,
+      "loss": 1.2195,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 1.0999,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7368421052631583e-06,
+      "loss": 1.1316,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.842105263157895e-06,
+      "loss": 1.0481,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9473684210526317e-06,
+      "loss": 1.0964,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.052631578947369e-06,
+      "loss": 1.0063,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.04,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2631578947368423e-06,
+      "loss": 1.0618,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.368421052631579e-06,
+      "loss": 1.0781,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.473684210526316e-06,
+      "loss": 1.1125,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.578947368421053e-06,
+      "loss": 1.0112,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 1.0356,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.789473684210527e-06,
+      "loss": 0.4022,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.894736842105263e-06,
+      "loss": 1.0195,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.4171,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.105263157894737e-06,
+      "loss": 1.0437,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.9595,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.315789473684211e-06,
+      "loss": 0.9731,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4210526315789476e-06,
+      "loss": 0.9434,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.526315789473685e-06,
+      "loss": 0.9761,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.631578947368421e-06,
+      "loss": 0.9719,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 1.0146,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.842105263157895e-06,
+      "loss": 0.9639,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947368421052632e-06,
+      "loss": 0.4208,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 0.8945,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.157894736842106e-06,
+      "loss": 0.9619,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.9065,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.36842105263158e-06,
+      "loss": 0.9424,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4736842105263165e-06,
+      "loss": 0.9263,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.578947368421052e-06,
+      "loss": 0.927,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 0.8857,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.925,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8947368421052634e-06,
+      "loss": 0.8389,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.9006,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.105263157894738e-06,
+      "loss": 0.9648,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2105263157894745e-06,
+      "loss": 0.9294,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.9016,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.421052631578948e-06,
+      "loss": 0.9224,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.526315789473685e-06,
+      "loss": 0.9563,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.631578947368421e-06,
+      "loss": 0.9321,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.736842105263158e-06,
+      "loss": 0.925,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.8647,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.947368421052632e-06,
+      "loss": 0.9258,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.052631578947369e-06,
+      "loss": 0.9341,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.157894736842106e-06,
+      "loss": 0.8833,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.263157894736843e-06,
+      "loss": 0.8987,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.8953,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.473684210526316e-06,
+      "loss": 0.3749,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.578947368421054e-06,
+      "loss": 0.8933,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.68421052631579e-06,
+      "loss": 0.8994,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.789473684210526e-06,
+      "loss": 0.4019,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.9065,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8771,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.105263157894736e-06,
+      "loss": 0.9048,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 0.875,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.315789473684212e-06,
+      "loss": 0.8374,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.8669,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.526315789473685e-06,
+      "loss": 0.34,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.631578947368422e-06,
+      "loss": 0.9331,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.736842105263158e-06,
+      "loss": 0.8364,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.842105263157895e-06,
+      "loss": 0.8535,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.8396,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.05263157894737e-06,
+      "loss": 0.8584,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.157894736842105e-06,
+      "loss": 0.8914,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.263157894736842e-06,
+      "loss": 0.9382,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.36842105263158e-06,
+      "loss": 0.8926,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.9128,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.578947368421054e-06,
+      "loss": 0.8567,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.68421052631579e-06,
+      "loss": 0.8474,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.789473684210527e-06,
+      "loss": 0.8977,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.894736842105264e-06,
+      "loss": 0.8616,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.8748,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 0.8911,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0210526315789476e-05,
+      "loss": 0.8618,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0315789473684213e-05,
+      "loss": 0.8442,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0421052631578948e-05,
+      "loss": 0.9104,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.9404,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0631578947368421e-05,
+      "loss": 0.8411,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.073684210526316e-05,
+      "loss": 0.8423,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0842105263157896e-05,
+      "loss": 0.9055,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0947368421052633e-05,
+      "loss": 0.9001,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.105263157894737e-05,
+      "loss": 0.8855,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1157894736842105e-05,
+      "loss": 0.8171,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1263157894736843e-05,
+      "loss": 0.843,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 0.8247,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1473684210526317e-05,
+      "loss": 0.8694,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.8577,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1684210526315792e-05,
+      "loss": 0.8945,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1789473684210527e-05,
+      "loss": 0.8523,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1894736842105264e-05,
+      "loss": 0.803,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8562,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 0.9028,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2210526315789475e-05,
+      "loss": 0.884,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2315789473684212e-05,
+      "loss": 0.8767,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2421052631578949e-05,
+      "loss": 0.8638,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2526315789473684e-05,
+      "loss": 0.8684,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.8467,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2736842105263159e-05,
+      "loss": 0.8721,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2842105263157896e-05,
+      "loss": 0.8369,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2947368421052633e-05,
+      "loss": 0.3404,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.305263157894737e-05,
+      "loss": 0.8979,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3157894736842108e-05,
+      "loss": 0.8599,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 0.8596,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.336842105263158e-05,
+      "loss": 0.8821,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3473684210526316e-05,
+      "loss": 0.8894,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3578947368421055e-05,
+      "loss": 0.7644,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.8794,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3789473684210528e-05,
+      "loss": 0.781,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3894736842105265e-05,
+      "loss": 0.818,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 0.8672,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4105263157894738e-05,
+      "loss": 0.8195,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 0.8879,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4315789473684212e-05,
+      "loss": 0.8767,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4421052631578948e-05,
+      "loss": 0.8452,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4526315789473687e-05,
+      "loss": 0.8108,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4631578947368424e-05,
+      "loss": 0.8823,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.8826,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4842105263157895e-05,
+      "loss": 0.785,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4947368421052632e-05,
+      "loss": 0.885,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.505263157894737e-05,
+      "loss": 0.8787,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5157894736842107e-05,
+      "loss": 0.8582,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5263157894736846e-05,
+      "loss": 0.8958,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536842105263158e-05,
+      "loss": 0.8595,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5473684210526316e-05,
+      "loss": 0.8452,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5578947368421052e-05,
+      "loss": 0.832,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.568421052631579e-05,
+      "loss": 0.8105,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.7502,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5894736842105266e-05,
+      "loss": 0.851,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.8264,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6105263157894736e-05,
+      "loss": 0.8218,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6210526315789473e-05,
+      "loss": 0.8835,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 0.8218,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.642105263157895e-05,
+      "loss": 0.8894,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6526315789473686e-05,
+      "loss": 0.7953,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6631578947368423e-05,
+      "loss": 0.811,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673684210526316e-05,
+      "loss": 0.9004,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.8687,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6947368421052633e-05,
+      "loss": 0.8582,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.705263157894737e-05,
+      "loss": 0.3281,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7157894736842107e-05,
+      "loss": 0.8152,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7263157894736843e-05,
+      "loss": 0.9023,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 0.8254,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7473684210526317e-05,
+      "loss": 0.8232,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7578947368421054e-05,
+      "loss": 0.8455,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.768421052631579e-05,
+      "loss": 0.8635,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7789473684210527e-05,
+      "loss": 0.8835,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.8414,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 0.8831,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.810526315789474e-05,
+      "loss": 0.844,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8210526315789477e-05,
+      "loss": 0.8108,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.831578947368421e-05,
+      "loss": 0.876,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.2955,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8526315789473684e-05,
+      "loss": 0.8391,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8631578947368424e-05,
+      "loss": 0.8782,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.873684210526316e-05,
+      "loss": 0.9197,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8842105263157898e-05,
+      "loss": 0.8459,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.7957,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9052631578947368e-05,
+      "loss": 0.8257,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9157894736842108e-05,
+      "loss": 0.8143,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9263157894736845e-05,
+      "loss": 0.8521,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.936842105263158e-05,
+      "loss": 0.8286,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9473684210526318e-05,
+      "loss": 0.8755,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9578947368421055e-05,
+      "loss": 0.8433,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.968421052631579e-05,
+      "loss": 0.7854,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9789473684210528e-05,
+      "loss": 0.8435,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9894736842105265e-05,
+      "loss": 0.824,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9229,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998687174187e-05,
+      "loss": 0.9075,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994748697086e-05,
+      "loss": 0.8372,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999988184569735e-05,
+      "loss": 0.8628,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999978994793855e-05,
+      "loss": 0.8379,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999967179371862e-05,
+      "loss": 0.2758,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999952738306857e-05,
+      "loss": 0.8057,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993567160263e-05,
+      "loss": 0.8809,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991597926366e-05,
+      "loss": 0.8667,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999893661295124e-05,
+      "loss": 0.8123,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986871770288e-05,
+      "loss": 0.9062,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999841148493475e-05,
+      "loss": 0.8767,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999810953674146e-05,
+      "loss": 0.2788,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999977813325283e-05,
+      "loss": 0.7548,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999742687238134e-05,
+      "loss": 0.8376,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999704615639367e-05,
+      "loss": 0.8213,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999663918466532e-05,
+      "loss": 0.8101,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999620595730312e-05,
+      "loss": 0.8413,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999574647442074e-05,
+      "loss": 0.8777,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999526073613894e-05,
+      "loss": 0.8796,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999947487425852e-05,
+      "loss": 0.3039,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999421049389395e-05,
+      "loss": 0.8069,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999936459902065e-05,
+      "loss": 0.8235,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999305523167112e-05,
+      "loss": 0.8547,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999924382184429e-05,
+      "loss": 0.866,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999179495068378e-05,
+      "loss": 0.8125,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999911254285628e-05,
+      "loss": 0.8616,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999904296522556e-05,
+      "loss": 0.8267,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998970762194497e-05,
+      "loss": 0.8108,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999889593378205e-05,
+      "loss": 0.874,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998818480007857e-05,
+      "loss": 0.8628,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999873840089226e-05,
+      "loss": 0.7722,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999865569645629e-05,
+      "loss": 0.8469,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999857036672165e-05,
+      "loss": 0.8264,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998482411710758e-05,
+      "loss": 0.8484,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99983918314467e-05,
+      "loss": 0.8391,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999829862595326e-05,
+      "loss": 0.8511,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999820279525491e-05,
+      "loss": 0.8674,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998104339376817e-05,
+      "loss": 0.9333,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998003258344827e-05,
+      "loss": 0.8613,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997899552185482e-05,
+      "loss": 0.7761,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999779322092601e-05,
+      "loss": 0.8469,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997684264594332e-05,
+      "loss": 0.7822,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997572683219055e-05,
+      "loss": 0.8694,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997458476829477e-05,
+      "loss": 0.8704,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997341645455587e-05,
+      "loss": 0.8479,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997222189128053e-05,
+      "loss": 0.802,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999710010787825e-05,
+      "loss": 0.7883,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996975401738227e-05,
+      "loss": 0.8459,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996848070740725e-05,
+      "loss": 0.8904,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996718114919183e-05,
+      "loss": 0.692,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999658553430772e-05,
+      "loss": 0.8195,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996450328941143e-05,
+      "loss": 0.891,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999631249885496e-05,
+      "loss": 0.9224,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996172044085355e-05,
+      "loss": 0.8156,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996028964669204e-05,
+      "loss": 0.8591,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995883260644082e-05,
+      "loss": 0.8315,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999573493204824e-05,
+      "loss": 0.8206,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995583978920628e-05,
+      "loss": 0.2943,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999543040130088e-05,
+      "loss": 0.844,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995274199229322e-05,
+      "loss": 0.8755,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999511537274696e-05,
+      "loss": 0.8184,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994953921895504e-05,
+      "loss": 0.8298,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999478984671734e-05,
+      "loss": 0.8708,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999462314725555e-05,
+      "loss": 0.8225,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994453823553907e-05,
+      "loss": 0.8062,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994281875656864e-05,
+      "loss": 0.8613,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994107303609573e-05,
+      "loss": 0.8252,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999393010745787e-05,
+      "loss": 0.8032,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993750287248278e-05,
+      "loss": 0.8391,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999356784302801e-05,
+      "loss": 0.8674,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993382774844976e-05,
+      "loss": 0.8462,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999319508274776e-05,
+      "loss": 0.8788,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993004766785656e-05,
+      "loss": 0.8014,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999281182700862e-05,
+      "loss": 0.7977,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999261626346732e-05,
+      "loss": 0.8105,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992418076213102e-05,
+      "loss": 0.8102,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992217265298e-05,
+      "loss": 0.8279,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999201383077475e-05,
+      "loss": 0.8457,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991807772696753e-05,
+      "loss": 0.8699,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991599091118123e-05,
+      "loss": 0.8551,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991387786093647e-05,
+      "loss": 0.7823,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999117385767881e-05,
+      "loss": 0.8899,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999095730592978e-05,
+      "loss": 0.8391,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990738130903415e-05,
+      "loss": 0.8479,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990516332657267e-05,
+      "loss": 0.8464,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990291911249567e-05,
+      "loss": 0.8601,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990064866739245e-05,
+      "loss": 0.8516,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998983519918591e-05,
+      "loss": 0.8491,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998960290864987e-05,
+      "loss": 0.8442,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989367995192113e-05,
+      "loss": 0.8416,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989130458874323e-05,
+      "loss": 0.7878,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988890299758862e-05,
+      "loss": 0.7893,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998864751790879e-05,
+      "loss": 0.8477,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988402113387858e-05,
+      "loss": 0.8108,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988154086260494e-05,
+      "loss": 0.7961,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987903436591827e-05,
+      "loss": 0.8267,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987650164447664e-05,
+      "loss": 0.7551,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987394269894507e-05,
+      "loss": 0.8494,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998713575299955e-05,
+      "loss": 0.791,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998687461383066e-05,
+      "loss": 0.8247,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998661085245641e-05,
+      "loss": 0.7869,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986344468946058e-05,
+      "loss": 0.3262,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998607546336954e-05,
+      "loss": 0.8569,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998580383579749e-05,
+      "loss": 0.8486,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998552958630123e-05,
+      "loss": 0.8848,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985252714952764e-05,
+      "loss": 0.8354,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984973221824795e-05,
+      "loss": 0.8789,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99846911069907e-05,
+      "loss": 0.7899,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998440637052456e-05,
+      "loss": 0.8525,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998411901250113e-05,
+      "loss": 0.2974,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998382903299587e-05,
+      "loss": 0.8096,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983536432084908e-05,
+      "loss": 0.8301,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983241209845078e-05,
+      "loss": 0.8325,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982943366353892e-05,
+      "loss": 0.8047,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998264290168955e-05,
+      "loss": 0.802,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998233981593095e-05,
+      "loss": 0.7412,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998203410915767e-05,
+      "loss": 0.7938,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981725781449975e-05,
+      "loss": 0.7778,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981414832888825e-05,
+      "loss": 0.8105,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981101263555864e-05,
+      "loss": 0.8606,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998078507353342e-05,
+      "loss": 0.8186,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980466262904515e-05,
+      "loss": 0.8381,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998014483175286e-05,
+      "loss": 0.8206,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979820780162853e-05,
+      "loss": 0.702,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979494108219576e-05,
+      "loss": 0.7853,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99791648160088e-05,
+      "loss": 0.8186,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978832903616988e-05,
+      "loss": 0.7899,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978498371131284e-05,
+      "loss": 0.7783,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978161218639534e-05,
+      "loss": 0.7717,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977821446230253e-05,
+      "loss": 0.8809,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997747905399266e-05,
+      "loss": 0.8428,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997713404201665e-05,
+      "loss": 0.8416,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997678641039282e-05,
+      "loss": 0.9065,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997643615921244e-05,
+      "loss": 0.8423,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976083288567466e-05,
+      "loss": 0.7843,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975727798550563e-05,
+      "loss": 0.8586,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975369689255063e-05,
+      "loss": 0.8501,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975008960774994e-05,
+      "loss": 0.8269,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997464561320507e-05,
+      "loss": 0.8501,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99742796466407e-05,
+      "loss": 0.8027,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973911061177963e-05,
+      "loss": 0.8623,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973539856913645e-05,
+      "loss": 0.8225,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997316603394521e-05,
+      "loss": 0.8582,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972789592370807e-05,
+      "loss": 0.8391,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997241053228928e-05,
+      "loss": 0.7844,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972028853800156e-05,
+      "loss": 0.8539,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997164455700365e-05,
+      "loss": 0.8645,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971257642000666e-05,
+      "loss": 0.8066,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970868108892795e-05,
+      "loss": 0.7947,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997047595778231e-05,
+      "loss": 0.8701,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970081188772184e-05,
+      "loss": 0.8391,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996968380196606e-05,
+      "loss": 0.7979,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996928379746829e-05,
+      "loss": 0.8433,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968881175383893e-05,
+      "loss": 0.8174,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968475935818584e-05,
+      "loss": 0.83,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968068078878766e-05,
+      "loss": 0.8281,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996765760467153e-05,
+      "loss": 0.8684,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967244513304647e-05,
+      "loss": 0.7776,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966828804886583e-05,
+      "loss": 0.8333,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966410479526493e-05,
+      "loss": 0.8662,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996598953733421e-05,
+      "loss": 0.7655,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965565978420258e-05,
+      "loss": 0.2867,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965139802895856e-05,
+      "loss": 0.8386,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996471101087289e-05,
+      "loss": 0.7964,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964279602463955e-05,
+      "loss": 0.8821,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963845577782325e-05,
+      "loss": 0.7914,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963408936941952e-05,
+      "loss": 0.864,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962969680057492e-05,
+      "loss": 0.8293,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996252780724427e-05,
+      "loss": 0.8057,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962083318618314e-05,
+      "loss": 0.8684,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961636214296324e-05,
+      "loss": 0.808,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961186494395702e-05,
+      "loss": 0.2842,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960734159034522e-05,
+      "loss": 0.8267,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960279208331553e-05,
+      "loss": 0.756,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995982164240625e-05,
+      "loss": 0.8054,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959361461378756e-05,
+      "loss": 0.8403,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958898665369894e-05,
+      "loss": 0.8025,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958433254501185e-05,
+      "loss": 0.8774,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995796522889482e-05,
+      "loss": 0.8135,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957494588673692e-05,
+      "loss": 0.8406,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995702133396138e-05,
+      "loss": 0.7979,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995654546488213e-05,
+      "loss": 0.8181,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956066981560906e-05,
+      "loss": 0.8281,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955585884123325e-05,
+      "loss": 0.8762,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955102172695717e-05,
+      "loss": 0.8713,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954615847405084e-05,
+      "loss": 0.7668,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954126908379116e-05,
+      "loss": 0.7773,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953635355746198e-05,
+      "loss": 0.823,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953141189635385e-05,
+      "loss": 0.7939,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952644410176436e-05,
+      "loss": 0.8479,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952145017499787e-05,
+      "loss": 0.8706,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951643011736558e-05,
+      "loss": 0.7484,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951138393018558e-05,
+      "loss": 0.8054,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950631161478283e-05,
+      "loss": 0.8125,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950121317248918e-05,
+      "loss": 0.8403,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949608860464326e-05,
+      "loss": 0.813,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949093791259063e-05,
+      "loss": 0.7766,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948576109768363e-05,
+      "loss": 0.7739,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948055816128157e-05,
+      "loss": 0.8442,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947532910475057e-05,
+      "loss": 0.8433,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947007392946353e-05,
+      "loss": 0.8259,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946479263680034e-05,
+      "loss": 0.8945,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945948522814766e-05,
+      "loss": 0.871,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945415170489902e-05,
+      "loss": 0.8213,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994487920684548e-05,
+      "loss": 0.8374,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944340632022234e-05,
+      "loss": 0.8468,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943799446161565e-05,
+      "loss": 0.7815,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943255649405574e-05,
+      "loss": 0.8252,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942709241897044e-05,
+      "loss": 0.7739,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942160223779438e-05,
+      "loss": 0.8059,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941608595196914e-05,
+      "loss": 0.8804,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994105435629431e-05,
+      "loss": 0.8048,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940497507217147e-05,
+      "loss": 0.7633,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939938048111637e-05,
+      "loss": 0.8186,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993937597912467e-05,
+      "loss": 0.8315,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938811300403833e-05,
+      "loss": 0.8186,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938244012097384e-05,
+      "loss": 0.7694,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993767411435428e-05,
+      "loss": 0.8433,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937101607324146e-05,
+      "loss": 0.8496,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936526491157315e-05,
+      "loss": 0.7828,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993594876600478e-05,
+      "loss": 0.7817,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935368432018243e-05,
+      "loss": 0.77,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993478548935007e-05,
+      "loss": 0.7598,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993419993815333e-05,
+      "loss": 0.8119,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993361177858176e-05,
+      "loss": 0.8806,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933021010789795e-05,
+      "loss": 0.7095,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993242763493255e-05,
+      "loss": 0.814,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931831651165824e-05,
+      "loss": 0.7969,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931233059646102e-05,
+      "loss": 0.8599,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930631860530553e-05,
+      "loss": 0.8899,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993002805397703e-05,
+      "loss": 0.8669,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929421640144073e-05,
+      "loss": 0.771,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928812619190904e-05,
+      "loss": 0.8442,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992820099127743e-05,
+      "loss": 0.803,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927586756564245e-05,
+      "loss": 0.865,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926969915212627e-05,
+      "loss": 0.8005,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926350467384533e-05,
+      "loss": 0.8713,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925728413242612e-05,
+      "loss": 0.8162,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992510375295019e-05,
+      "loss": 0.8601,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924476486671284e-05,
+      "loss": 0.8188,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923846614570592e-05,
+      "loss": 0.7957,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923214136813498e-05,
+      "loss": 0.8124,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922579053566063e-05,
+      "loss": 0.7859,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921941364995044e-05,
+      "loss": 0.8843,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921301071267873e-05,
+      "loss": 0.8606,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992065817255267e-05,
+      "loss": 0.8398,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992001266901823e-05,
+      "loss": 0.8376,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919364560834053e-05,
+      "loss": 0.8137,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918713848170304e-05,
+      "loss": 0.8293,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918060531197833e-05,
+      "loss": 0.8887,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917404610088183e-05,
+      "loss": 0.2843,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916746085013578e-05,
+      "loss": 0.843,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991608495614692e-05,
+      "loss": 0.8044,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915421223661794e-05,
+      "loss": 0.9055,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914754887732483e-05,
+      "loss": 0.7465,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914085948533937e-05,
+      "loss": 0.7847,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99134144062418e-05,
+      "loss": 0.8264,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991274026103239e-05,
+      "loss": 0.8337,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991206351308272e-05,
+      "loss": 0.8148,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911384162570478e-05,
+      "loss": 0.854,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991070220967404e-05,
+      "loss": 0.8237,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910017654572456e-05,
+      "loss": 0.7991,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990933049744547e-05,
+      "loss": 0.8257,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908640738473512e-05,
+      "loss": 0.7798,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907948377837682e-05,
+      "loss": 0.8408,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990725341571977e-05,
+      "loss": 0.8018,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906555852302248e-05,
+      "loss": 0.9124,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905855687768275e-05,
+      "loss": 0.8464,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990515292230169e-05,
+      "loss": 0.8567,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904447556087014e-05,
+      "loss": 0.7533,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903739589309448e-05,
+      "loss": 0.7692,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903029022154887e-05,
+      "loss": 0.8381,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902315854809894e-05,
+      "loss": 0.751,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901600087461725e-05,
+      "loss": 0.83,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900881720298315e-05,
+      "loss": 0.7726,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900160753508282e-05,
+      "loss": 0.7823,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899437187280925e-05,
+      "loss": 0.7766,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898711021806233e-05,
+      "loss": 0.7794,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897982257274866e-05,
+      "loss": 0.8206,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897250893878172e-05,
+      "loss": 0.8171,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896516931808188e-05,
+      "loss": 0.7888,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989578037125762e-05,
+      "loss": 0.8062,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895041212419865e-05,
+      "loss": 0.806,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894299455489004e-05,
+      "loss": 0.8464,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893555100659792e-05,
+      "loss": 0.7979,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989280814812767e-05,
+      "loss": 0.8873,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989205859808877e-05,
+      "loss": 0.8374,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891306450739886e-05,
+      "loss": 0.8079,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890551706278516e-05,
+      "loss": 0.8445,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988979436490282e-05,
+      "loss": 0.814,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889034426811657e-05,
+      "loss": 0.8435,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888271892204562e-05,
+      "loss": 0.8267,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988750676128174e-05,
+      "loss": 0.793,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886739034244096e-05,
+      "loss": 0.7634,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885968711293205e-05,
+      "loss": 0.8281,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885195792631328e-05,
+      "loss": 0.786,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884420278461407e-05,
+      "loss": 0.7784,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883642168987063e-05,
+      "loss": 0.7201,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882861464412602e-05,
+      "loss": 0.7671,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882078164943014e-05,
+      "loss": 0.7917,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881292270783955e-05,
+      "loss": 0.8098,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880503782141786e-05,
+      "loss": 0.8013,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879712699223524e-05,
+      "loss": 0.8252,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878919022236894e-05,
+      "loss": 0.8438,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878122751390275e-05,
+      "loss": 0.8223,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987732388689275e-05,
+      "loss": 0.8044,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876522428954068e-05,
+      "loss": 0.7791,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875718377784663e-05,
+      "loss": 0.8374,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987491173359565e-05,
+      "loss": 0.8237,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874102496598834e-05,
+      "loss": 0.7812,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873290667006683e-05,
+      "loss": 0.8064,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987247624503236e-05,
+      "loss": 0.8069,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98716592308897e-05,
+      "loss": 0.8003,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870839624793226e-05,
+      "loss": 0.8474,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870017426958136e-05,
+      "loss": 0.7361,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869192637600315e-05,
+      "loss": 0.7969,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868365256936317e-05,
+      "loss": 0.7764,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867535285183388e-05,
+      "loss": 0.7527,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986670272255945e-05,
+      "loss": 0.7583,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98658675692831e-05,
+      "loss": 0.8586,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865029825573625e-05,
+      "loss": 0.8247,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864189491650986e-05,
+      "loss": 0.8041,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863346567735823e-05,
+      "loss": 0.7893,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862501054049465e-05,
+      "loss": 0.8213,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986165295081391e-05,
+      "loss": 0.803,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860802258251835e-05,
+      "loss": 0.7877,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985994897658661e-05,
+      "loss": 0.8713,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985909310604228e-05,
+      "loss": 0.8328,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858234646843556e-05,
+      "loss": 0.7919,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857373599215847e-05,
+      "loss": 0.8173,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856509963385228e-05,
+      "loss": 0.8367,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855643739578467e-05,
+      "loss": 0.8845,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854774928023003e-05,
+      "loss": 0.7717,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985390352894695e-05,
+      "loss": 0.7632,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853029542579115e-05,
+      "loss": 0.8,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985215296914897e-05,
+      "loss": 0.8335,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851273808886673e-05,
+      "loss": 0.7369,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850392062023066e-05,
+      "loss": 0.8184,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849507728789658e-05,
+      "loss": 0.8599,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984862080941865e-05,
+      "loss": 0.8506,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847731304142912e-05,
+      "loss": 0.8201,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846839213196002e-05,
+      "loss": 0.7966,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845944536812144e-05,
+      "loss": 0.8157,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984504727522626e-05,
+      "loss": 0.8058,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984414742867393e-05,
+      "loss": 0.8193,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843244997391423e-05,
+      "loss": 0.757,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842339981615693e-05,
+      "loss": 0.7955,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984143238158436e-05,
+      "loss": 0.7777,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984052219753573e-05,
+      "loss": 0.7402,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839609429708786e-05,
+      "loss": 0.801,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838694078343186e-05,
+      "loss": 0.8199,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837776143679274e-05,
+      "loss": 0.8464,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836855625958063e-05,
+      "loss": 0.807,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983593252542125e-05,
+      "loss": 0.8259,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983500684231121e-05,
+      "loss": 0.771,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834078576871e-05,
+      "loss": 0.8169,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833147729344343e-05,
+      "loss": 0.8223,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983221429997565e-05,
+      "loss": 0.3116,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831278289010004e-05,
+      "loss": 0.7871,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830339696693173e-05,
+      "loss": 0.8315,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829398523271603e-05,
+      "loss": 0.7986,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98284547689924e-05,
+      "loss": 0.8013,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982750843410337e-05,
+      "loss": 0.7991,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826559518852985e-05,
+      "loss": 0.8081,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825608023490403e-05,
+      "loss": 0.8274,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824653948265444e-05,
+      "loss": 0.761,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982369729342862e-05,
+      "loss": 0.8137,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822738059231116e-05,
+      "loss": 0.7399,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821776245924793e-05,
+      "loss": 0.8591,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982081185376219e-05,
+      "loss": 0.803,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819844882996523e-05,
+      "loss": 0.771,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981887533388168e-05,
+      "loss": 0.8389,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817903206672235e-05,
+      "loss": 0.8115,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816928501623435e-05,
+      "loss": 0.7791,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98159512189912e-05,
+      "loss": 0.8179,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814971359032136e-05,
+      "loss": 0.8549,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981398892200352e-05,
+      "loss": 0.7686,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813003908163298e-05,
+      "loss": 0.8284,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812016317770105e-05,
+      "loss": 0.7966,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811026151083253e-05,
+      "loss": 0.8586,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810033408362717e-05,
+      "loss": 0.8303,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980903808986916e-05,
+      "loss": 0.8137,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808040195863918e-05,
+      "loss": 0.3023,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807039726609005e-05,
+      "loss": 0.7959,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806036682367105e-05,
+      "loss": 0.8125,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805031063401587e-05,
+      "loss": 0.7684,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804022869976487e-05,
+      "loss": 0.2958,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803012102356528e-05,
+      "loss": 0.7819,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801998760807096e-05,
+      "loss": 0.7942,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980098284559426e-05,
+      "loss": 0.8147,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799964356984768e-05,
+      "loss": 0.8364,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979894329524604e-05,
+      "loss": 0.8298,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797919660646167e-05,
+      "loss": 0.8187,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796893453453916e-05,
+      "loss": 0.8513,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795864673938742e-05,
+      "loss": 0.8276,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794833322370764e-05,
+      "loss": 0.8098,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793799399020776e-05,
+      "loss": 0.7268,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792762904160255e-05,
+      "loss": 0.79,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791723838061342e-05,
+      "loss": 0.7632,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979068220099687e-05,
+      "loss": 0.7922,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789637993240323e-05,
+      "loss": 0.782,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788591215065884e-05,
+      "loss": 0.7944,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787541866748397e-05,
+      "loss": 0.783,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786489948563383e-05,
+      "loss": 0.8289,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785435460787042e-05,
+      "loss": 0.8594,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978437840369624e-05,
+      "loss": 0.8171,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783318777568534e-05,
+      "loss": 0.7532,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782256582682135e-05,
+      "loss": 0.8237,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781191819315944e-05,
+      "loss": 0.7826,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780124487749526e-05,
+      "loss": 0.8372,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779054588263127e-05,
+      "loss": 0.8274,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777982121137665e-05,
+      "loss": 0.7563,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977690708665474e-05,
+      "loss": 0.8384,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775829485096606e-05,
+      "loss": 0.8164,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774749316746212e-05,
+      "loss": 0.7902,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977366658188717e-05,
+      "loss": 0.7681,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977258128080377e-05,
+      "loss": 0.8218,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977149341378097e-05,
+      "loss": 0.785,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977040298110441e-05,
+      "loss": 0.7365,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97693099830604e-05,
+      "loss": 0.8132,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768214419935923e-05,
+      "loss": 0.7397,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767116292018633e-05,
+      "loss": 0.7778,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976601559959686e-05,
+      "loss": 0.8735,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764912342959612e-05,
+      "loss": 0.8342,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763806522396564e-05,
+      "loss": 0.8152,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976269813819806e-05,
+      "loss": 0.8315,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761587190655134e-05,
+      "loss": 0.8141,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976047368005947e-05,
+      "loss": 0.7837,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759357606703447e-05,
+      "loss": 0.8164,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758238970880105e-05,
+      "loss": 0.3185,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757117772883152e-05,
+      "loss": 0.7667,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755994013006983e-05,
+      "loss": 0.8718,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754867691546656e-05,
+      "loss": 0.8721,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97537388087979e-05,
+      "loss": 0.7701,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975260736505713e-05,
+      "loss": 0.8137,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751473360621413e-05,
+      "loss": 0.8779,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750336795788506e-05,
+      "loss": 0.8503,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749197670856825e-05,
+      "loss": 0.8142,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748055986125473e-05,
+      "loss": 0.7709,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974691174189421e-05,
+      "loss": 0.8291,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745764938463477e-05,
+      "loss": 0.8481,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744615576134384e-05,
+      "loss": 0.748,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743463655208713e-05,
+      "loss": 0.7869,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974230917598892e-05,
+      "loss": 0.8328,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974115213877813e-05,
+      "loss": 0.8221,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739992543880146e-05,
+      "loss": 0.8083,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738830391599425e-05,
+      "loss": 0.8079,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737665682241115e-05,
+      "loss": 0.8184,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973649841611103e-05,
+      "loss": 0.8213,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973532859351565e-05,
+      "loss": 0.7886,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734156214762135e-05,
+      "loss": 0.7649,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732981280158302e-05,
+      "loss": 0.3423,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731803790012656e-05,
+      "loss": 0.837,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730623744634362e-05,
+      "loss": 0.7983,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729441144333255e-05,
+      "loss": 0.8447,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972825598941985e-05,
+      "loss": 0.7908,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727068280205325e-05,
+      "loss": 0.8188,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972587801700153e-05,
+      "loss": 0.8407,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972468520012099e-05,
+      "loss": 0.8364,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723489829876893e-05,
+      "loss": 0.8208,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722291906583105e-05,
+      "loss": 0.813,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721091430554157e-05,
+      "loss": 0.8002,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719888402105257e-05,
+      "loss": 0.7751,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971868282155227e-05,
+      "loss": 0.757,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717474689211747e-05,
+      "loss": 0.8088,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716264005400897e-05,
+      "loss": 0.8273,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715050770437605e-05,
+      "loss": 0.8308,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971383498464042e-05,
+      "loss": 0.7648,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712616648328574e-05,
+      "loss": 0.8672,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711395761821953e-05,
+      "loss": 0.7937,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971017232544112e-05,
+      "loss": 0.7793,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970894633950731e-05,
+      "loss": 0.7318,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707717804342416e-05,
+      "loss": 0.792,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970648672026902e-05,
+      "loss": 0.7915,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705253087610353e-05,
+      "loss": 0.8347,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970401690669033e-05,
+      "loss": 0.3354,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970277817783352e-05,
+      "loss": 0.3099,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970153690136518e-05,
+      "loss": 0.8296,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700293077611226e-05,
+      "loss": 0.7937,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699046706898236e-05,
+      "loss": 0.8335,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697797789553466e-05,
+      "loss": 0.7947,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696546325904837e-05,
+      "loss": 0.8032,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695292316280943e-05,
+      "loss": 0.7964,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969403576101104e-05,
+      "loss": 0.8368,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969277666042506e-05,
+      "loss": 0.2897,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691515014853594e-05,
+      "loss": 0.8182,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969025082462791e-05,
+      "loss": 0.3116,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688984090079938e-05,
+      "loss": 0.8262,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968771481154228e-05,
+      "loss": 0.8616,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96864429893482e-05,
+      "loss": 0.821,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968516862383164e-05,
+      "loss": 0.8157,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683891715327203e-05,
+      "loss": 0.2811,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968261226417016e-05,
+      "loss": 0.299,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681330270696442e-05,
+      "loss": 0.8015,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968004573524267e-05,
+      "loss": 0.7588,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967875865814611e-05,
+      "loss": 0.8914,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677469039744705e-05,
+      "loss": 0.8322,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676176880377067e-05,
+      "loss": 0.8337,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674882180382467e-05,
+      "loss": 0.8503,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967358494010085e-05,
+      "loss": 0.8093,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672285159872827e-05,
+      "loss": 0.7854,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670982840039673e-05,
+      "loss": 0.8749,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669677980943333e-05,
+      "loss": 0.7865,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966837058292642e-05,
+      "loss": 0.7574,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667060646332208e-05,
+      "loss": 0.823,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966574817150464e-05,
+      "loss": 0.7803,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966443315878833e-05,
+      "loss": 0.8064,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966311560852855e-05,
+      "loss": 0.7979,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966179552107125e-05,
+      "loss": 0.8085,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660472896763027e-05,
+      "loss": 0.7743,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659147735951168e-05,
+      "loss": 0.803,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965782003898361e-05,
+      "loss": 0.7947,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656489806208955e-05,
+      "loss": 0.7622,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655157037976482e-05,
+      "loss": 0.8403,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965382173463613e-05,
+      "loss": 0.7996,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652483896538498e-05,
+      "loss": 0.7788,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965114352403486e-05,
+      "loss": 0.7913,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649800617477145e-05,
+      "loss": 0.8057,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964845517721796e-05,
+      "loss": 0.7815,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647107203610572e-05,
+      "loss": 0.8088,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645756697008905e-05,
+      "loss": 0.842,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644403657767558e-05,
+      "loss": 0.8193,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643048086241797e-05,
+      "loss": 0.7959,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964168998278754e-05,
+      "loss": 0.751,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640329347761382e-05,
+      "loss": 0.8169,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638966181520577e-05,
+      "loss": 0.8103,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963760048442305e-05,
+      "loss": 0.864,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636232256827374e-05,
+      "loss": 0.8379,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634861499092806e-05,
+      "loss": 0.7593,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963348821157926e-05,
+      "loss": 0.7693,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632112394647308e-05,
+      "loss": 0.7947,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96307340486582e-05,
+      "loss": 0.7911,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629353173973832e-05,
+      "loss": 0.7595,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627969770956782e-05,
+      "loss": 0.7679,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626583839970278e-05,
+      "loss": 0.7548,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962519538137822e-05,
+      "loss": 0.8469,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623804395545167e-05,
+      "loss": 0.7977,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622410882836343e-05,
+      "loss": 0.7883,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621014843617638e-05,
+      "loss": 0.8547,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619616278255603e-05,
+      "loss": 0.783,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618215187117448e-05,
+      "loss": 0.834,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961681157057106e-05,
+      "loss": 0.8406,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961540542898497e-05,
+      "loss": 0.8239,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961399676272839e-05,
+      "loss": 0.8142,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612585572171184e-05,
+      "loss": 0.8291,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961117185768388e-05,
+      "loss": 0.8169,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609755619637662e-05,
+      "loss": 0.8589,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608336858404404e-05,
+      "loss": 0.7998,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606915574356607e-05,
+      "loss": 0.7646,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960549176786746e-05,
+      "loss": 0.7981,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604065439310794e-05,
+      "loss": 0.7948,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602636589061124e-05,
+      "loss": 0.8267,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601205217493614e-05,
+      "loss": 0.7407,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959977132498409e-05,
+      "loss": 0.8547,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959833491190905e-05,
+      "loss": 0.8921,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959689597864563e-05,
+      "loss": 0.8401,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595454525571655e-05,
+      "loss": 0.2927,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594010553065602e-05,
+      "loss": 0.7974,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95925640615066e-05,
+      "loss": 0.8606,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591115051274456e-05,
+      "loss": 0.8093,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958966352274962e-05,
+      "loss": 0.845,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958820947631322e-05,
+      "loss": 0.8237,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586752912347037e-05,
+      "loss": 0.8176,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585293831233514e-05,
+      "loss": 0.766,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583832233355752e-05,
+      "loss": 0.8384,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582368119097518e-05,
+      "loss": 0.78,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580901488843236e-05,
+      "loss": 0.8652,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579432342977994e-05,
+      "loss": 0.7791,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577960681887535e-05,
+      "loss": 0.7651,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957648650595827e-05,
+      "loss": 0.8293,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575009815577267e-05,
+      "loss": 0.8091,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957353061113225e-05,
+      "loss": 0.7673,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572048893011603e-05,
+      "loss": 0.8026,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570564661604384e-05,
+      "loss": 0.7798,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956907791730029e-05,
+      "loss": 0.8806,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567588660489695e-05,
+      "loss": 0.8359,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566096891563625e-05,
+      "loss": 0.7645,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564602610913763e-05,
+      "loss": 0.808,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956310581893246e-05,
+      "loss": 0.8414,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561606516012718e-05,
+      "loss": 0.824,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560104702548202e-05,
+      "loss": 0.8669,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558600378933236e-05,
+      "loss": 0.7754,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557093545562802e-05,
+      "loss": 0.8071,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555584202832545e-05,
+      "loss": 0.7759,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554072351138765e-05,
+      "loss": 0.7993,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955255799087842e-05,
+      "loss": 0.7763,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551041122449128e-05,
+      "loss": 0.834,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954952174624917e-05,
+      "loss": 0.7539,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547999862677477e-05,
+      "loss": 0.7703,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546475472133642e-05,
+      "loss": 0.2872,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544948575017916e-05,
+      "loss": 0.7717,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543419171731216e-05,
+      "loss": 0.8254,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541887262675106e-05,
+      "loss": 0.7086,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540352848251806e-05,
+      "loss": 0.8268,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953881592886421e-05,
+      "loss": 0.8037,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537276504915853e-05,
+      "loss": 0.827,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535734576810938e-05,
+      "loss": 0.8004,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953419014495432e-05,
+      "loss": 0.7935,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532643209751508e-05,
+      "loss": 0.7836,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531093771608683e-05,
+      "loss": 0.7625,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529541830932666e-05,
+      "loss": 0.7778,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527987388130945e-05,
+      "loss": 0.7976,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952643044361166e-05,
+      "loss": 0.759,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524870997783614e-05,
+      "loss": 0.7957,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523309051056264e-05,
+      "loss": 0.7988,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952174460383972e-05,
+      "loss": 0.8643,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520177656544748e-05,
+      "loss": 0.816,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518608209582783e-05,
+      "loss": 0.7886,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95170362633659e-05,
+      "loss": 0.2944,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951546181830684e-05,
+      "loss": 0.814,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513884874818995e-05,
+      "loss": 0.8142,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512305433316418e-05,
+      "loss": 0.752,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510723494213813e-05,
+      "loss": 0.7844,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509139057926546e-05,
+      "loss": 0.8738,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950755212487063e-05,
+      "loss": 0.792,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505962695462743e-05,
+      "loss": 0.7401,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504370770120208e-05,
+      "loss": 0.877,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502776349261015e-05,
+      "loss": 0.7773,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95011794333038e-05,
+      "loss": 0.8024,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499580022667856e-05,
+      "loss": 0.7686,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497978117773137e-05,
+      "loss": 0.8088,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496373719040246e-05,
+      "loss": 0.7621,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494766826890445e-05,
+      "loss": 0.8257,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949315744174564e-05,
+      "loss": 0.8503,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491545564028402e-05,
+      "loss": 0.8323,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948993119416196e-05,
+      "loss": 0.7971,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488314332570183e-05,
+      "loss": 0.7306,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948669497967761e-05,
+      "loss": 0.8103,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948507313590942e-05,
+      "loss": 0.7986,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483448801691462e-05,
+      "loss": 0.7793,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948182197745022e-05,
+      "loss": 0.7902,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480192663612843e-05,
+      "loss": 0.8137,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478560860607135e-05,
+      "loss": 0.8022,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476926568861548e-05,
+      "loss": 0.8186,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475289788805195e-05,
+      "loss": 0.7996,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473650520867837e-05,
+      "loss": 0.8176,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947200876547988e-05,
+      "loss": 0.8125,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94703645230724e-05,
+      "loss": 0.7629,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468717794077113e-05,
+      "loss": 0.7579,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467068578926397e-05,
+      "loss": 0.7832,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465416878053275e-05,
+      "loss": 0.7754,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946376269189143e-05,
+      "loss": 0.8262,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946210602087519e-05,
+      "loss": 0.8523,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946044686543954e-05,
+      "loss": 0.7241,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458785226020117e-05,
+      "loss": 0.8469,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457121103053208e-05,
+      "loss": 0.3026,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455454496975756e-05,
+      "loss": 0.7848,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453785408225353e-05,
+      "loss": 0.7999,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452113837240242e-05,
+      "loss": 0.8239,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450439784459323e-05,
+      "loss": 0.8765,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944876325032214e-05,
+      "loss": 0.733,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447084235268895e-05,
+      "loss": 0.7833,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445402739740436e-05,
+      "loss": 0.7382,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944371876417827e-05,
+      "loss": 0.2951,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442032309024543e-05,
+      "loss": 0.7412,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440343374722065e-05,
+      "loss": 0.8008,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438651961714293e-05,
+      "loss": 0.7715,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436958070445327e-05,
+      "loss": 0.7568,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943526170135993e-05,
+      "loss": 0.8596,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433562854903502e-05,
+      "loss": 0.7937,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431861531522108e-05,
+      "loss": 0.8236,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430157731662453e-05,
+      "loss": 0.8279,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428451455771896e-05,
+      "loss": 0.8035,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426742704298446e-05,
+      "loss": 0.8198,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425031477690762e-05,
+      "loss": 0.8181,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423317776398152e-05,
+      "loss": 0.8245,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421601600870573e-05,
+      "loss": 0.7565,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419882951558633e-05,
+      "loss": 0.78,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418161828913594e-05,
+      "loss": 0.7649,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416438233387355e-05,
+      "loss": 0.7749,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941471216543248e-05,
+      "loss": 0.8237,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412983625502165e-05,
+      "loss": 0.8418,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411252614050274e-05,
+      "loss": 0.8038,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409519131531308e-05,
+      "loss": 0.7827,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407783178400413e-05,
+      "loss": 0.8208,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406044755113397e-05,
+      "loss": 0.8311,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404303862126707e-05,
+      "loss": 0.8325,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940256049989744e-05,
+      "loss": 0.8335,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400814668883342e-05,
+      "loss": 0.8391,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399066369542808e-05,
+      "loss": 0.8269,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397315602334882e-05,
+      "loss": 0.792,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939556236771925e-05,
+      "loss": 0.282,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393806666156255e-05,
+      "loss": 0.8069,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392048498106884e-05,
+      "loss": 0.2765,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390287864032764e-05,
+      "loss": 0.8259,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388524764396184e-05,
+      "loss": 0.7944,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938675919966007e-05,
+      "loss": 0.8276,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938499117028799e-05,
+      "loss": 0.7539,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938322067674418e-05,
+      "loss": 0.8013,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93814477194935e-05,
+      "loss": 0.8391,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379672299001472e-05,
+      "loss": 0.8425,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377894415734257e-05,
+      "loss": 0.8273,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376114070158666e-05,
+      "loss": 0.8176,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374331262742156e-05,
+      "loss": 0.8374,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372545993952832e-05,
+      "loss": 0.7694,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937075826425944e-05,
+      "loss": 0.8066,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368968074131374e-05,
+      "loss": 0.8877,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936717542403868e-05,
+      "loss": 0.8328,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936538031445205e-05,
+      "loss": 0.8232,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363582745842803e-05,
+      "loss": 0.8567,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361782718682928e-05,
+      "loss": 0.801,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359980233445048e-05,
+      "loss": 0.3008,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358175290602434e-05,
+      "loss": 0.2917,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356367890628997e-05,
+      "loss": 0.8315,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93545580339993e-05,
+      "loss": 0.7676,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935274572118855e-05,
+      "loss": 0.8101,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350930952672593e-05,
+      "loss": 0.756,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349113728927928e-05,
+      "loss": 0.752,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934729405043169e-05,
+      "loss": 0.8225,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345471917661672e-05,
+      "loss": 0.7955,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343647331096293e-05,
+      "loss": 0.8223,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341820291214626e-05,
+      "loss": 0.7887,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339990798496395e-05,
+      "loss": 0.8469,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338158853421957e-05,
+      "loss": 0.8281,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933632445647232e-05,
+      "loss": 0.7264,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334487608129128e-05,
+      "loss": 0.7711,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332648308874674e-05,
+      "loss": 0.8088,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330806559191898e-05,
+      "loss": 0.8552,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328962359564375e-05,
+      "loss": 0.8098,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932711571047633e-05,
+      "loss": 0.8423,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325266612412627e-05,
+      "loss": 0.8231,
+      "step": 911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932341506585878e-05,
+      "loss": 0.8074,
+      "step": 912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932156107130093e-05,
+      "loss": 0.7126,
+      "step": 913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931970462922588e-05,
+      "loss": 0.907,
+      "step": 914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931784574012107e-05,
+      "loss": 0.8545,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931598440447457e-05,
+      "loss": 0.7896,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314120622775106e-05,
+      "loss": 0.7897,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312254395512045e-05,
+      "loss": 0.2995,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310385723175388e-05,
+      "loss": 0.7604,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308514606255787e-05,
+      "loss": 0.8085,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930664104524453e-05,
+      "loss": 0.7466,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304765040633552e-05,
+      "loss": 0.7471,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302886592915426e-05,
+      "loss": 0.7639,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930100570258336e-05,
+      "loss": 0.2759,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929912237013122e-05,
+      "loss": 0.7581,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297236596053498e-05,
+      "loss": 0.8162,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295348380845335e-05,
+      "loss": 0.8364,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293457725002506e-05,
+      "loss": 0.8562,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291564629021437e-05,
+      "loss": 0.8428,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289669093399187e-05,
+      "loss": 0.8037,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287771118633455e-05,
+      "loss": 0.2648,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285870705222585e-05,
+      "loss": 0.7034,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928396785366556e-05,
+      "loss": 0.8254,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282062564462003e-05,
+      "loss": 0.7373,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928015483811218e-05,
+      "loss": 0.8281,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927824467511698e-05,
+      "loss": 0.8533,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276332075977962e-05,
+      "loss": 0.818,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274417041197297e-05,
+      "loss": 0.7479,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927249957127781e-05,
+      "loss": 0.8221,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927057966672296e-05,
+      "loss": 0.2922,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268657328036853e-05,
+      "loss": 0.8101,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266732555724223e-05,
+      "loss": 0.8145,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264805350290447e-05,
+      "loss": 0.7661,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262875712241548e-05,
+      "loss": 0.7339,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260943642084172e-05,
+      "loss": 0.7927,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259009140325622e-05,
+      "loss": 0.824,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257072207473828e-05,
+      "loss": 0.8014,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925513284403736e-05,
+      "loss": 0.8186,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925319105052543e-05,
+      "loss": 0.7689,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925124682744788e-05,
+      "loss": 0.8523,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249300175315202e-05,
+      "loss": 0.7773,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247351094638514e-05,
+      "loss": 0.8008,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245399585929577e-05,
+      "loss": 0.7573,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243445649700795e-05,
+      "loss": 0.8757,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241489286465196e-05,
+      "loss": 0.793,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239530496736458e-05,
+      "loss": 0.7804,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237569281028888e-05,
+      "loss": 0.8345,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235605639857433e-05,
+      "loss": 0.7482,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923363957373768e-05,
+      "loss": 0.8245,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231671083185847e-05,
+      "loss": 0.8169,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922970016871879e-05,
+      "loss": 0.8293,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227726830854005e-05,
+      "loss": 0.7493,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225751070109616e-05,
+      "loss": 0.8036,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92237728870044e-05,
+      "loss": 0.7666,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221792282057747e-05,
+      "loss": 0.7488,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219809255789706e-05,
+      "loss": 0.79,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921782380872094e-05,
+      "loss": 0.845,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215835941372765e-05,
+      "loss": 0.7938,
+      "step": 968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213845654267126e-05,
+      "loss": 0.8281,
+      "step": 969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211852947926596e-05,
+      "loss": 0.8208,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209857822874398e-05,
+      "loss": 0.7769,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207860279634382e-05,
+      "loss": 0.8584,
+      "step": 972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205860318731023e-05,
+      "loss": 0.822,
+      "step": 973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203857940689455e-05,
+      "loss": 0.8813,
+      "step": 974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201853146035424e-05,
+      "loss": 0.7756,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199845935295324e-05,
+      "loss": 0.8257,
+      "step": 976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197836308996175e-05,
+      "loss": 0.8545,
+      "step": 977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195824267665636e-05,
+      "loss": 0.7993,
+      "step": 978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193809811832e-05,
+      "loss": 0.7859,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191792942024192e-05,
+      "loss": 0.7842,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189773658771774e-05,
+      "loss": 0.7146,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187751962604933e-05,
+      "loss": 0.8005,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185727854054504e-05,
+      "loss": 0.7965,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183701333651943e-05,
+      "loss": 0.8083,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181672401929344e-05,
+      "loss": 0.823,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179641059419433e-05,
+      "loss": 0.825,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917760730665557e-05,
+      "loss": 0.8152,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917557114417175e-05,
+      "loss": 0.7759,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91735325725026e-05,
+      "loss": 0.813,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171491592183373e-05,
+      "loss": 0.7886,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169448203749955e-05,
+      "loss": 0.2891,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167402407738883e-05,
+      "loss": 0.792,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91653542046873e-05,
+      "loss": 0.7764,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163303595132997e-05,
+      "loss": 0.7631,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916125057961439e-05,
+      "loss": 0.8506,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159195158670532e-05,
+      "loss": 0.8184,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157137332841103e-05,
+      "loss": 0.8296,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915507710266642e-05,
+      "loss": 0.8389,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153014468687425e-05,
+      "loss": 0.7708,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150949431445695e-05,
+      "loss": 0.8186,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148881991483432e-05,
+      "loss": 0.7803,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914681214934348e-05,
+      "loss": 0.7859,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144739905569307e-05,
+      "loss": 0.2759,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142665260705008e-05,
+      "loss": 0.8643,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140588215295316e-05,
+      "loss": 0.7852,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138508769885588e-05,
+      "loss": 0.8279,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136426925021815e-05,
+      "loss": 0.853,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134342681250614e-05,
+      "loss": 0.7361,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132256039119244e-05,
+      "loss": 0.812,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130166999175578e-05,
+      "loss": 0.7922,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912807556196812e-05,
+      "loss": 0.7235,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912598172804602e-05,
+      "loss": 0.7695,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123885497959035e-05,
+      "loss": 0.8247,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912178687225757e-05,
+      "loss": 0.7981,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911968585149265e-05,
+      "loss": 0.8527,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117582436215922e-05,
+      "loss": 0.8601,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911547662697968e-05,
+      "loss": 0.7932,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911336842433683e-05,
+      "loss": 0.7408,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111257828840916e-05,
+      "loss": 0.78,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109144841046105e-05,
+      "loss": 0.8018,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910702946150719e-05,
+      "loss": 0.7855,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104911690779605e-05,
+      "loss": 0.8052,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102791529419392e-05,
+      "loss": 0.7834,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100668977983242e-05,
+      "loss": 0.7852,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098544037028456e-05,
+      "loss": 0.7986,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096416707112973e-05,
+      "loss": 0.7758,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094286988795356e-05,
+      "loss": 0.8203,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092154882634793e-05,
+      "loss": 0.8335,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90900203891911e-05,
+      "loss": 0.7607,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908788350902472e-05,
+      "loss": 0.7091,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085744242696728e-05,
+      "loss": 0.7308,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083602590768818e-05,
+      "loss": 0.76,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908145855380331e-05,
+      "loss": 0.7371,
+      "step": 1033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907931213236316e-05,
+      "loss": 0.7817,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077163327011936e-05,
+      "loss": 0.7467,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075012138313845e-05,
+      "loss": 0.8364,
+      "step": 1036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072858566833713e-05,
+      "loss": 0.8276,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070702613136995e-05,
+      "loss": 0.7942,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068544277789763e-05,
+      "loss": 0.2924,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066383561358727e-05,
+      "loss": 0.8149,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064220464411215e-05,
+      "loss": 0.7573,
+      "step": 1041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062054987515173e-05,
+      "loss": 0.8043,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905988713123919e-05,
+      "loss": 0.3183,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905771689615247e-05,
+      "loss": 0.7937,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905554428282483e-05,
+      "loss": 0.7925,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053369291826735e-05,
+      "loss": 0.7957,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905119192372925e-05,
+      "loss": 0.7742,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049012179104086e-05,
+      "loss": 0.7345,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904683005852356e-05,
+      "loss": 0.7455,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044645562560632e-05,
+      "loss": 0.8047,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904245869178886e-05,
+      "loss": 0.8379,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040269446782453e-05,
+      "loss": 0.8188,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038077828116225e-05,
+      "loss": 0.811,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035883836365613e-05,
+      "loss": 0.7668,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033687472106693e-05,
+      "loss": 0.7933,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903148873591615e-05,
+      "loss": 0.8195,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029287628371292e-05,
+      "loss": 0.3271,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027084150050063e-05,
+      "loss": 0.7589,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024878301531005e-05,
+      "loss": 0.7836,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902267008339331e-05,
+      "loss": 0.729,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902045949621677e-05,
+      "loss": 0.8032,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018246540581814e-05,
+      "loss": 0.7731,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016031217069483e-05,
+      "loss": 0.8258,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013813526261448e-05,
+      "loss": 0.7769,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011593468739993e-05,
+      "loss": 0.7535,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009371045088033e-05,
+      "loss": 0.8411,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007146255889093e-05,
+      "loss": 0.2977,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900491910172733e-05,
+      "loss": 0.8215,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002689583187514e-05,
+      "loss": 0.7395,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900045770085504e-05,
+      "loss": 0.8035,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998223455315924e-05,
+      "loss": 0.7903,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89959868471568e-05,
+      "loss": 0.8394,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899374787696492e-05,
+      "loss": 0.7925,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991506545328166e-05,
+      "loss": 0.7866,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989262852835032e-05,
+      "loss": 0.3082,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987016800074627e-05,
+      "loss": 0.8105,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984768387636694e-05,
+      "loss": 0.7561,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982517616111586e-05,
+      "loss": 0.8167,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898026448609027e-05,
+      "loss": 0.7957,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897800899816435e-05,
+      "loss": 0.8621,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975751152926034e-05,
+      "loss": 0.7856,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973490950968152e-05,
+      "loss": 0.7991,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971228392884153e-05,
+      "loss": 0.814,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968963479268114e-05,
+      "loss": 0.8306,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966696210714714e-05,
+      "loss": 0.2463,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964426587819265e-05,
+      "loss": 0.8511,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962154611177684e-05,
+      "loss": 0.8364,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959880281386517e-05,
+      "loss": 0.7571,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957603599042925e-05,
+      "loss": 0.7581,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955324564744683e-05,
+      "loss": 0.8252,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895304317909019e-05,
+      "loss": 0.8652,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950759442678452e-05,
+      "loss": 0.791,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948473356109103e-05,
+      "loss": 0.7773,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946184919982387e-05,
+      "loss": 0.8148,
+      "step": 1094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894389413489917e-05,
+      "loss": 0.7852,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941601001460933e-05,
+      "loss": 0.7852,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893930552026977e-05,
+      "loss": 0.7764,
+      "step": 1097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937007691928398e-05,
+      "loss": 0.7543,
+      "step": 1098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934707517040142e-05,
+      "loss": 0.8691,
+      "step": 1099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893240499620895e-05,
+      "loss": 0.8092,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893010013003939e-05,
+      "loss": 0.8696,
+      "step": 1101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927792919136627e-05,
+      "loss": 0.7832,
+      "step": 1102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925483364106464e-05,
+      "loss": 0.8145,
+      "step": 1103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923171465555304e-05,
+      "loss": 0.8054,
+      "step": 1104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920857224090175e-05,
+      "loss": 0.8408,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918540640318715e-05,
+      "loss": 0.7773,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916221714849178e-05,
+      "loss": 0.7866,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913900448290438e-05,
+      "loss": 0.7644,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911576841251966e-05,
+      "loss": 0.7979,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909250894343873e-05,
+      "loss": 0.7703,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906922608176863e-05,
+      "loss": 0.3006,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890459198336227e-05,
+      "loss": 0.7754,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890225902051203e-05,
+      "loss": 0.7842,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88999237202387e-05,
+      "loss": 0.7705,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897586083155442e-05,
+      "loss": 0.7439,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895246109876054e-05,
+      "loss": 0.8286,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892903801014913e-05,
+      "loss": 0.8611,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889055915718704e-05,
+      "loss": 0.814,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888212179008053e-05,
+      "loss": 0.8203,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888586286709418e-05,
+      "loss": 0.7687,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888351122206228e-05,
+      "loss": 0.8368,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888115724452981e-05,
+      "loss": 0.8186,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878800935114836e-05,
+      "loss": 0.7334,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887644229443605e-05,
+      "loss": 0.7635,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874081323112746e-05,
+      "loss": 0.7377,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887171802176483e-05,
+      "loss": 0.7947,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886935239101283e-05,
+      "loss": 0.804,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866984431477872e-05,
+      "loss": 0.7642,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88646141437817e-05,
+      "loss": 0.7789,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886224152854667e-05,
+      "loss": 0.8174,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885986658639575e-05,
+      "loss": 0.791,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857489317952516e-05,
+      "loss": 0.2816,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855109723841153e-05,
+      "loss": 0.7749,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852727804686464e-05,
+      "loss": 0.7792,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850343561113854e-05,
+      "loss": 0.78,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847956993749346e-05,
+      "loss": 0.7571,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845568103219568e-05,
+      "loss": 0.8538,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884317689015176e-05,
+      "loss": 0.8049,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840783355173767e-05,
+      "loss": 0.7444,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838387498914055e-05,
+      "loss": 0.7949,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835989322001683e-05,
+      "loss": 0.7769,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883358882506634e-05,
+      "loss": 0.7942,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831186008738306e-05,
+      "loss": 0.8313,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828780873648478e-05,
+      "loss": 0.8496,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882637342042836e-05,
+      "loss": 0.8813,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823963649710066e-05,
+      "loss": 0.8323,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882155156212632e-05,
+      "loss": 0.7771,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819137158310446e-05,
+      "loss": 0.8091,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881672043889639e-05,
+      "loss": 0.7849,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814301404518693e-05,
+      "loss": 0.7239,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811880055812513e-05,
+      "loss": 0.7964,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880945639341361e-05,
+      "loss": 0.7412,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880703041795835e-05,
+      "loss": 0.8193,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804602130083713e-05,
+      "loss": 0.729,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802171530427283e-05,
+      "loss": 0.7893,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799738619627254e-05,
+      "loss": 0.8159,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797303398322416e-05,
+      "loss": 0.8618,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794865867152178e-05,
+      "loss": 0.781,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792426026756552e-05,
+      "loss": 0.843,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789983877776152e-05,
+      "loss": 0.8118,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87875394208522e-05,
+      "loss": 0.7236,
+      "step": 1161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785092656626533e-05,
+      "loss": 0.7686,
+      "step": 1162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782643585741572e-05,
+      "loss": 0.2865,
+      "step": 1163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780192208840373e-05,
+      "loss": 0.7898,
+      "step": 1164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777738526566572e-05,
+      "loss": 0.749,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775282539564426e-05,
+      "loss": 0.7522,
+      "step": 1166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877282424847879e-05,
+      "loss": 0.761,
+      "step": 1167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770363653955125e-05,
+      "loss": 0.755,
+      "step": 1168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767900756639496e-05,
+      "loss": 0.3061,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876543555717858e-05,
+      "loss": 0.7615,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762968056219642e-05,
+      "loss": 0.7827,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876049825441057e-05,
+      "loss": 0.7803,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875802615239985e-05,
+      "loss": 0.8418,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875555175083656e-05,
+      "loss": 0.8579,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753075050370403e-05,
+      "loss": 0.833,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750596051651665e-05,
+      "loss": 0.6711,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874811475533125e-05,
+      "loss": 0.7383,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745631162060653e-05,
+      "loss": 0.8181,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874314527249199e-05,
+      "loss": 0.7509,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874065708727796e-05,
+      "loss": 0.8218,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738166607071878e-05,
+      "loss": 0.8137,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735673832527657e-05,
+      "loss": 0.8289,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733178764299813e-05,
+      "loss": 0.7983,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730681403043465e-05,
+      "loss": 0.8009,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728181749414327e-05,
+      "loss": 0.8313,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725679804068728e-05,
+      "loss": 0.8101,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723175567663587e-05,
+      "loss": 0.7676,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872066904085643e-05,
+      "loss": 0.8157,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718160224305387e-05,
+      "loss": 0.7944,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871564911866919e-05,
+      "loss": 0.8396,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713135724607153e-05,
+      "loss": 0.792,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710620042779214e-05,
+      "loss": 0.8022,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708102073845908e-05,
+      "loss": 0.7932,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705581818468358e-05,
+      "loss": 0.7341,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703059277308302e-05,
+      "loss": 0.2856,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700534451028064e-05,
+      "loss": 0.7585,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698007340290583e-05,
+      "loss": 0.8396,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695477945759385e-05,
+      "loss": 0.803,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692946268098607e-05,
+      "loss": 0.8003,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690412307972973e-05,
+      "loss": 0.7563,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868787606604781e-05,
+      "loss": 0.2927,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685337542989055e-05,
+      "loss": 0.8601,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682796739463235e-05,
+      "loss": 0.8191,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680253656137467e-05,
+      "loss": 0.793,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867770829367949e-05,
+      "loss": 0.7795,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675160652757617e-05,
+      "loss": 0.7606,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672610734040774e-05,
+      "loss": 0.8247,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867005853819848e-05,
+      "loss": 0.7605,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667504065900852e-05,
+      "loss": 0.7825,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664947317818607e-05,
+      "loss": 0.78,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662388294623057e-05,
+      "loss": 0.7626,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865982699698611e-05,
+      "loss": 0.834,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865726342558028e-05,
+      "loss": 0.761,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654697581078664e-05,
+      "loss": 0.8254,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865212946415497e-05,
+      "loss": 0.7661,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649559075483486e-05,
+      "loss": 0.7698,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864698641573912e-05,
+      "loss": 0.8494,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644411485597355e-05,
+      "loss": 0.8308,
+      "step": 1218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864183428573428e-05,
+      "loss": 0.76,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639254816826574e-05,
+      "loss": 0.7765,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636673079551525e-05,
+      "loss": 0.8547,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634089074586996e-05,
+      "loss": 0.7283,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863150280261146e-05,
+      "loss": 0.7294,
+      "step": 1223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862891426430399e-05,
+      "loss": 0.7766,
+      "step": 1224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626323460344234e-05,
+      "loss": 0.7817,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623730391412456e-05,
+      "loss": 0.7922,
+      "step": 1226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621135058189503e-05,
+      "loss": 0.7531,
+      "step": 1227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618537461356813e-05,
+      "loss": 0.775,
+      "step": 1228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615937601596433e-05,
+      "loss": 0.8152,
+      "step": 1229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613335479590992e-05,
+      "loss": 0.8208,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610731096023717e-05,
+      "loss": 0.7738,
+      "step": 1231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608124451578426e-05,
+      "loss": 0.8552,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605515546939536e-05,
+      "loss": 0.8452,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602904382792057e-05,
+      "loss": 0.7834,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600290959821583e-05,
+      "loss": 0.7493,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597675278714312e-05,
+      "loss": 0.769,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859505734015703e-05,
+      "loss": 0.8159,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592437144837118e-05,
+      "loss": 0.7806,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589814693442544e-05,
+      "loss": 0.7183,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587189986661877e-05,
+      "loss": 0.8323,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858456302518427e-05,
+      "loss": 0.7714,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858193380969947e-05,
+      "loss": 0.7422,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579302340897824e-05,
+      "loss": 0.77,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576668619470262e-05,
+      "loss": 0.3442,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574032646108305e-05,
+      "loss": 0.3115,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857139442150407e-05,
+      "loss": 0.7697,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568753946350256e-05,
+      "loss": 0.7937,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856611122134017e-05,
+      "loss": 0.855,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563466247167693e-05,
+      "loss": 0.8331,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560819024527305e-05,
+      "loss": 0.8174,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558169554114073e-05,
+      "loss": 0.8235,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855551783662366e-05,
+      "loss": 0.7866,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552863872752312e-05,
+      "loss": 0.8835,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550207663196863e-05,
+      "loss": 0.8098,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547549208654748e-05,
+      "loss": 0.7667,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854488850982398e-05,
+      "loss": 0.7184,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542225567403172e-05,
+      "loss": 0.8474,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539560382091513e-05,
+      "loss": 0.8159,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853689295458879e-05,
+      "loss": 0.8042,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534223285595375e-05,
+      "loss": 0.7616,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531551375812232e-05,
+      "loss": 0.8237,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528877225940914e-05,
+      "loss": 0.7964,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526200836683555e-05,
+      "loss": 0.7229,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523522208742884e-05,
+      "loss": 0.8286,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520841342822216e-05,
+      "loss": 0.8452,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851815823962545e-05,
+      "loss": 0.7668,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515472899857082e-05,
+      "loss": 0.74,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512785324222178e-05,
+      "loss": 0.7645,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510095513426413e-05,
+      "loss": 0.7682,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507403468176028e-05,
+      "loss": 0.7974,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504709189177864e-05,
+      "loss": 0.7881,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850201267713935e-05,
+      "loss": 0.8049,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499313932768487e-05,
+      "loss": 0.7061,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849661295677388e-05,
+      "loss": 0.8274,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493909749864706e-05,
+      "loss": 0.7693,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491204312750732e-05,
+      "loss": 0.8298,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488496646142313e-05,
+      "loss": 0.7681,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848578675075039e-05,
+      "loss": 0.833,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483074627286487e-05,
+      "loss": 0.8333,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480360276462708e-05,
+      "loss": 0.8076,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477643698991756e-05,
+      "loss": 0.8201,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474924895586902e-05,
+      "loss": 0.7346,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472203866962007e-05,
+      "loss": 0.7944,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469480613831528e-05,
+      "loss": 0.7566,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466755136910487e-05,
+      "loss": 0.7694,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464027436914507e-05,
+      "loss": 0.8254,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461297514559784e-05,
+      "loss": 0.7554,
+      "step": 1287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84585653705631e-05,
+      "loss": 0.8232,
+      "step": 1288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455831005641814e-05,
+      "loss": 0.7948,
+      "step": 1289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845309442051389e-05,
+      "loss": 0.8088,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450355615897845e-05,
+      "loss": 0.7069,
+      "step": 1291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447614592512802e-05,
+      "loss": 0.8134,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844487135107846e-05,
+      "loss": 0.8053,
+      "step": 1293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442125892315094e-05,
+      "loss": 0.7863,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843937821694356e-05,
+      "loss": 0.7786,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436628325685316e-05,
+      "loss": 0.7812,
+      "step": 1296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843387621926238e-05,
+      "loss": 0.7922,
+      "step": 1297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843112189839736e-05,
+      "loss": 0.7681,
+      "step": 1298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428365363813443e-05,
+      "loss": 0.7988,
+      "step": 1299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84256066162344e-05,
+      "loss": 0.7625,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422845656384583e-05,
+      "loss": 0.7012,
+      "step": 1301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420082484988925e-05,
+      "loss": 0.7522,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417317102772934e-05,
+      "loss": 0.7764,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841454951046271e-05,
+      "loss": 0.8313,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411779708784917e-05,
+      "loss": 0.7557,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409007698466816e-05,
+      "loss": 0.7898,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406233480236236e-05,
+      "loss": 0.7595,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403457054821595e-05,
+      "loss": 0.906,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400678422951883e-05,
+      "loss": 0.7622,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839789758535667e-05,
+      "loss": 0.7963,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395114542766106e-05,
+      "loss": 0.8225,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392329295910926e-05,
+      "loss": 0.7473,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838954184552244e-05,
+      "loss": 0.8006,
+      "step": 1313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386752192332526e-05,
+      "loss": 0.7507,
+      "step": 1314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383960337073657e-05,
+      "loss": 0.7419,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838116628047888e-05,
+      "loss": 0.8433,
+      "step": 1316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378370023281806e-05,
+      "loss": 0.3164,
+      "step": 1317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375571566216644e-05,
+      "loss": 0.2921,
+      "step": 1318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372770910018167e-05,
+      "loss": 0.7815,
+      "step": 1319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369968055421733e-05,
+      "loss": 0.8333,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367163003163276e-05,
+      "loss": 0.7463,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364355753979294e-05,
+      "loss": 0.8167,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361546308606882e-05,
+      "loss": 0.3222,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358734667783703e-05,
+      "loss": 0.7893,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355920832247994e-05,
+      "loss": 0.7485,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353104802738565e-05,
+      "loss": 0.7722,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350286579994817e-05,
+      "loss": 0.7893,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347466164756708e-05,
+      "loss": 0.7971,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344643557764787e-05,
+      "loss": 0.7998,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341818759760168e-05,
+      "loss": 0.7393,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338991771484548e-05,
+      "loss": 0.8115,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336162593680194e-05,
+      "loss": 0.8218,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833333122708995e-05,
+      "loss": 0.8223,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330497672457236e-05,
+      "loss": 0.8091,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832766193052604e-05,
+      "loss": 0.7717,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324824002040933e-05,
+      "loss": 0.8564,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321983887747055e-05,
+      "loss": 0.8328,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831914158839012e-05,
+      "loss": 0.7694,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316297104716418e-05,
+      "loss": 0.8271,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313450437472812e-05,
+      "loss": 0.783,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310601587406737e-05,
+      "loss": 0.7504,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83077505552662e-05,
+      "loss": 0.842,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304897341799783e-05,
+      "loss": 0.2637,
+      "step": 1343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302041947756644e-05,
+      "loss": 0.8335,
+      "step": 1344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299184373886508e-05,
+      "loss": 0.7804,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296324620939673e-05,
+      "loss": 0.7654,
+      "step": 1346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293462689667012e-05,
+      "loss": 0.7849,
+      "step": 1347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290598580819967e-05,
+      "loss": 0.7961,
+      "step": 1348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287732295150557e-05,
+      "loss": 0.783,
+      "step": 1349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828486383341136e-05,
+      "loss": 0.2827,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828199319635555e-05,
+      "loss": 0.8435,
+      "step": 1351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827912038473684e-05,
+      "loss": 0.8352,
+      "step": 1352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276245399309535e-05,
+      "loss": 0.731,
+      "step": 1353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273368240828512e-05,
+      "loss": 0.7821,
+      "step": 1354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270488910049205e-05,
+      "loss": 0.8159,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826760740772763e-05,
+      "loss": 0.8425,
+      "step": 1356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826472373462037e-05,
+      "loss": 0.748,
+      "step": 1357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261837891484572e-05,
+      "loss": 0.7988,
+      "step": 1358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258949879077965e-05,
+      "loss": 0.7549,
+      "step": 1359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256059698158837e-05,
+      "loss": 0.7673,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253167349486044e-05,
+      "loss": 0.8142,
+      "step": 1361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250272833819024e-05,
+      "loss": 0.845,
+      "step": 1362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247376151917773e-05,
+      "loss": 0.8381,
+      "step": 1363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244477304542858e-05,
+      "loss": 0.7664,
+      "step": 1364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241576292455417e-05,
+      "loss": 0.8679,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823867311641715e-05,
+      "loss": 0.7825,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235767777190333e-05,
+      "loss": 0.8208,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232860275537812e-05,
+      "loss": 0.7993,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229950612222986e-05,
+      "loss": 0.7681,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227038788009836e-05,
+      "loss": 0.2816,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822412480366291e-05,
+      "loss": 0.8557,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822120865994731e-05,
+      "loss": 0.7163,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218290357628722e-05,
+      "loss": 0.8513,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821536989747338e-05,
+      "loss": 0.7451,
+      "step": 1374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821244728024811e-05,
+      "loss": 0.7637,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209522506720275e-05,
+      "loss": 0.7864,
+      "step": 1376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206595577657828e-05,
+      "loss": 0.7925,
+      "step": 1377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203666493829272e-05,
+      "loss": 0.8259,
+      "step": 1378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820073525600369e-05,
+      "loss": 0.8062,
+      "step": 1379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197801864950716e-05,
+      "loss": 0.7959,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819486632144056e-05,
+      "loss": 0.7776,
+      "step": 1381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191928626243992e-05,
+      "loss": 0.7729,
+      "step": 1382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818898878013235e-05,
+      "loss": 0.7723,
+      "step": 1383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186046783877535e-05,
+      "loss": 0.8267,
+      "step": 1384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183102638252012e-05,
+      "loss": 0.7852,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180156344028807e-05,
+      "loss": 0.7856,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817720790198152e-05,
+      "loss": 0.8594,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817425731288431e-05,
+      "loss": 0.7357,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171304577511892e-05,
+      "loss": 0.7749,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816834969663956e-05,
+      "loss": 0.7157,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165392671043155e-05,
+      "loss": 0.7959,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816243350149909e-05,
+      "loss": 0.812,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815947218878435e-05,
+      "loss": 0.7979,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815650873367646e-05,
+      "loss": 0.8038,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153543136953528e-05,
+      "loss": 0.7388,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815057539939421e-05,
+      "loss": 0.2953,
+      "step": 1396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147605521777733e-05,
+      "loss": 0.812,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144633504883884e-05,
+      "loss": 0.7802,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814165934949301e-05,
+      "loss": 0.782,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138683056386028e-05,
+      "loss": 0.7759,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135704626344395e-05,
+      "loss": 0.7832,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132724060150157e-05,
+      "loss": 0.769,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81297413585859e-05,
+      "loss": 0.8008,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812675652243477e-05,
+      "loss": 0.822,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123769552480495e-05,
+      "loss": 0.7826,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120780449507347e-05,
+      "loss": 0.7473,
+      "step": 1406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117789214300153e-05,
+      "loss": 0.8501,
+      "step": 1407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811479584764431e-05,
+      "loss": 0.7808,
+      "step": 1408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811180035032577e-05,
+      "loss": 0.7517,
+      "step": 1409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108802723131055e-05,
+      "loss": 0.8032,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105802966847226e-05,
+      "loss": 0.7766,
+      "step": 1411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102801082261924e-05,
+      "loss": 0.804,
+      "step": 1412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809979707016333e-05,
+      "loss": 0.7557,
+      "step": 1413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096790931340202e-05,
+      "loss": 0.8284,
+      "step": 1414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809378266658184e-05,
+      "loss": 0.7502,
+      "step": 1415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090772276678118e-05,
+      "loss": 0.7937,
+      "step": 1416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808775976241945e-05,
+      "loss": 0.3055,
+      "step": 1417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808474512459682e-05,
+      "loss": 0.2905,
+      "step": 1418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081728364001777e-05,
+      "loss": 0.8113,
+      "step": 1419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80787094814264e-05,
+      "loss": 0.7988,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8075688477663354e-05,
+      "loss": 0.7888,
+      "step": 1421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072665353505848e-05,
+      "loss": 0.7839,
+      "step": 1422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069640109747647e-05,
+      "loss": 0.7993,
+      "step": 1423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066612747183074e-05,
+      "loss": 0.7788,
+      "step": 1424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063583266607013e-05,
+      "loss": 0.8684,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806055166881489e-05,
+      "loss": 0.7786,
+      "step": 1426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805751795460271e-05,
+      "loss": 0.7874,
+      "step": 1427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054482124767013e-05,
+      "loss": 0.7209,
+      "step": 1428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051444180104905e-05,
+      "loss": 0.8066,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048404121414047e-05,
+      "loss": 0.79,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045361949492643e-05,
+      "loss": 0.8303,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804231766513947e-05,
+      "loss": 0.7596,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803927126915385e-05,
+      "loss": 0.7852,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036222762335655e-05,
+      "loss": 0.7394,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033172145485324e-05,
+      "loss": 0.8176,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030119419403835e-05,
+      "loss": 0.8123,
+      "step": 1436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027064584892734e-05,
+      "loss": 0.731,
+      "step": 1437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802400764275411e-05,
+      "loss": 0.7704,
+      "step": 1438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020948593790618e-05,
+      "loss": 0.7952,
+      "step": 1439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017887438805442e-05,
+      "loss": 0.8518,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801482417860235e-05,
+      "loss": 0.7754,
+      "step": 1441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801175881398564e-05,
+      "loss": 0.7634,
+      "step": 1442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800869134576017e-05,
+      "loss": 0.7792,
+      "step": 1443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005621774731353e-05,
+      "loss": 0.8088,
+      "step": 1444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002550101705146e-05,
+      "loss": 0.7971,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799947632748807e-05,
+      "loss": 0.7703,
+      "step": 1446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996400452887188e-05,
+      "loss": 0.8159,
+      "step": 1447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993322478710116e-05,
+      "loss": 0.7625,
+      "step": 1448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990242405765025e-05,
+      "loss": 0.7881,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987160234860637e-05,
+      "loss": 0.8298,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984075966806216e-05,
+      "loss": 0.7307,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798098960241159e-05,
+      "loss": 0.7577,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977901142487126e-05,
+      "loss": 0.8052,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974810587843752e-05,
+      "loss": 0.7664,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797171793929293e-05,
+      "loss": 0.7471,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968623197646694e-05,
+      "loss": 0.7227,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965526363717605e-05,
+      "loss": 0.7542,
+      "step": 1457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962427438318788e-05,
+      "loss": 0.835,
+      "step": 1458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959326422263916e-05,
+      "loss": 0.8452,
+      "step": 1459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956223316367204e-05,
+      "loss": 0.7397,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795311812144342e-05,
+      "loss": 0.7783,
+      "step": 1461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950010838307878e-05,
+      "loss": 0.829,
+      "step": 1462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946901467776444e-05,
+      "loss": 0.7922,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943790010665532e-05,
+      "loss": 0.8162,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79406764677921e-05,
+      "loss": 0.8376,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793756083997366e-05,
+      "loss": 0.8064,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793444312802826e-05,
+      "loss": 0.793,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793132333277451e-05,
+      "loss": 0.7947,
+      "step": 1468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928201455031552e-05,
+      "loss": 0.7993,
+      "step": 1469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925077495619092e-05,
+      "loss": 0.7727,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921951455357365e-05,
+      "loss": 0.8115,
+      "step": 1471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918823335067164e-05,
+      "loss": 0.7937,
+      "step": 1472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791569313556982e-05,
+      "loss": 0.8083,
+      "step": 1473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791256085768722e-05,
+      "loss": 0.7834,
+      "step": 1474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790942650224179e-05,
+      "loss": 0.8218,
+      "step": 1475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79062900700565e-05,
+      "loss": 0.8354,
+      "step": 1476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903151561954865e-05,
+      "loss": 0.7954,
+      "step": 1477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790001097876096e-05,
+      "loss": 0.7319,
+      "step": 1478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896868321299377e-05,
+      "loss": 0.7937,
+      "step": 1479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789372359039528e-05,
+      "loss": 0.781,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890576786874362e-05,
+      "loss": 0.6702,
+      "step": 1481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788742791156286e-05,
+      "loss": 0.7913,
+      "step": 1482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884276965287565e-05,
+      "loss": 0.7883,
+      "step": 1483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881123948875805e-05,
+      "loss": 0.7836,
+      "step": 1484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877968863155447e-05,
+      "loss": 0.8076,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787481170895492e-05,
+      "loss": 0.7744,
+      "step": 1486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787165248710316e-05,
+      "loss": 0.8574,
+      "step": 1487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868491198429688e-05,
+      "loss": 0.7458,
+      "step": 1488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865327843764542e-05,
+      "loss": 0.8191,
+      "step": 1489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786216242393831e-05,
+      "loss": 0.7539,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858994939782116e-05,
+      "loss": 0.7876,
+      "step": 1491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855825392127636e-05,
+      "loss": 0.6617,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785265378180708e-05,
+      "loss": 0.8241,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849480109653207e-05,
+      "loss": 0.7305,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846304376499308e-05,
+      "loss": 0.7942,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784312658317922e-05,
+      "loss": 0.7656,
+      "step": 1496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839946730527324e-05,
+      "loss": 0.7756,
+      "step": 1497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836764819378534e-05,
+      "loss": 0.8076,
+      "step": 1498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833580850568317e-05,
+      "loss": 0.3073,
+      "step": 1499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830394824932663e-05,
+      "loss": 0.7681,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827206743308116e-05,
+      "loss": 0.7617,
+      "step": 1501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824016606531757e-05,
+      "loss": 0.8662,
+      "step": 1502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78208244154412e-05,
+      "loss": 0.8271,
+      "step": 1503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781763017087461e-05,
+      "loss": 0.8171,
+      "step": 1504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781443387367067e-05,
+      "loss": 0.8584,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811235524668635e-05,
+      "loss": 0.7565,
+      "step": 1506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808035124708266e-05,
+      "loss": 0.7715,
+      "step": 1507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804832674629887e-05,
+      "loss": 0.7871,
+      "step": 1508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801628175274338e-05,
+      "loss": 0.7615,
+      "step": 1509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779842162748302e-05,
+      "loss": 0.7197,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795213032097857e-05,
+      "loss": 0.7825,
+      "step": 1511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792002389961308e-05,
+      "loss": 0.8296,
+      "step": 1512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788789701916388e-05,
+      "loss": 0.7939,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785574968806626e-05,
+      "loss": 0.7274,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782358191476107e-05,
+      "loss": 0.7764,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777913937076944e-05,
+      "loss": 0.813,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777591850753177e-05,
+      "loss": 0.7852,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772695602608796e-05,
+      "loss": 0.7725,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769470656846733e-05,
+      "loss": 0.7733,
+      "step": 1519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776624367109234e-05,
+      "loss": 0.8125,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763014646192912e-05,
+      "loss": 0.7491,
+      "step": 1521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759783582996276e-05,
+      "loss": 0.7695,
+      "step": 1522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77565504823508e-05,
+      "loss": 0.798,
+      "step": 1523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753315345105383e-05,
+      "loss": 0.7728,
+      "step": 1524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750078172109456e-05,
+      "loss": 0.7648,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746838964212994e-05,
+      "loss": 0.7048,
+      "step": 1526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743597722266494e-05,
+      "loss": 0.7169,
+      "step": 1527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740354447120996e-05,
+      "loss": 0.8508,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737109139628073e-05,
+      "loss": 0.8049,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733861800639828e-05,
+      "loss": 0.8191,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730612431008896e-05,
+      "loss": 0.7488,
+      "step": 1531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727361031588453e-05,
+      "loss": 0.792,
+      "step": 1532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77241076032322e-05,
+      "loss": 0.7903,
+      "step": 1533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720852146794374e-05,
+      "loss": 0.8042,
+      "step": 1534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771759466312975e-05,
+      "loss": 0.7795,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771433515309362e-05,
+      "loss": 0.6736,
+      "step": 1536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711073617541825e-05,
+      "loss": 0.7083,
+      "step": 1537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707810057330728e-05,
+      "loss": 0.8315,
+      "step": 1538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770454447331723e-05,
+      "loss": 0.8066,
+      "step": 1539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701276866358753e-05,
+      "loss": 0.8315,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698007237313263e-05,
+      "loss": 0.7825,
+      "step": 1541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694735587039245e-05,
+      "loss": 0.8188,
+      "step": 1542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691461916395725e-05,
+      "loss": 0.7864,
+      "step": 1543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688186226242252e-05,
+      "loss": 0.7756,
+      "step": 1544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768490851743891e-05,
+      "loss": 0.3177,
+      "step": 1545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768162879084631e-05,
+      "loss": 0.7847,
+      "step": 1546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678347047325596e-05,
+      "loss": 0.2943,
+      "step": 1547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675063287738436e-05,
+      "loss": 0.7896,
+      "step": 1548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767177751294703e-05,
+      "loss": 0.7836,
+      "step": 1549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766848972381412e-05,
+      "loss": 0.8418,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766519992120295e-05,
+      "loss": 0.3033,
+      "step": 1551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661908105977313e-05,
+      "loss": 0.7372,
+      "step": 1552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658614279001525e-05,
+      "loss": 0.7543,
+      "step": 1553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765531844114043e-05,
+      "loss": 0.8384,
+      "step": 1554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652020593259404e-05,
+      "loss": 0.7917,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648720736224342e-05,
+      "loss": 0.7358,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764541887090167e-05,
+      "loss": 0.7367,
+      "step": 1557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764211499815835e-05,
+      "loss": 0.6898,
+      "step": 1558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638809118861856e-05,
+      "loss": 0.7638,
+      "step": 1559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635501233880203e-05,
+      "loss": 0.7681,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632191344081923e-05,
+      "loss": 0.7534,
+      "step": 1561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762887945033608e-05,
+      "loss": 0.7925,
+      "step": 1562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762556555351226e-05,
+      "loss": 0.7709,
+      "step": 1563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762224965448058e-05,
+      "loss": 0.7444,
+      "step": 1564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761893175411167e-05,
+      "loss": 0.7501,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761561185327671e-05,
+      "loss": 0.7947,
+      "step": 1566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761228995284738e-05,
+      "loss": 0.7128,
+      "step": 1567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608966053695896e-05,
+      "loss": 0.7861,
+      "step": 1568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605640156695e-05,
+      "loss": 0.7831,
+      "step": 1569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602312262717957e-05,
+      "loss": 0.7822,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598982372638555e-05,
+      "loss": 0.8096,
+      "step": 1571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595650487331107e-05,
+      "loss": 0.7725,
+      "step": 1572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759231660767045e-05,
+      "loss": 0.8599,
+      "step": 1573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588980734531948e-05,
+      "loss": 0.7815,
+      "step": 1574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758564286879148e-05,
+      "loss": 0.7676,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582303011325458e-05,
+      "loss": 0.3301,
+      "step": 1576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578961163010812e-05,
+      "loss": 0.8131,
+      "step": 1577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575617324724986e-05,
+      "loss": 0.7754,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572271497345968e-05,
+      "loss": 0.3202,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756892368175225e-05,
+      "loss": 0.2939,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756557387882285e-05,
+      "loss": 0.7396,
+      "step": 1581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756222208943731e-05,
+      "loss": 0.7654,
+      "step": 1582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558868314475694e-05,
+      "loss": 0.8459,
+      "step": 1583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755551255481859e-05,
+      "loss": 0.8101,
+      "step": 1584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75521548113471e-05,
+      "loss": 0.7916,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754879508494285e-05,
+      "loss": 0.3027,
+      "step": 1586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545433376487985e-05,
+      "loss": 0.7358,
+      "step": 1587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542069686865178e-05,
+      "loss": 0.7815,
+      "step": 1588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538704016957616e-05,
+      "loss": 0.8071,
+      "step": 1589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535336367649006e-05,
+      "loss": 0.8074,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531966739823568e-05,
+      "loss": 0.2999,
+      "step": 1591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528595134366063e-05,
+      "loss": 0.7568,
+      "step": 1592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525221552161744e-05,
+      "loss": 0.8342,
+      "step": 1593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521845994096402e-05,
+      "loss": 0.7961,
+      "step": 1594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751846846105634e-05,
+      "loss": 0.8044,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515088953928385e-05,
+      "loss": 0.7961,
+      "step": 1596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751170747359987e-05,
+      "loss": 0.7321,
+      "step": 1597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750832402095866e-05,
+      "loss": 0.7681,
+      "step": 1598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750493859689313e-05,
+      "loss": 0.8003,
+      "step": 1599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501551202292174e-05,
+      "loss": 0.8062,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74981618380452e-05,
+      "loss": 0.7693,
+      "step": 1601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494770505042143e-05,
+      "loss": 0.7733,
+      "step": 1602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491377204173447e-05,
+      "loss": 0.8188,
+      "step": 1603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487981936330075e-05,
+      "loss": 0.7697,
+      "step": 1604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484584702403503e-05,
+      "loss": 0.2951,
+      "step": 1605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748118550328573e-05,
+      "loss": 0.7461,
+      "step": 1606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477784339869267e-05,
+      "loss": 0.7908,
+      "step": 1607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474381213047138e-05,
+      "loss": 0.7437,
+      "step": 1608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747097612371289e-05,
+      "loss": 0.7827,
+      "step": 1609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467569072760574e-05,
+      "loss": 0.7485,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464160061084772e-05,
+      "loss": 0.2731,
+      "step": 1611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460749089580566e-05,
+      "loss": 0.717,
+      "step": 1612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457336159143557e-05,
+      "loss": 0.7837,
+      "step": 1613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453921270669866e-05,
+      "loss": 0.2872,
+      "step": 1614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450504425056118e-05,
+      "loss": 0.8123,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447085623199463e-05,
+      "loss": 0.781,
+      "step": 1616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744366486599756e-05,
+      "loss": 0.7405,
+      "step": 1617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744024215434857e-05,
+      "loss": 0.7114,
+      "step": 1618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436817489151195e-05,
+      "loss": 0.7577,
+      "step": 1619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433390871304623e-05,
+      "loss": 0.322,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429962301708562e-05,
+      "loss": 0.7819,
+      "step": 1621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742653178126324e-05,
+      "loss": 0.7759,
+      "step": 1622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423099310869386e-05,
+      "loss": 0.7556,
+      "step": 1623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419664891428255e-05,
+      "loss": 0.7971,
+      "step": 1624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74162285238416e-05,
+      "loss": 0.7228,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412790209011698e-05,
+      "loss": 0.7263,
+      "step": 1626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409349947841327e-05,
+      "loss": 0.776,
+      "step": 1627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405907741233773e-05,
+      "loss": 0.7325,
+      "step": 1628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402463590092852e-05,
+      "loss": 0.7908,
+      "step": 1629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739901749532287e-05,
+      "loss": 0.7832,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739556945782865e-05,
+      "loss": 0.7306,
+      "step": 1631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392119478515534e-05,
+      "loss": 0.309,
+      "step": 1632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388667558289358e-05,
+      "loss": 0.7422,
+      "step": 1633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385213698056484e-05,
+      "loss": 0.7891,
+      "step": 1634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738175789872377e-05,
+      "loss": 0.8245,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737830016119859e-05,
+      "loss": 0.7556,
+      "step": 1636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374840486388823e-05,
+      "loss": 0.7565,
+      "step": 1637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371378875202865e-05,
+      "loss": 0.7098,
+      "step": 1638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736791532854961e-05,
+      "loss": 0.7717,
+      "step": 1639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736444984733847e-05,
+      "loss": 0.759,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736098243247935e-05,
+      "loss": 0.285,
+      "step": 1641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357513084882676e-05,
+      "loss": 0.8064,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354041805459383e-05,
+      "loss": 0.7487,
+      "step": 1643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350568595120906e-05,
+      "loss": 0.8002,
+      "step": 1644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347093454779185e-05,
+      "loss": 0.7426,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343616385346674e-05,
+      "loss": 0.775,
+      "step": 1646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734013738773633e-05,
+      "loss": 0.7559,
+      "step": 1647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336656462861617e-05,
+      "loss": 0.7189,
+      "step": 1648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333173611636502e-05,
+      "loss": 0.7527,
+      "step": 1649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329688834975465e-05,
+      "loss": 0.8142,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7326202133793485e-05,
+      "loss": 0.7406,
+      "step": 1651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732271350900604e-05,
+      "loss": 0.7969,
+      "step": 1652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319222961529137e-05,
+      "loss": 0.7734,
+      "step": 1653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315730492279262e-05,
+      "loss": 0.813,
+      "step": 1654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312236102173416e-05,
+      "loss": 0.7861,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308739792129108e-05,
+      "loss": 0.2587,
+      "step": 1656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305241563064346e-05,
+      "loss": 0.7782,
+      "step": 1657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730174141589764e-05,
+      "loss": 0.8428,
+      "step": 1658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729823935154801e-05,
+      "loss": 0.7656,
+      "step": 1659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294735370934974e-05,
+      "loss": 0.7637,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291229474978557e-05,
+      "loss": 0.8081,
+      "step": 1661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728772166459928e-05,
+      "loss": 0.8097,
+      "step": 1662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284211940718182e-05,
+      "loss": 0.752,
+      "step": 1663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280700304256788e-05,
+      "loss": 0.7659,
+      "step": 1664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727718675613713e-05,
+      "loss": 0.7883,
+      "step": 1665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273671297281745e-05,
+      "loss": 0.7787,
+      "step": 1666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270153928613673e-05,
+      "loss": 0.7986,
+      "step": 1667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266634651056447e-05,
+      "loss": 0.7954,
+      "step": 1668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726311346553411e-05,
+      "loss": 0.7832,
+      "step": 1669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72595903729712e-05,
+      "loss": 0.7939,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256065374292763e-05,
+      "loss": 0.7551,
+      "step": 1671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252538470424337e-05,
+      "loss": 0.7255,
+      "step": 1672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249009662291966e-05,
+      "loss": 0.8153,
+      "step": 1673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724547895082219e-05,
+      "loss": 0.8391,
+      "step": 1674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7241946336942054e-05,
+      "loss": 0.7045,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72384118215791e-05,
+      "loss": 0.8083,
+      "step": 1676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234875405661363e-05,
+      "loss": 0.7594,
+      "step": 1677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231337090117387e-05,
+      "loss": 0.7755,
+      "step": 1678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227796875876205e-05,
+      "loss": 0.7949,
+      "step": 1679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224254763867364e-05,
+      "loss": 0.811,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722071075502089e-05,
+      "loss": 0.8357,
+      "step": 1681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217164850267322e-05,
+      "loss": 0.8274,
+      "step": 1682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213617050537686e-05,
+      "loss": 0.2925,
+      "step": 1683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210067356763514e-05,
+      "loss": 0.8396,
+      "step": 1684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720651576987683e-05,
+      "loss": 0.7897,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202962290810162e-05,
+      "loss": 0.7972,
+      "step": 1686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719940692049652e-05,
+      "loss": 0.7192,
+      "step": 1687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195849659869433e-05,
+      "loss": 0.7831,
+      "step": 1688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192290509862905e-05,
+      "loss": 0.7905,
+      "step": 1689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188729471411446e-05,
+      "loss": 0.7812,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718516654545006e-05,
+      "loss": 0.8074,
+      "step": 1691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718160173291425e-05,
+      "loss": 0.803,
+      "step": 1692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178035034740012e-05,
+      "loss": 0.7627,
+      "step": 1693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174466451863834e-05,
+      "loss": 0.7123,
+      "step": 1694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71708959852227e-05,
+      "loss": 0.8174,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167323635754098e-05,
+      "loss": 0.7721,
+      "step": 1696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163749404395993e-05,
+      "loss": 0.8026,
+      "step": 1697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160173292086862e-05,
+      "loss": 0.7264,
+      "step": 1698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715659529976566e-05,
+      "loss": 0.7135,
+      "step": 1699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153015428371847e-05,
+      "loss": 0.7555,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714943367884537e-05,
+      "loss": 0.7653,
+      "step": 1701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145850052126676e-05,
+      "loss": 0.7407,
+      "step": 1702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142264549156703e-05,
+      "loss": 0.7644,
+      "step": 1703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713867717087687e-05,
+      "loss": 0.8462,
+      "step": 1704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71350879182291e-05,
+      "loss": 0.729,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713149679215581e-05,
+      "loss": 0.7246,
+      "step": 1706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127903793599904e-05,
+      "loss": 0.7467,
+      "step": 1707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7124308923504772e-05,
+      "loss": 0.8145,
+      "step": 1708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120712182814308e-05,
+      "loss": 0.8245,
+      "step": 1709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711711357247289e-05,
+      "loss": 0.7723,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113513093425386e-05,
+      "loss": 0.7888,
+      "step": 1711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710991074661716e-05,
+      "loss": 0.7852,
+      "step": 1712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106306532994055e-05,
+      "loss": 0.7482,
+      "step": 1713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102700453502417e-05,
+      "loss": 0.7548,
+      "step": 1714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709909250908908e-05,
+      "loss": 0.7953,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709548270070136e-05,
+      "loss": 0.8242,
+      "step": 1716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709187102928707e-05,
+      "loss": 0.8,
+      "step": 1717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088257495794506e-05,
+      "loss": 0.8254,
+      "step": 1718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708464210117246e-05,
+      "loss": 0.7451,
+      "step": 1719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70810248463702e-05,
+      "loss": 0.7885,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077405732337497e-05,
+      "loss": 0.7898,
+      "step": 1721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073784760024607e-05,
+      "loss": 0.7935,
+      "step": 1722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7070161930382268e-05,
+      "loss": 0.7571,
+      "step": 1723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706653724436171e-05,
+      "loss": 0.7661,
+      "step": 1724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7062910702914645e-05,
+      "loss": 0.8203,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059282306993287e-05,
+      "loss": 0.739,
+      "step": 1726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055652057550314e-05,
+      "loss": 0.8047,
+      "step": 1727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705201995553891e-05,
+      "loss": 0.3207,
+      "step": 1728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048386001912735e-05,
+      "loss": 0.7595,
+      "step": 1729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044750197625942e-05,
+      "loss": 0.7942,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041112543633164e-05,
+      "loss": 0.8025,
+      "step": 1731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037473040889526e-05,
+      "loss": 0.7812,
+      "step": 1732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703383169035063e-05,
+      "loss": 0.7604,
+      "step": 1733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703018849297257e-05,
+      "loss": 0.7731,
+      "step": 1734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026543449711924e-05,
+      "loss": 0.7546,
+      "step": 1735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702289656152575e-05,
+      "loss": 0.7996,
+      "step": 1736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019247829371595e-05,
+      "loss": 0.8174,
+      "step": 1737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701559725420749e-05,
+      "loss": 0.6667,
+      "step": 1738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701194483699195e-05,
+      "loss": 0.7693,
+      "step": 1739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700829057868397e-05,
+      "loss": 0.7688,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700463448024303e-05,
+      "loss": 0.7839,
+      "step": 1741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000976542629096e-05,
+      "loss": 0.7832,
+      "step": 1742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997316766802618e-05,
+      "loss": 0.762,
+      "step": 1743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993655153724523e-05,
+      "loss": 0.7874,
+      "step": 1744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989991704356218e-05,
+      "loss": 0.7664,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986326419659605e-05,
+      "loss": 0.7747,
+      "step": 1746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982659300597055e-05,
+      "loss": 0.7517,
+      "step": 1747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978990348131426e-05,
+      "loss": 0.7406,
+      "step": 1748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975319563226063e-05,
+      "loss": 0.8542,
+      "step": 1749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697164694684478e-05,
+      "loss": 0.7771,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696797249995188e-05,
+      "loss": 0.7979,
+      "step": 1751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964296223512146e-05,
+      "loss": 0.8018,
+      "step": 1752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960618118490833e-05,
+      "loss": 0.8098,
+      "step": 1753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956938185853697e-05,
+      "loss": 0.7539,
+      "step": 1754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953256426566946e-05,
+      "loss": 0.7391,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949572841597294e-05,
+      "loss": 0.8413,
+      "step": 1756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945887431911913e-05,
+      "loss": 0.741,
+      "step": 1757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942200198478463e-05,
+      "loss": 0.7549,
+      "step": 1758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693851114226509e-05,
+      "loss": 0.8228,
+      "step": 1759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693482026424041e-05,
+      "loss": 0.7859,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693112756537351e-05,
+      "loss": 0.7917,
+      "step": 1761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692743304663398e-05,
+      "loss": 0.7992,
+      "step": 1762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923736708991856e-05,
+      "loss": 0.7684,
+      "step": 1763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920038553417675e-05,
+      "loss": 0.7585,
+      "step": 1764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916338580882446e-05,
+      "loss": 0.7915,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912636792357648e-05,
+      "loss": 0.8376,
+      "step": 1766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908933188815245e-05,
+      "loss": 0.8279,
+      "step": 1767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905227771227672e-05,
+      "loss": 0.7679,
+      "step": 1768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901520540567844e-05,
+      "loss": 0.8809,
+      "step": 1769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897811497809154e-05,
+      "loss": 0.7407,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894100643925456e-05,
+      "loss": 0.7917,
+      "step": 1771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890387979891106e-05,
+      "loss": 0.8049,
+      "step": 1772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688667350668091e-05,
+      "loss": 0.7466,
+      "step": 1773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882957225270164e-05,
+      "loss": 0.8,
+      "step": 1774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687923913663463e-05,
+      "loss": 0.7854,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687551924175055e-05,
+      "loss": 0.7483,
+      "step": 1776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687179754159464e-05,
+      "loss": 0.8408,
+      "step": 1777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686807403714409e-05,
+      "loss": 0.801,
+      "step": 1778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686434872937656e-05,
+      "loss": 0.8672,
+      "step": 1779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686062161927019e-05,
+      "loss": 0.7599,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685689270780358e-05,
+      "loss": 0.7805,
+      "step": 1781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853161995955823e-05,
+      "loss": 0.7769,
+      "step": 1782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849429484706467e-05,
+      "loss": 0.7445,
+      "step": 1783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684569517503554e-05,
+      "loss": 0.7253,
+      "step": 1784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684195906792355e-05,
+      "loss": 0.7804,
+      "step": 1785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838221164351458e-05,
+      "loss": 0.8069,
+      "step": 1786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683448146530071e-05,
+      "loss": 0.804,
+      "step": 1787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830739971753224e-05,
+      "loss": 0.7524,
+      "step": 1788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826996684691384e-05,
+      "loss": 0.8215,
+      "step": 1789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823251605098045e-05,
+      "loss": 0.7645,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681950473395654e-05,
+      "loss": 0.7798,
+      "step": 1791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681575607225066e-05,
+      "loss": 0.2791,
+      "step": 1792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812005620964676e-05,
+      "loss": 0.8149,
+      "step": 1793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808253381083326e-05,
+      "loss": 0.7849,
+      "step": 1794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804499353591815e-05,
+      "loss": 0.739,
+      "step": 1795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800743539475827e-05,
+      "loss": 0.834,
+      "step": 1796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.67969859397215e-05,
+      "loss": 0.7501,
+      "step": 1797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793226555315452e-05,
+      "loss": 0.741,
+      "step": 1798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789465387244764e-05,
+      "loss": 0.7535,
+      "step": 1799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785702436496992e-05,
+      "loss": 0.7933,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781937704060156e-05,
+      "loss": 0.8589,
+      "step": 1801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677817119092274e-05,
+      "loss": 0.772,
+      "step": 1802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67744028980737e-05,
+      "loss": 0.6964,
+      "step": 1803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770632826502457e-05,
+      "loss": 0.3174,
+      "step": 1804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766860977198903e-05,
+      "loss": 0.7732,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763087351153395e-05,
+      "loss": 0.8101,
+      "step": 1806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675931194935675e-05,
+      "loss": 0.8579,
+      "step": 1807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755534772800266e-05,
+      "loss": 0.7805,
+      "step": 1808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751755822475695e-05,
+      "loss": 0.6913,
+      "step": 1809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674797509937525e-05,
+      "loss": 0.8011,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674419260449163e-05,
+      "loss": 0.7769,
+      "step": 1811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674040833881797e-05,
+      "loss": 0.8,
+      "step": 1812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67366223033479e-05,
+      "loss": 0.8501,
+      "step": 1813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732834499075502e-05,
+      "loss": 0.7944,
+      "step": 1814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672904492699531e-05,
+      "loss": 0.7692,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725253588102343e-05,
+      "loss": 0.7831,
+      "step": 1816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672146048339207e-05,
+      "loss": 0.7732,
+      "step": 1817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671766561386043e-05,
+      "loss": 0.8333,
+      "step": 1818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713868980503824e-05,
+      "loss": 0.7402,
+      "step": 1819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710070584319115e-05,
+      "loss": 0.7379,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706270426303626e-05,
+      "loss": 0.2991,
+      "step": 1821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702468507455152e-05,
+      "loss": 0.7402,
+      "step": 1822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669866482877194e-05,
+      "loss": 0.7743,
+      "step": 1823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669485939125271e-05,
+      "loss": 0.8135,
+      "step": 1824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691052195896626e-05,
+      "loss": 0.8042,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687243243703335e-05,
+      "loss": 0.7715,
+      "step": 1826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668343253567293e-05,
+      "loss": 0.7861,
+      "step": 1827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679620072805975e-05,
+      "loss": 0.7863,
+      "step": 1828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675805856103485e-05,
+      "loss": 0.7966,
+      "step": 1829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671989886566945e-05,
+      "loss": 0.291,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666817216519829e-05,
+      "loss": 0.8213,
+      "step": 1831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6664352692999922e-05,
+      "loss": 0.7402,
+      "step": 1832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660531470974705e-05,
+      "loss": 0.8635,
+      "step": 1833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656708500125956e-05,
+      "loss": 0.8184,
+      "step": 1834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652883781457453e-05,
+      "loss": 0.7786,
+      "step": 1835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664905731597344e-05,
+      "loss": 0.7468,
+      "step": 1836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6645229104678606e-05,
+      "loss": 0.7697,
+      "step": 1837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6641399148578107e-05,
+      "loss": 0.7949,
+      "step": 1838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663756744867756e-05,
+      "loss": 0.8059,
+      "step": 1839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633734005983037e-05,
+      "loss": 0.2979,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662989882150106e-05,
+      "loss": 0.792,
+      "step": 1841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626061896238616e-05,
+      "loss": 0.7964,
+      "step": 1842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622223231203153e-05,
+      "loss": 0.7183,
+      "step": 1843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661838282740257e-05,
+      "loss": 0.7146,
+      "step": 1844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614540685845218e-05,
+      "loss": 0.7637,
+      "step": 1845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610696807539913e-05,
+      "loss": 0.7112,
+      "step": 1846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606851193495923e-05,
+      "loss": 0.7209,
+      "step": 1847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660300384472297e-05,
+      "loss": 0.7803,
+      "step": 1848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659915476223124e-05,
+      "loss": 0.7141,
+      "step": 1849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659530394703137e-05,
+      "loss": 0.8071,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591451400134432e-05,
+      "loss": 0.7722,
+      "step": 1851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587597122551988e-05,
+      "loss": 0.7673,
+      "step": 1852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658374111529603e-05,
+      "loss": 0.7561,
+      "step": 1853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579883379379017e-05,
+      "loss": 0.7809,
+      "step": 1854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657602391581385e-05,
+      "loss": 0.761,
+      "step": 1855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657216272561389e-05,
+      "loss": 0.8008,
+      "step": 1856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568299809792957e-05,
+      "loss": 0.72,
+      "step": 1857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564435169365308e-05,
+      "loss": 0.7979,
+      "step": 1858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560568805345673e-05,
+      "loss": 0.7395,
+      "step": 1859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6556700718749218e-05,
+      "loss": 0.7708,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655283091059157e-05,
+      "loss": 0.7408,
+      "step": 1861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548959381888807e-05,
+      "loss": 0.7969,
+      "step": 1862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654508613365746e-05,
+      "loss": 0.7479,
+      "step": 1863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65412111669145e-05,
+      "loss": 0.7498,
+      "step": 1864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537334482677362e-05,
+      "loss": 0.7605,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6533456081963932e-05,
+      "loss": 0.7932,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529575965792543e-05,
+      "loss": 0.7424,
+      "step": 1867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525694135181976e-05,
+      "loss": 0.7637,
+      "step": 1868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521810591151462e-05,
+      "loss": 0.3295,
+      "step": 1869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651792533472069e-05,
+      "loss": 0.7593,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651403836690979e-05,
+      "loss": 0.7297,
+      "step": 1871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510149688739342e-05,
+      "loss": 0.7598,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650625930123038e-05,
+      "loss": 0.8264,
+      "step": 1873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502367205404385e-05,
+      "loss": 0.7717,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498473402283286e-05,
+      "loss": 0.7986,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494577892889458e-05,
+      "loss": 0.7504,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490680678245725e-05,
+      "loss": 0.7803,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486781759375365e-05,
+      "loss": 0.7786,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482881137302095e-05,
+      "loss": 0.7847,
+      "step": 1879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647897881305008e-05,
+      "loss": 0.8181,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475074787643937e-05,
+      "loss": 0.8467,
+      "step": 1881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6471169062108726e-05,
+      "loss": 0.7859,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467261637469956e-05,
+      "loss": 0.8135,
+      "step": 1883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646335251475358e-05,
+      "loss": 0.8442,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459441694986002e-05,
+      "loss": 0.74,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645552917919406e-05,
+      "loss": 0.7194,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451614968405042e-05,
+      "loss": 0.7531,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447699063646695e-05,
+      "loss": 0.7512,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443781465947188e-05,
+      "loss": 0.8503,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643986217633515e-05,
+      "loss": 0.7477,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435941195839654e-05,
+      "loss": 0.702,
+      "step": 1891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432018525490205e-05,
+      "loss": 0.7471,
+      "step": 1892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428094166316762e-05,
+      "loss": 0.7673,
+      "step": 1893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642416811934973e-05,
+      "loss": 0.7979,
+      "step": 1894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420240385619947e-05,
+      "loss": 0.7771,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64163109661587e-05,
+      "loss": 0.7979,
+      "step": 1896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641237986199772e-05,
+      "loss": 0.7859,
+      "step": 1897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6408447074169175e-05,
+      "loss": 0.7419,
+      "step": 1898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640451260370568e-05,
+      "loss": 0.8098,
+      "step": 1899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640057645164029e-05,
+      "loss": 0.7882,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63966386190065e-05,
+      "loss": 0.7913,
+      "step": 1901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392699106838248e-05,
+      "loss": 0.8186,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638875791616991e-05,
+      "loss": 0.7584,
+      "step": 1903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384815048036316e-05,
+      "loss": 0.7755,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380870503472713e-05,
+      "loss": 0.803,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376924283514807e-05,
+      "loss": 0.7142,
+      "step": 1906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372976389198736e-05,
+      "loss": 0.7797,
+      "step": 1907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636902682156108e-05,
+      "loss": 0.7599,
+      "step": 1908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636507558163886e-05,
+      "loss": 0.77,
+      "step": 1909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361122670469534e-05,
+      "loss": 0.8333,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357168089090992e-05,
+      "loss": 0.7474,
+      "step": 1911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353211838541582e-05,
+      "loss": 0.804,
+      "step": 1912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349253919860065e-05,
+      "loss": 0.842,
+      "step": 1913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634529433408566e-05,
+      "loss": 0.8176,
+      "step": 1914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634133308225801e-05,
+      "loss": 0.76,
+      "step": 1915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633737016541721e-05,
+      "loss": 0.8235,
+      "step": 1916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333405584603775e-05,
+      "loss": 0.8372,
+      "step": 1917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329439340858672e-05,
+      "loss": 0.834,
+      "step": 1918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325471435223298e-05,
+      "loss": 0.7275,
+      "step": 1919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321501868739483e-05,
+      "loss": 0.7347,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317530642449503e-05,
+      "loss": 0.7228,
+      "step": 1921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313557757396056e-05,
+      "loss": 0.8071,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630958321462229e-05,
+      "loss": 0.7572,
+      "step": 1923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305607015171777e-05,
+      "loss": 0.7914,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6301629160088535e-05,
+      "loss": 0.725,
+      "step": 1925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297649650417002e-05,
+      "loss": 0.7852,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293668487202064e-05,
+      "loss": 0.8137,
+      "step": 1927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628968567148903e-05,
+      "loss": 0.748,
+      "step": 1928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285701204323654e-05,
+      "loss": 0.7655,
+      "step": 1929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281715086752116e-05,
+      "loss": 0.7339,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6277727319821032e-05,
+      "loss": 0.7572,
+      "step": 1931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273737904577453e-05,
+      "loss": 0.7664,
+      "step": 1932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626974684206886e-05,
+      "loss": 0.7806,
+      "step": 1933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265754133343162e-05,
+      "loss": 0.7397,
+      "step": 1934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626175977944871e-05,
+      "loss": 0.774,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257763781434278e-05,
+      "loss": 0.7361,
+      "step": 1936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253766140349084e-05,
+      "loss": 0.719,
+      "step": 1937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249766857242762e-05,
+      "loss": 0.8257,
+      "step": 1938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245765933165384e-05,
+      "loss": 0.3234,
+      "step": 1939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241763369167458e-05,
+      "loss": 0.7834,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237759166299913e-05,
+      "loss": 0.8044,
+      "step": 1941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623375332561412e-05,
+      "loss": 0.7494,
+      "step": 1942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6229745848161864e-05,
+      "loss": 0.7284,
+      "step": 1943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622573673499538e-05,
+      "loss": 0.6952,
+      "step": 1944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622172598716731e-05,
+      "loss": 0.3092,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217713605730745e-05,
+      "loss": 0.8435,
+      "step": 1946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213699591739193e-05,
+      "loss": 0.7173,
+      "step": 1947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209683946246592e-05,
+      "loss": 0.8606,
+      "step": 1948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205666670307316e-05,
+      "loss": 0.7446,
+      "step": 1949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620164776497616e-05,
+      "loss": 0.8,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6197627231308346e-05,
+      "loss": 0.7266,
+      "step": 1951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193605070359528e-05,
+      "loss": 0.7742,
+      "step": 1952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189581283185784e-05,
+      "loss": 0.7473,
+      "step": 1953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185555870843623e-05,
+      "loss": 0.7845,
+      "step": 1954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618152883438998e-05,
+      "loss": 0.775,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177500174882202e-05,
+      "loss": 0.7714,
+      "step": 1956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617346989337809e-05,
+      "loss": 0.2966,
+      "step": 1957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169437990935848e-05,
+      "loss": 0.7566,
+      "step": 1958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165404468614117e-05,
+      "loss": 0.7874,
+      "step": 1959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161369327471953e-05,
+      "loss": 0.6752,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615733256856885e-05,
+      "loss": 0.7906,
+      "step": 1961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153294192964714e-05,
+      "loss": 0.7732,
+      "step": 1962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614925420171989e-05,
+      "loss": 0.842,
+      "step": 1963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614521259589513e-05,
+      "loss": 0.766,
+      "step": 1964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141169376551623e-05,
+      "loss": 0.7338,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613712454475098e-05,
+      "loss": 0.7684,
+      "step": 1966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133078101555225e-05,
+      "loss": 0.7761,
+      "step": 1967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129030048026825e-05,
+      "loss": 0.8103,
+      "step": 1968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612498038522865e-05,
+      "loss": 0.7507,
+      "step": 1969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120929114224e-05,
+      "loss": 0.7363,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116876236076603e-05,
+      "loss": 0.7832,
+      "step": 1971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6112821751850595e-05,
+      "loss": 0.7651,
+      "step": 1972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610876566261055e-05,
+      "loss": 0.7274,
+      "step": 1973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104707969421457e-05,
+      "loss": 0.7246,
+      "step": 1974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100648673348714e-05,
+      "loss": 0.3281,
+      "step": 1975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609658777545816e-05,
+      "loss": 0.777,
+      "step": 1976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092525276816043e-05,
+      "loss": 0.7407,
+      "step": 1977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088461178489036e-05,
+      "loss": 0.7646,
+      "step": 1978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084395481544222e-05,
+      "loss": 0.7986,
+      "step": 1979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080328187049123e-05,
+      "loss": 0.7522,
+      "step": 1980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6076259296071658e-05,
+      "loss": 0.8083,
+      "step": 1981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607218880968018e-05,
+      "loss": 0.8204,
+      "step": 1982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606811672894346e-05,
+      "loss": 0.2963,
+      "step": 1983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606404305493068e-05,
+      "loss": 0.7614,
+      "step": 1984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059967788711448e-05,
+      "loss": 0.7024,
+      "step": 1985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055890931355788e-05,
+      "loss": 0.7954,
+      "step": 1986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051812483934135e-05,
+      "loss": 0.714,
+      "step": 1987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6047732447517352e-05,
+      "loss": 0.8088,
+      "step": 1988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604365082317671e-05,
+      "loss": 0.7312,
+      "step": 1989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039567611983908e-05,
+      "loss": 0.765,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035482815011053e-05,
+      "loss": 0.7471,
+      "step": 1991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031396433330664e-05,
+      "loss": 0.7584,
+      "step": 1992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027308468015692e-05,
+      "loss": 0.8037,
+      "step": 1993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602321892013949e-05,
+      "loss": 0.8013,
+      "step": 1994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6019127790775825e-05,
+      "loss": 0.771,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015035080998895e-05,
+      "loss": 0.7659,
+      "step": 1996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010940791883295e-05,
+      "loss": 0.7794,
+      "step": 1997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600684492450405e-05,
+      "loss": 0.8062,
+      "step": 1998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002747479936586e-05,
+      "loss": 0.7643,
+      "step": 1999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998648459256752e-05,
+      "loss": 0.8379,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599454786354081e-05,
+      "loss": 0.7791,
+      "step": 2001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990445693865428e-05,
+      "loss": 0.7812,
+      "step": 2002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986341951307696e-05,
+      "loss": 0.8258,
+      "step": 2003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982236636945113e-05,
+      "loss": 0.7834,
+      "step": 2004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978129751855593e-05,
+      "loss": 0.7595,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597402129711746e-05,
+      "loss": 0.8052,
+      "step": 2006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969911273809453e-05,
+      "loss": 0.7938,
+      "step": 2007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596579968301072e-05,
+      "loss": 0.7725,
+      "step": 2008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596168652580082e-05,
+      "loss": 0.7679,
+      "step": 2009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957571803259723e-05,
+      "loss": 0.8047,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595345551646782e-05,
+      "loss": 0.7949,
+      "step": 2011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594933766650589e-05,
+      "loss": 0.8005,
+      "step": 2012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945218254455152e-05,
+      "loss": 0.793,
+      "step": 2013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594109728139721e-05,
+      "loss": 0.8099,
+      "step": 2014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59369747484141e-05,
+      "loss": 0.7717,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932850656588242e-05,
+      "loss": 0.7271,
+      "step": 2016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928725007002483e-05,
+      "loss": 0.7733,
+      "step": 2017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924597800740074e-05,
+      "loss": 0.7893,
+      "step": 2018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920469038884684e-05,
+      "loss": 0.7581,
+      "step": 2019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916338722520374e-05,
+      "loss": 0.8174,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912206852731622e-05,
+      "loss": 0.3077,
+      "step": 2021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590807343060331e-05,
+      "loss": 0.7753,
+      "step": 2022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590393845722074e-05,
+      "loss": 0.8169,
+      "step": 2023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899801933669604e-05,
+      "loss": 0.7854,
+      "step": 2024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589566386103601e-05,
+      "loss": 0.7539,
+      "step": 2025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589152424040648e-05,
+      "loss": 0.7944,
+      "step": 2026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887383072867923e-05,
+      "loss": 0.803,
+      "step": 2027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588324035950767e-05,
+      "loss": 0.7599,
+      "step": 2028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5879096101413453e-05,
+      "loss": 0.7585,
+      "step": 2029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874950299673412e-05,
+      "loss": 0.7935,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870802955376087e-05,
+      "loss": 0.322,
+      "step": 2031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866654069610424e-05,
+      "loss": 0.7952,
+      "step": 2032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862503643465784e-05,
+      "loss": 0.7211,
+      "step": 2033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858351678031914e-05,
+      "loss": 0.7378,
+      "step": 2034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854198174398984e-05,
+      "loss": 0.3147,
+      "step": 2035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850043133657554e-05,
+      "loss": 0.8137,
+      "step": 2036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845886556898595e-05,
+      "loss": 0.804,
+      "step": 2037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584172844521348e-05,
+      "loss": 0.741,
+      "step": 2038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837568799693988e-05,
+      "loss": 0.8076,
+      "step": 2039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833407621432287e-05,
+      "loss": 0.8259,
+      "step": 2040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829244911520965e-05,
+      "loss": 0.7205,
+      "step": 2041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825080671053e-05,
+      "loss": 0.7634,
+      "step": 2042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820914901121782e-05,
+      "loss": 0.769,
+      "step": 2043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816747602821093e-05,
+      "loss": 0.6456,
+      "step": 2044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812578777245125e-05,
+      "loss": 0.7452,
+      "step": 2045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580840842548846e-05,
+      "loss": 0.7883,
+      "step": 2046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804236548646088e-05,
+      "loss": 0.7487,
+      "step": 2047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800063147813402e-05,
+      "loss": 0.8037,
+      "step": 2048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579588822408619e-05,
+      "loss": 0.7573,
+      "step": 2049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791711778560644e-05,
+      "loss": 0.7661,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578753381233335e-05,
+      "loss": 0.7617,
+      "step": 2051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783354326501294e-05,
+      "loss": 0.6851,
+      "step": 2052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577917332216187e-05,
+      "loss": 0.7275,
+      "step": 2053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774990800412857e-05,
+      "loss": 0.7502,
+      "step": 2054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5770806762352446e-05,
+      "loss": 0.8228,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766621209079217e-05,
+      "loss": 0.7886,
+      "step": 2056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576243414169215e-05,
+      "loss": 0.8135,
+      "step": 2057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758245561290625e-05,
+      "loss": 0.7484,
+      "step": 2058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754055468974413e-05,
+      "loss": 0.811,
+      "step": 2059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749863865843687e-05,
+      "loss": 0.7917,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574567075299902e-05,
+      "loss": 0.8176,
+      "step": 2061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5741476131541377e-05,
+      "loss": 0.7976,
+      "step": 2062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737280002572115e-05,
+      "loss": 0.7971,
+      "step": 2063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573308236719299e-05,
+      "loss": 0.7566,
+      "step": 2064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572888322650617e-05,
+      "loss": 0.7981,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724682581614183e-05,
+      "loss": 0.7451,
+      "step": 2066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720480433619983e-05,
+      "loss": 0.7744,
+      "step": 2067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571627678362691e-05,
+      "loss": 0.7476,
+      "step": 2068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712071632738683e-05,
+      "loss": 0.7268,
+      "step": 2069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707864982059442e-05,
+      "loss": 0.7473,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57036568326937e-05,
+      "loss": 0.7798,
+      "step": 2071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699447185746373e-05,
+      "loss": 0.7981,
+      "step": 2072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695236042322767e-05,
+      "loss": 0.8457,
+      "step": 2073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569102340352858e-05,
+      "loss": 0.7874,
+      "step": 2074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686809270469904e-05,
+      "loss": 0.7882,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682593644253228e-05,
+      "loss": 0.7645,
+      "step": 2076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678376525985424e-05,
+      "loss": 0.7051,
+      "step": 2077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567415791677376e-05,
+      "loss": 0.7771,
+      "step": 2078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56699378177259e-05,
+      "loss": 0.7439,
+      "step": 2079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566571622994989e-05,
+      "loss": 0.8193,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566149315455418e-05,
+      "loss": 0.74,
+      "step": 2081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657268592647594e-05,
+      "loss": 0.8051,
+      "step": 2082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653042545339357e-05,
+      "loss": 0.7681,
+      "step": 2083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648815013739084e-05,
+      "loss": 0.733,
+      "step": 2084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644585998956775e-05,
+      "loss": 0.7339,
+      "step": 2085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640355502102826e-05,
+      "loss": 0.7437,
+      "step": 2086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563612352428801e-05,
+      "loss": 0.7852,
+      "step": 2087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5631890066623506e-05,
+      "loss": 0.7568,
+      "step": 2088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627655130220863e-05,
+      "loss": 0.8474,
+      "step": 2089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623418716192042e-05,
+      "loss": 0.7573,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619180825649363e-05,
+      "loss": 0.6454,
+      "step": 2091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561494145970556e-05,
+      "loss": 0.2855,
+      "step": 2092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610700619473734e-05,
+      "loss": 0.3084,
+      "step": 2093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5606458306067382e-05,
+      "loss": 0.7769,
+      "step": 2094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602214520600398e-05,
+      "loss": 0.8042,
+      "step": 2095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559796926418704e-05,
+      "loss": 0.3,
+      "step": 2096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559372253794197e-05,
+      "loss": 0.3135,
+      "step": 2097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5589474342980235e-05,
+      "loss": 0.6705,
+      "step": 2098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558522468041725e-05,
+      "loss": 0.7651,
+      "step": 2099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558097355136884e-05,
+      "loss": 0.7668,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55767209569512e-05,
+      "loss": 0.8335,
+      "step": 2101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572466898280915e-05,
+      "loss": 0.7966,
+      "step": 2102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568211376474952e-05,
+      "loss": 0.8096,
+      "step": 2103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563954392650657e-05,
+      "loss": 0.7595,
+      "step": 2104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559695947925768e-05,
+      "loss": 0.7456,
+      "step": 2105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555543604341841e-05,
+      "loss": 0.7384,
+      "step": 2106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555117468024708e-05,
+      "loss": 0.7924,
+      "step": 2107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546911859530666e-05,
+      "loss": 0.7308,
+      "step": 2108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542647582388438e-05,
+      "loss": 0.7113,
+      "step": 2109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553838184994004e-05,
+      "loss": 0.8018,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534114663305513e-05,
+      "loss": 0.7913,
+      "step": 2111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529846023605264e-05,
+      "loss": 0.8215,
+      "step": 2112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552557593196009e-05,
+      "loss": 0.6633,
+      "step": 2113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521304389491178e-05,
+      "loss": 0.7793,
+      "step": 2114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551703139732007e-05,
+      "loss": 0.7617,
+      "step": 2115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512756956568722e-05,
+      "loss": 0.7111,
+      "step": 2116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508481068359442e-05,
+      "loss": 0.694,
+      "step": 2117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504203733814933e-05,
+      "loss": 0.3134,
+      "step": 2118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499924954058273e-05,
+      "loss": 0.8474,
+      "step": 2119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549564473021292e-05,
+      "loss": 0.7277,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491363063402714e-05,
+      "loss": 0.7441,
+      "step": 2121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548707995475187e-05,
+      "loss": 0.825,
+      "step": 2122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548279540538498e-05,
+      "loss": 0.3003,
+      "step": 2123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478509416427027e-05,
+      "loss": 0.8049,
+      "step": 2124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547422198900335e-05,
+      "loss": 0.7979,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469933124239685e-05,
+      "loss": 0.7712,
+      "step": 2126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546564282326214e-05,
+      "loss": 0.7766,
+      "step": 2127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461351087197197e-05,
+      "loss": 0.8293,
+      "step": 2128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457057917171714e-05,
+      "loss": 0.771,
+      "step": 2129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5452763314312925e-05,
+      "loss": 0.8121,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448467279748453e-05,
+      "loss": 0.7943,
+      "step": 2131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544416981460628e-05,
+      "loss": 0.7666,
+      "step": 2132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439870920014775e-05,
+      "loss": 0.3289,
+      "step": 2133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435570597102672e-05,
+      "loss": 0.7893,
+      "step": 2134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543126884699909e-05,
+      "loss": 0.8145,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426965670833517e-05,
+      "loss": 0.7745,
+      "step": 2136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422661069735817e-05,
+      "loss": 0.743,
+      "step": 2137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418355044836234e-05,
+      "loss": 0.76,
+      "step": 2138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414047597265374e-05,
+      "loss": 0.8115,
+      "step": 2139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540973872815422e-05,
+      "loss": 0.8157,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5405428438634138e-05,
+      "loss": 0.7212,
+      "step": 2141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401116729836857e-05,
+      "loss": 0.7605,
+      "step": 2142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396803602894484e-05,
+      "loss": 0.7833,
+      "step": 2143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392489058939485e-05,
+      "loss": 0.8049,
+      "step": 2144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388173099104727e-05,
+      "loss": 0.7416,
+      "step": 2145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383855724523415e-05,
+      "loss": 0.7784,
+      "step": 2146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537953693632915e-05,
+      "loss": 0.7725,
+      "step": 2147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375216735655892e-05,
+      "loss": 0.772,
+      "step": 2148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370895123637972e-05,
+      "loss": 0.7378,
+      "step": 2149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5366572101410104e-05,
+      "loss": 0.7451,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362247670107353e-05,
+      "loss": 0.822,
+      "step": 2151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535792183086517e-05,
+      "loss": 0.6847,
+      "step": 2152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535359458481937e-05,
+      "loss": 0.7733,
+      "step": 2153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534926593310613e-05,
+      "loss": 0.772,
+      "step": 2154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344935876862014e-05,
+      "loss": 0.8237,
+      "step": 2155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340604417223935e-05,
+      "loss": 0.7415,
+      "step": 2156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336271555329188e-05,
+      "loss": 0.7783,
+      "step": 2157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533193729231543e-05,
+      "loss": 0.3184,
+      "step": 2158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327601629320687e-05,
+      "loss": 0.7213,
+      "step": 2159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5323264567483354e-05,
+      "loss": 0.7466,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531892610794219e-05,
+      "loss": 0.775,
+      "step": 2161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531458625183633e-05,
+      "loss": 0.7687,
+      "step": 2162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310245000305265e-05,
+      "loss": 0.8042,
+      "step": 2163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530590235448885e-05,
+      "loss": 0.7008,
+      "step": 2164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301558315527323e-05,
+      "loss": 0.7759,
+      "step": 2165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529721288456127e-05,
+      "loss": 0.7756,
+      "step": 2166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292866062731654e-05,
+      "loss": 0.7657,
+      "step": 2167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52885178511798e-05,
+      "loss": 0.7291,
+      "step": 2168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528416825104739e-05,
+      "loss": 0.7557,
+      "step": 2169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279817263476485e-05,
+      "loss": 0.834,
+      "step": 2170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275464889609497e-05,
+      "loss": 0.7866,
+      "step": 2171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527111113058921e-05,
+      "loss": 0.8022,
+      "step": 2172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266755987558774e-05,
+      "loss": 0.8213,
+      "step": 2173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526239946166169e-05,
+      "loss": 0.769,
+      "step": 2174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258041554041834e-05,
+      "loss": 0.7542,
+      "step": 2175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253682265843442e-05,
+      "loss": 0.7249,
+      "step": 2176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249321598211109e-05,
+      "loss": 0.8229,
+      "step": 2177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244959552289792e-05,
+      "loss": 0.7754,
+      "step": 2178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5240596129224815e-05,
+      "loss": 0.7568,
+      "step": 2179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236231330161869e-05,
+      "loss": 0.7708,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523186515624698e-05,
+      "loss": 0.7169,
+      "step": 2181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227497608626569e-05,
+      "loss": 0.7705,
+      "step": 2182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223128688447397e-05,
+      "loss": 0.7908,
+      "step": 2183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218758396856585e-05,
+      "loss": 0.7755,
+      "step": 2184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214386735001627e-05,
+      "loss": 0.7577,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521001370403037e-05,
+      "loss": 0.7748,
+      "step": 2186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205639305091007e-05,
+      "loss": 0.792,
+      "step": 2187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201263539332117e-05,
+      "loss": 0.8088,
+      "step": 2188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196886407902617e-05,
+      "loss": 0.2844,
+      "step": 2189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519250791195179e-05,
+      "loss": 0.7732,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188128052629277e-05,
+      "loss": 0.7955,
+      "step": 2191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183746831085076e-05,
+      "loss": 0.7832,
+      "step": 2192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179364248469543e-05,
+      "loss": 0.7876,
+      "step": 2193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517498030593339e-05,
+      "loss": 0.7683,
+      "step": 2194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170595004627693e-05,
+      "loss": 0.8413,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166208345703872e-05,
+      "loss": 0.7546,
+      "step": 2196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161820330313717e-05,
+      "loss": 0.7622,
+      "step": 2197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157430959609367e-05,
+      "loss": 0.738,
+      "step": 2198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153040234743314e-05,
+      "loss": 0.7656,
+      "step": 2199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148648156868414e-05,
+      "loss": 0.7028,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144254727137865e-05,
+      "loss": 0.7269,
+      "step": 2201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139859946705242e-05,
+      "loss": 0.8173,
+      "step": 2202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135463816724451e-05,
+      "loss": 0.7781,
+      "step": 2203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131066338349769e-05,
+      "loss": 0.7373,
+      "step": 2204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126667512735816e-05,
+      "loss": 0.7572,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122267341037571e-05,
+      "loss": 0.7822,
+      "step": 2206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117865824410367e-05,
+      "loss": 0.7537,
+      "step": 2207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511346296400989e-05,
+      "loss": 0.8263,
+      "step": 2208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109058760992173e-05,
+      "loss": 0.7632,
+      "step": 2209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104653216513615e-05,
+      "loss": 0.7625,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100246331730946e-05,
+      "loss": 0.7159,
+      "step": 2211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095838107801273e-05,
+      "loss": 0.7867,
+      "step": 2212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091428545882034e-05,
+      "loss": 0.7534,
+      "step": 2213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508701764713103e-05,
+      "loss": 0.7903,
+      "step": 2214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082605412706408e-05,
+      "loss": 0.7583,
+      "step": 2215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078191843766669e-05,
+      "loss": 0.2883,
+      "step": 2216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073776941470659e-05,
+      "loss": 0.8062,
+      "step": 2217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069360706977576e-05,
+      "loss": 0.7231,
+      "step": 2218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064943141446979e-05,
+      "loss": 0.7734,
+      "step": 2219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060524246038756e-05,
+      "loss": 0.7444,
+      "step": 2220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056104021913163e-05,
+      "loss": 0.8613,
+      "step": 2221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505168247023079e-05,
+      "loss": 0.7336,
+      "step": 2222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047259592152587e-05,
+      "loss": 0.8235,
+      "step": 2223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042835388839848e-05,
+      "loss": 0.3133,
+      "step": 2224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038409861454208e-05,
+      "loss": 0.8054,
+      "step": 2225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033983011157669e-05,
+      "loss": 0.7258,
+      "step": 2226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029554839112557e-05,
+      "loss": 0.749,
+      "step": 2227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025125346481556e-05,
+      "loss": 0.7361,
+      "step": 2228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5020694534427702e-05,
+      "loss": 0.7788,
+      "step": 2229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501626240411437e-05,
+      "loss": 0.7484,
+      "step": 2230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011828956705282e-05,
+      "loss": 0.8271,
+      "step": 2231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007394193364501e-05,
+      "loss": 0.6582,
+      "step": 2232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002958115256454e-05,
+      "loss": 0.7637,
+      "step": 2233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998520723545893e-05,
+      "loss": 0.752,
+      "step": 2234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994082019397922e-05,
+      "loss": 0.7627,
+      "step": 2235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989642003977993e-05,
+      "loss": 0.7532,
+      "step": 2236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985200678451897e-05,
+      "loss": 0.7896,
+      "step": 2237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498075804398577e-05,
+      "loss": 0.7605,
+      "step": 2238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976314101746102e-05,
+      "loss": 0.7839,
+      "step": 2239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4971868852899705e-05,
+      "loss": 0.7627,
+      "step": 2240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967422298613751e-05,
+      "loss": 0.7712,
+      "step": 2241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962974440055755e-05,
+      "loss": 0.79,
+      "step": 2242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958525278393566e-05,
+      "loss": 0.7427,
+      "step": 2243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4954074814795378e-05,
+      "loss": 0.8103,
+      "step": 2244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949623050429729e-05,
+      "loss": 0.8145,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945169986465496e-05,
+      "loss": 0.7144,
+      "step": 2246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940715624071902e-05,
+      "loss": 0.7588,
+      "step": 2247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49362599644185e-05,
+      "loss": 0.7886,
+      "step": 2248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931803008675201e-05,
+      "loss": 0.7303,
+      "step": 2249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927344758012238e-05,
+      "loss": 0.7394,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922885213600196e-05,
+      "loss": 0.7844,
+      "step": 2251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918424376609996e-05,
+      "loss": 0.7595,
+      "step": 2252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49139622482129e-05,
+      "loss": 0.7789,
+      "step": 2253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909498829580502e-05,
+      "loss": 0.7805,
+      "step": 2254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905034121884743e-05,
+      "loss": 0.7559,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900568126297903e-05,
+      "loss": 0.7783,
+      "step": 2256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896100843992593e-05,
+      "loss": 0.7423,
+      "step": 2257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891632276141766e-05,
+      "loss": 0.7542,
+      "step": 2258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887162423918712e-05,
+      "loss": 0.7695,
+      "step": 2259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882691288497065e-05,
+      "loss": 0.7531,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878218871050778e-05,
+      "loss": 0.7983,
+      "step": 2261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487374517275416e-05,
+      "loss": 0.7927,
+      "step": 2262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869270194781848e-05,
+      "loss": 0.7834,
+      "step": 2263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864793938308812e-05,
+      "loss": 0.7255,
+      "step": 2264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4860316404510362e-05,
+      "loss": 0.8102,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855837594562144e-05,
+      "loss": 0.3163,
+      "step": 2266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851357509640136e-05,
+      "loss": 0.8005,
+      "step": 2267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846876150920652e-05,
+      "loss": 0.7974,
+      "step": 2268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842393519580345e-05,
+      "loss": 0.7722,
+      "step": 2269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483790961679619e-05,
+      "loss": 0.7715,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483342444374551e-05,
+      "loss": 0.698,
+      "step": 2271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828938001605951e-05,
+      "loss": 0.7825,
+      "step": 2272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824450291555498e-05,
+      "loss": 0.2906,
+      "step": 2273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819961314772475e-05,
+      "loss": 0.8171,
+      "step": 2274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815471072435515e-05,
+      "loss": 0.7765,
+      "step": 2275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481097956572361e-05,
+      "loss": 0.7495,
+      "step": 2276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806486795816075e-05,
+      "loss": 0.7908,
+      "step": 2277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480199276389255e-05,
+      "loss": 0.7823,
+      "step": 2278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797497471133007e-05,
+      "loss": 0.3053,
+      "step": 2279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793000918717765e-05,
+      "loss": 0.7479,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788503107827452e-05,
+      "loss": 0.8528,
+      "step": 2281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4784004039643041e-05,
+      "loss": 0.7922,
+      "step": 2282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779503715345827e-05,
+      "loss": 0.7605,
+      "step": 2283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775002136117445e-05,
+      "loss": 0.7643,
+      "step": 2284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770499303139845e-05,
+      "loss": 0.7371,
+      "step": 2285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476599521759532e-05,
+      "loss": 0.7123,
+      "step": 2286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761489880666483e-05,
+      "loss": 0.7705,
+      "step": 2287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475698329353628e-05,
+      "loss": 0.762,
+      "step": 2288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752475457387983e-05,
+      "loss": 0.7161,
+      "step": 2289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747966373405192e-05,
+      "loss": 0.7723,
+      "step": 2290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743456042771838e-05,
+      "loss": 0.749,
+      "step": 2291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738944466672173e-05,
+      "loss": 0.7944,
+      "step": 2292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734431646290783e-05,
+      "loss": 0.7615,
+      "step": 2293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729917582812575e-05,
+      "loss": 0.7404,
+      "step": 2294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725402277422787e-05,
+      "loss": 0.7419,
+      "step": 2295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472088573130698e-05,
+      "loss": 0.7081,
+      "step": 2296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471636794565104e-05,
+      "loss": 0.6974,
+      "step": 2297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711848921641183e-05,
+      "loss": 0.781,
+      "step": 2298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707328660463944e-05,
+      "loss": 0.7367,
+      "step": 2299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470280716330619e-05,
+      "loss": 0.733,
+      "step": 2300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698284431355105e-05,
+      "loss": 0.748,
+      "step": 2301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46937604657982e-05,
+      "loss": 0.7747,
+      "step": 2302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689235267823318e-05,
+      "loss": 0.7737,
+      "step": 2303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684708838618616e-05,
+      "loss": 0.7261,
+      "step": 2304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680181179372568e-05,
+      "loss": 0.7334,
+      "step": 2305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675652291273988e-05,
+      "loss": 0.7939,
+      "step": 2306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4671122175512e-05,
+      "loss": 0.7307,
+      "step": 2307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466659083327606e-05,
+      "loss": 0.712,
+      "step": 2308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662058265755935e-05,
+      "loss": 0.74,
+      "step": 2309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657524474141722e-05,
+      "loss": 0.6953,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652989459623838e-05,
+      "loss": 0.7681,
+      "step": 2311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648453223393016e-05,
+      "loss": 0.7642,
+      "step": 2312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643915766640313e-05,
+      "loss": 0.764,
+      "step": 2313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639377090557114e-05,
+      "loss": 0.7797,
+      "step": 2314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634837196335112e-05,
+      "loss": 0.79,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630296085166327e-05,
+      "loss": 0.7874,
+      "step": 2316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625753758243093e-05,
+      "loss": 0.7302,
+      "step": 2317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621210216758073e-05,
+      "loss": 0.7712,
+      "step": 2318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616665461904233e-05,
+      "loss": 0.7578,
+      "step": 2319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612119494874878e-05,
+      "loss": 0.7758,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607572316863618e-05,
+      "loss": 0.7295,
+      "step": 2321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460302392906438e-05,
+      "loss": 0.7029,
+      "step": 2322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598474332671412e-05,
+      "loss": 0.7843,
+      "step": 2323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593923528879282e-05,
+      "loss": 0.2999,
+      "step": 2324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589371518882873e-05,
+      "loss": 0.771,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458481830387738e-05,
+      "loss": 0.8,
+      "step": 2326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580263885058323e-05,
+      "loss": 0.2933,
+      "step": 2327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575708263621536e-05,
+      "loss": 0.7205,
+      "step": 2328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571151440763158e-05,
+      "loss": 0.8397,
+      "step": 2329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566593417679659e-05,
+      "loss": 0.7786,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562034195567815e-05,
+      "loss": 0.7017,
+      "step": 2331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455747377562472e-05,
+      "loss": 0.702,
+      "step": 2332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552912159047778e-05,
+      "loss": 0.275,
+      "step": 2333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4548349347034716e-05,
+      "loss": 0.792,
+      "step": 2334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543785340783562e-05,
+      "loss": 0.7988,
+      "step": 2335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539220141492672e-05,
+      "loss": 0.73,
+      "step": 2336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534653750360704e-05,
+      "loss": 0.7391,
+      "step": 2337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530086168586636e-05,
+      "loss": 0.7515,
+      "step": 2338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525517397369756e-05,
+      "loss": 0.8235,
+      "step": 2339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520947437909661e-05,
+      "loss": 0.7125,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516376291406266e-05,
+      "loss": 0.8494,
+      "step": 2341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511803959059792e-05,
+      "loss": 0.8066,
+      "step": 2342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507230442070777e-05,
+      "loss": 0.7661,
+      "step": 2343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502655741640066e-05,
+      "loss": 0.8093,
+      "step": 2344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498079858968819e-05,
+      "loss": 0.7781,
+      "step": 2345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493502795258499e-05,
+      "loss": 0.7909,
+      "step": 2346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488924551710884e-05,
+      "loss": 0.7883,
+      "step": 2347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484345129528064e-05,
+      "loss": 0.7183,
+      "step": 2348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479764529912431e-05,
+      "loss": 0.7678,
+      "step": 2349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475182754066696e-05,
+      "loss": 0.7878,
+      "step": 2350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470599803193875e-05,
+      "loss": 0.824,
+      "step": 2351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446601567849728e-05,
+      "loss": 0.7498,
+      "step": 2352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4461430381180556e-05,
+      "loss": 0.7786,
+      "step": 2353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456843912447638e-05,
+      "loss": 0.7875,
+      "step": 2354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445225627350277e-05,
+      "loss": 0.6833,
+      "step": 2355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447667465550503e-05,
+      "loss": 0.7727,
+      "step": 2356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443077489795712e-05,
+      "loss": 0.7612,
+      "step": 2357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443848634744355e-05,
+      "loss": 0.7762,
+      "step": 2358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433894039699497e-05,
+      "loss": 0.726,
+      "step": 2359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4429300567769333e-05,
+      "loss": 0.7383,
+      "step": 2360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424705932859145e-05,
+      "loss": 0.7902,
+      "step": 2361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442011013617532e-05,
+      "loss": 0.7937,
+      "step": 2362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415513178924558e-05,
+      "loss": 0.823,
+      "step": 2363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441091506231386e-05,
+      "loss": 0.668,
+      "step": 2364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406315787550527e-05,
+      "loss": 0.7285,
+      "step": 2365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401715355842169e-05,
+      "loss": 0.7617,
+      "step": 2366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397113768396706e-05,
+      "loss": 0.7308,
+      "step": 2367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392511026422349e-05,
+      "loss": 0.7888,
+      "step": 2368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387907131127616e-05,
+      "loss": 0.7865,
+      "step": 2369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383302083721333e-05,
+      "loss": 0.7688,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378695885412623e-05,
+      "loss": 0.7332,
+      "step": 2371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374088537410918e-05,
+      "loss": 0.8008,
+      "step": 2372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369480040925936e-05,
+      "loss": 0.7292,
+      "step": 2373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364870397167721e-05,
+      "loss": 0.7832,
+      "step": 2374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360259607346597e-05,
+      "loss": 0.74,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43556476726732e-05,
+      "loss": 0.7974,
+      "step": 2376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351034594358457e-05,
+      "loss": 0.7649,
+      "step": 2377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434642037361361e-05,
+      "loss": 0.7495,
+      "step": 2378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434180501165019e-05,
+      "loss": 0.8013,
+      "step": 2379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337188509680024e-05,
+      "loss": 0.7432,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332570868915256e-05,
+      "loss": 0.7759,
+      "step": 2381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327952090568311e-05,
+      "loss": 0.7642,
+      "step": 2382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323332175851915e-05,
+      "loss": 0.7667,
+      "step": 2383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4318711125979108e-05,
+      "loss": 0.7043,
+      "step": 2384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314088942163207e-05,
+      "loss": 0.7368,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430946562561784e-05,
+      "loss": 0.7362,
+      "step": 2386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304841177556928e-05,
+      "loss": 0.7559,
+      "step": 2387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300215599194692e-05,
+      "loss": 0.7605,
+      "step": 2388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295588891745644e-05,
+      "loss": 0.8081,
+      "step": 2389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290961056424598e-05,
+      "loss": 0.3224,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286332094446664e-05,
+      "loss": 0.7791,
+      "step": 2391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281702007027248e-05,
+      "loss": 0.8162,
+      "step": 2392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277070795382042e-05,
+      "loss": 0.7683,
+      "step": 2393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272438460727047e-05,
+      "loss": 0.7297,
+      "step": 2394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267805004278553e-05,
+      "loss": 0.744,
+      "step": 2395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426317042725314e-05,
+      "loss": 0.7294,
+      "step": 2396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258534730867688e-05,
+      "loss": 0.7399,
+      "step": 2397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253897916339374e-05,
+      "loss": 0.6699,
+      "step": 2398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249259984885657e-05,
+      "loss": 0.8108,
+      "step": 2399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244620937724301e-05,
+      "loss": 0.6945,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239980776073355e-05,
+      "loss": 0.7166,
+      "step": 2401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235339501151169e-05,
+      "loss": 0.7837,
+      "step": 2402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4230697114176371e-05,
+      "loss": 0.7247,
+      "step": 2403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42260536163679e-05,
+      "loss": 0.8076,
+      "step": 2404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221409008944972e-05,
+      "loss": 0.7979,
+      "step": 2405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216763293127096e-05,
+      "loss": 0.7346,
+      "step": 2406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212116470134083e-05,
+      "loss": 0.7585,
+      "step": 2407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420746854118602e-05,
+      "loss": 0.7861,
+      "step": 2408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202819507503293e-05,
+      "loss": 0.7764,
+      "step": 2409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419816937030658e-05,
+      "loss": 0.7892,
+      "step": 2410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193518130816839e-05,
+      "loss": 0.7864,
+      "step": 2411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418886579025533e-05,
+      "loss": 0.8154,
+      "step": 2412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184212349843588e-05,
+      "loss": 0.772,
+      "step": 2413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179557810803449e-05,
+      "loss": 0.6923,
+      "step": 2414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4174902174357032e-05,
+      "loss": 0.7792,
+      "step": 2415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170245441726747e-05,
+      "loss": 0.7294,
+      "step": 2416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165587614135285e-05,
+      "loss": 0.7672,
+      "step": 2417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160928692805636e-05,
+      "loss": 0.801,
+      "step": 2418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156268678961063e-05,
+      "loss": 0.8325,
+      "step": 2419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151607573825126e-05,
+      "loss": 0.7649,
+      "step": 2420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146945378621672e-05,
+      "loss": 0.7693,
+      "step": 2421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142282094574831e-05,
+      "loss": 0.3067,
+      "step": 2422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137617722909013e-05,
+      "loss": 0.2719,
+      "step": 2423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132952264848924e-05,
+      "loss": 0.715,
+      "step": 2424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412828572161955e-05,
+      "loss": 0.75,
+      "step": 2425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123618094446165e-05,
+      "loss": 0.8005,
+      "step": 2426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118949384554319e-05,
+      "loss": 0.7119,
+      "step": 2427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411427959316986e-05,
+      "loss": 0.7008,
+      "step": 2428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4109608721518905e-05,
+      "loss": 0.8247,
+      "step": 2429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104936770827868e-05,
+      "loss": 0.7515,
+      "step": 2430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100263742323439e-05,
+      "loss": 0.7634,
+      "step": 2431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4095589637232592e-05,
+      "loss": 0.7375,
+      "step": 2432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090914456782586e-05,
+      "loss": 0.7834,
+      "step": 2433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4086238202200957e-05,
+      "loss": 0.772,
+      "step": 2434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081560874715527e-05,
+      "loss": 0.7185,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076882475554403e-05,
+      "loss": 0.7534,
+      "step": 2436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072203005945965e-05,
+      "loss": 0.7917,
+      "step": 2437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067522467118884e-05,
+      "loss": 0.76,
+      "step": 2438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062840860302101e-05,
+      "loss": 0.7922,
+      "step": 2439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058158186724846e-05,
+      "loss": 0.7974,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053474447616627e-05,
+      "loss": 0.8774,
+      "step": 2441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048789644207225e-05,
+      "loss": 0.7245,
+      "step": 2442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044103777726709e-05,
+      "loss": 0.7498,
+      "step": 2443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039416849405431e-05,
+      "loss": 0.7747,
+      "step": 2444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403472886047401e-05,
+      "loss": 0.7708,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030039812163343e-05,
+      "loss": 0.7047,
+      "step": 2446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402534970570462e-05,
+      "loss": 0.7236,
+      "step": 2447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402065854232929e-05,
+      "loss": 0.8577,
+      "step": 2448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015966323269099e-05,
+      "loss": 0.8608,
+      "step": 2449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011273049756054e-05,
+      "loss": 0.7112,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4006578723022449e-05,
+      "loss": 0.7534,
+      "step": 2451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001883344300847e-05,
+      "loss": 0.7637,
+      "step": 2452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399718691482409e-05,
+      "loss": 0.7559,
+      "step": 2453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992489435825303e-05,
+      "loss": 0.7657,
+      "step": 2454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987790908537874e-05,
+      "loss": 0.7277,
+      "step": 2455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983091334195478e-05,
+      "loss": 0.7865,
+      "step": 2456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978390714032053e-05,
+      "loss": 0.8083,
+      "step": 2457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973689049281825e-05,
+      "loss": 0.7454,
+      "step": 2458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396898634117928e-05,
+      "loss": 0.7585,
+      "step": 2459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396428259095919e-05,
+      "loss": 0.7686,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959577799856598e-05,
+      "loss": 0.8152,
+      "step": 2461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954871969106815e-05,
+      "loss": 0.708,
+      "step": 2462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950165099945427e-05,
+      "loss": 0.7739,
+      "step": 2463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945457193608299e-05,
+      "loss": 0.7585,
+      "step": 2464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940748251331557e-05,
+      "loss": 0.7295,
+      "step": 2465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393603827435161e-05,
+      "loss": 0.7782,
+      "step": 2466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931327263905128e-05,
+      "loss": 0.6864,
+      "step": 2467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926615221229065e-05,
+      "loss": 0.7112,
+      "step": 2468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392190214756064e-05,
+      "loss": 0.6681,
+      "step": 2469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391718804413733e-05,
+      "loss": 0.8066,
+      "step": 2470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391247291219691e-05,
+      "loss": 0.7725,
+      "step": 2471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39077567529774e-05,
+      "loss": 0.7742,
+      "step": 2472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39030395677171e-05,
+      "loss": 0.7302,
+      "step": 2473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898321357654578e-05,
+      "loss": 0.7961,
+      "step": 2474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893602124028677e-05,
+      "loss": 0.8152,
+      "step": 2475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888881868078499e-05,
+      "loss": 0.7861,
+      "step": 2476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884160591043416e-05,
+      "loss": 0.7236,
+      "step": 2477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879438294163077e-05,
+      "loss": 0.7754,
+      "step": 2478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874714978677391e-05,
+      "loss": 0.7556,
+      "step": 2479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869990645826534e-05,
+      "loss": 0.7473,
+      "step": 2480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865265296850952e-05,
+      "loss": 0.792,
+      "step": 2481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860538932991356e-05,
+      "loss": 0.7797,
+      "step": 2482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855811555488727e-05,
+      "loss": 0.3066,
+      "step": 2483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385108316558431e-05,
+      "loss": 0.7924,
+      "step": 2484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384635376451961e-05,
+      "loss": 0.7363,
+      "step": 2485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841623353536412e-05,
+      "loss": 0.7627,
+      "step": 2486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836891933876748e-05,
+      "loss": 0.7686,
+      "step": 2487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383215950678293e-05,
+      "loss": 0.781,
+      "step": 2488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3827426073497525e-05,
+      "loss": 0.7319,
+      "step": 2489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382269163526337e-05,
+      "loss": 0.7887,
+      "step": 2490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817956193323561e-05,
+      "loss": 0.2903,
+      "step": 2491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813219748921462e-05,
+      "loss": 0.7623,
+      "step": 2492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808482303300703e-05,
+      "loss": 0.7419,
+      "step": 2493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803743857705162e-05,
+      "loss": 0.7749,
+      "step": 2494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799004413378996e-05,
+      "loss": 0.7505,
+      "step": 2495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794263971566618e-05,
+      "loss": 0.7522,
+      "step": 2496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789522533512698e-05,
+      "loss": 0.7617,
+      "step": 2497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784780100462183e-05,
+      "loss": 0.7703,
+      "step": 2498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378003667366026e-05,
+      "loss": 0.7797,
+      "step": 2499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377529225435239e-05,
+      "loss": 0.7754,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770546843784295e-05,
+      "loss": 0.2766,
+      "step": 2501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765800443201956e-05,
+      "loss": 0.7869,
+      "step": 2502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761053053851608e-05,
+      "loss": 0.7122,
+      "step": 2503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756304676979753e-05,
+      "loss": 0.7998,
+      "step": 2504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751555313833146e-05,
+      "loss": 0.7783,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746804965658808e-05,
+      "loss": 0.7634,
+      "step": 2506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742053633704013e-05,
+      "loss": 0.8047,
+      "step": 2507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737301319216293e-05,
+      "loss": 0.7593,
+      "step": 2508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732548023443448e-05,
+      "loss": 0.2942,
+      "step": 2509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372779374763352e-05,
+      "loss": 0.7212,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723038493034818e-05,
+      "loss": 0.7759,
+      "step": 2511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718282260895912e-05,
+      "loss": 0.7676,
+      "step": 2512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713525052465616e-05,
+      "loss": 0.8169,
+      "step": 2513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708766868993003e-05,
+      "loss": 0.783,
+      "step": 2514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704007711727418e-05,
+      "loss": 0.7856,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699247581918446e-05,
+      "loss": 0.7794,
+      "step": 2516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694486480815927e-05,
+      "loss": 0.3081,
+      "step": 2517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689724409669964e-05,
+      "loss": 0.8357,
+      "step": 2518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684961369730912e-05,
+      "loss": 0.8232,
+      "step": 2519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680197362249374e-05,
+      "loss": 0.7202,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675432388476213e-05,
+      "loss": 0.7614,
+      "step": 2521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670666449662556e-05,
+      "loss": 0.7092,
+      "step": 2522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665899547059757e-05,
+      "loss": 0.7328,
+      "step": 2523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661131681919448e-05,
+      "loss": 0.7192,
+      "step": 2524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3656362855493503e-05,
+      "loss": 0.7588,
+      "step": 2525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651593069034048e-05,
+      "loss": 0.7437,
+      "step": 2526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646822323793465e-05,
+      "loss": 0.746,
+      "step": 2527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642050621024383e-05,
+      "loss": 0.7499,
+      "step": 2528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637277961979687e-05,
+      "loss": 0.8162,
+      "step": 2529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363250434791251e-05,
+      "loss": 0.693,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627729780076234e-05,
+      "loss": 0.7816,
+      "step": 2531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622954259724499e-05,
+      "loss": 0.8264,
+      "step": 2532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618177788111189e-05,
+      "loss": 0.7891,
+      "step": 2533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613400366490437e-05,
+      "loss": 0.7168,
+      "step": 2534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3608621996116629e-05,
+      "loss": 0.7439,
+      "step": 2535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603842678244399e-05,
+      "loss": 0.7585,
+      "step": 2536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359906241412863e-05,
+      "loss": 0.7417,
+      "step": 2537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359428120502445e-05,
+      "loss": 0.2853,
+      "step": 2538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589499052187238e-05,
+      "loss": 0.3044,
+      "step": 2539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584715956872626e-05,
+      "loss": 0.8003,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357993192033648e-05,
+      "loss": 0.7356,
+      "step": 2541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575146943834928e-05,
+      "loss": 0.7605,
+      "step": 2542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570361028624336e-05,
+      "loss": 0.3037,
+      "step": 2543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356557417596132e-05,
+      "loss": 0.826,
+      "step": 2544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560786387102736e-05,
+      "loss": 0.2791,
+      "step": 2545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555997663305698e-05,
+      "loss": 0.7661,
+      "step": 2546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551208005827551e-05,
+      "loss": 0.8401,
+      "step": 2547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546417415925894e-05,
+      "loss": 0.7207,
+      "step": 2548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541625894858571e-05,
+      "loss": 0.7551,
+      "step": 2549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353683344388367e-05,
+      "loss": 0.7295,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532040064259515e-05,
+      "loss": 0.822,
+      "step": 2551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352724575724469e-05,
+      "loss": 0.751,
+      "step": 2552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522450524098005e-05,
+      "loss": 0.73,
+      "step": 2553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517654366078523e-05,
+      "loss": 0.7539,
+      "step": 2554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351285728444555e-05,
+      "loss": 0.7537,
+      "step": 2555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3508059280458633e-05,
+      "loss": 0.2896,
+      "step": 2556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3503260355377556e-05,
+      "loss": 0.7771,
+      "step": 2557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498460510462356e-05,
+      "loss": 0.7593,
+      "step": 2558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.34936597469733e-05,
+      "loss": 0.7301,
+      "step": 2559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488858066170904e-05,
+      "loss": 0.7245,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348405546931592e-05,
+      "loss": 0.741,
+      "step": 2561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3479251957669344e-05,
+      "loss": 0.8132,
+      "step": 2562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3474447532492413e-05,
+      "loss": 0.7908,
+      "step": 2563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469642195046597e-05,
+      "loss": 0.7949,
+      "step": 2564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346483594659361e-05,
+      "loss": 0.776,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460028788395412e-05,
+      "loss": 0.8032,
+      "step": 2566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345522072171419e-05,
+      "loss": 0.7386,
+      "step": 2567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450411747812374e-05,
+      "loss": 0.7578,
+      "step": 2568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344560186795264e-05,
+      "loss": 0.7378,
+      "step": 2569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440791083397885e-05,
+      "loss": 0.7385,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343597939541126e-05,
+      "loss": 0.8025,
+      "step": 2571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431166805256145e-05,
+      "loss": 0.2923,
+      "step": 2572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426353314196159e-05,
+      "loss": 0.8218,
+      "step": 2573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421538923495156e-05,
+      "loss": 0.7478,
+      "step": 2574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341672363441723e-05,
+      "loss": 0.7351,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411907448226703e-05,
+      "loss": 0.2997,
+      "step": 2576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3407090366188142e-05,
+      "loss": 0.7361,
+      "step": 2577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402272389566342e-05,
+      "loss": 0.783,
+      "step": 2578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339745351962634e-05,
+      "loss": 0.7797,
+      "step": 2579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3392633757633398e-05,
+      "loss": 0.7922,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387813104853022e-05,
+      "loss": 0.7916,
+      "step": 2581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382991562550946e-05,
+      "loss": 0.2971,
+      "step": 2582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337816913199314e-05,
+      "loss": 0.7844,
+      "step": 2583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373345814445803e-05,
+      "loss": 0.294,
+      "step": 2584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368521611175371e-05,
+      "loss": 0.3071,
+      "step": 2585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363696523448517e-05,
+      "loss": 0.7434,
+      "step": 2586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358870552532137e-05,
+      "loss": 0.7581,
+      "step": 2587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335404369969336e-05,
+      "loss": 0.7822,
+      "step": 2588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349215966199553e-05,
+      "loss": 0.7615,
+      "step": 2589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344387353318311e-05,
+      "loss": 0.7188,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339557862317461e-05,
+      "loss": 0.7338,
+      "step": 2591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333472749446505e-05,
+      "loss": 0.7346,
+      "step": 2592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329896251029374e-05,
+      "loss": 0.8282,
+      "step": 2593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325064133278949e-05,
+      "loss": 0.8154,
+      "step": 2594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320231142482514e-05,
+      "loss": 0.2827,
+      "step": 2595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315397279909049e-05,
+      "loss": 0.738,
+      "step": 2596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310562546827757e-05,
+      "loss": 0.7756,
+      "step": 2597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305726944508069e-05,
+      "loss": 0.7351,
+      "step": 2598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330089047421965e-05,
+      "loss": 0.7708,
+      "step": 2599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296053137232383e-05,
+      "loss": 0.77,
+      "step": 2600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291214934816385e-05,
+      "loss": 0.7262,
+      "step": 2601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286375868242005e-05,
+      "loss": 0.7656,
+      "step": 2602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281535938779808e-05,
+      "loss": 0.7529,
+      "step": 2603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276695147700593e-05,
+      "loss": 0.7233,
+      "step": 2604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271853496275381e-05,
+      "loss": 0.792,
+      "step": 2605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267010985775424e-05,
+      "loss": 0.7815,
+      "step": 2606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262167617472195e-05,
+      "loss": 0.7878,
+      "step": 2607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257323392637392e-05,
+      "loss": 0.7328,
+      "step": 2608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325247831254294e-05,
+      "loss": 0.7922,
+      "step": 2609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247632378460995e-05,
+      "loss": 0.7947,
+      "step": 2610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324278559166392e-05,
+      "loss": 0.7612,
+      "step": 2611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237937953424323e-05,
+      "loss": 0.7164,
+      "step": 2612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233089465015015e-05,
+      "loss": 0.7913,
+      "step": 2613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228240127709047e-05,
+      "loss": 0.7458,
+      "step": 2614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223389942779679e-05,
+      "loss": 0.7766,
+      "step": 2615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218538911500411e-05,
+      "loss": 0.7201,
+      "step": 2616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213687035144948e-05,
+      "loss": 0.7888,
+      "step": 2617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208834314987222e-05,
+      "loss": 0.7349,
+      "step": 2618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3203980752301394e-05,
+      "loss": 0.8015,
+      "step": 2619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3199126348361835e-05,
+      "loss": 0.7439,
+      "step": 2620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194271104443146e-05,
+      "loss": 0.7528,
+      "step": 2621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3189415021820145e-05,
+      "loss": 0.7495,
+      "step": 2622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3184558101767869e-05,
+      "loss": 0.7142,
+      "step": 2623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179700345561575e-05,
+      "loss": 0.7587,
+      "step": 2624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174841754476737e-05,
+      "loss": 0.7279,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3169982329789061e-05,
+      "loss": 0.7404,
+      "step": 2626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165122072774459e-05,
+      "loss": 0.7222,
+      "step": 2627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160260984709061e-05,
+      "loss": 0.7983,
+      "step": 2628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155399066869223e-05,
+      "loss": 0.7354,
+      "step": 2629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150536320531517e-05,
+      "loss": 0.7119,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145672746972727e-05,
+      "loss": 0.7238,
+      "step": 2631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140808347469857e-05,
+      "loss": 0.791,
+      "step": 2632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135943123300135e-05,
+      "loss": 0.7722,
+      "step": 2633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131077075740994e-05,
+      "loss": 0.7195,
+      "step": 2634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126210206070091e-05,
+      "loss": 0.7483,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121342515565295e-05,
+      "loss": 0.792,
+      "step": 2636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3116474005504692e-05,
+      "loss": 0.7759,
+      "step": 2637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111604677166586e-05,
+      "loss": 0.2963,
+      "step": 2638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3106734531829486e-05,
+      "loss": 0.76,
+      "step": 2639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101863570772133e-05,
+      "loss": 0.7911,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096991795273463e-05,
+      "loss": 0.7194,
+      "step": 2641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092119206612637e-05,
+      "loss": 0.7712,
+      "step": 2642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087245806069028e-05,
+      "loss": 0.6995,
+      "step": 2643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082371594922221e-05,
+      "loss": 0.7773,
+      "step": 2644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077496574452013e-05,
+      "loss": 0.7831,
+      "step": 2645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072620745938419e-05,
+      "loss": 0.7566,
+      "step": 2646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067744110661653e-05,
+      "loss": 0.7266,
+      "step": 2647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062866669902156e-05,
+      "loss": 0.8298,
+      "step": 2648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057988424940573e-05,
+      "loss": 0.7583,
+      "step": 2649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053109377057762e-05,
+      "loss": 0.6854,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048229527534789e-05,
+      "loss": 0.7683,
+      "step": 2651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043348877652933e-05,
+      "loss": 0.7727,
+      "step": 2652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038467428693682e-05,
+      "loss": 0.6718,
+      "step": 2653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3033585181938736e-05,
+      "loss": 0.7915,
+      "step": 2654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302870213867e-05,
+      "loss": 0.7865,
+      "step": 2655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023818300169598e-05,
+      "loss": 0.792,
+      "step": 2656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018933667719847e-05,
+      "loss": 0.7283,
+      "step": 2657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014048242603288e-05,
+      "loss": 0.7249,
+      "step": 2658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300916202610266e-05,
+      "loss": 0.7886,
+      "step": 2659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004275019500914e-05,
+      "loss": 0.7485,
+      "step": 2660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999387224081208e-05,
+      "loss": 0.7456,
+      "step": 2661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994498641126905e-05,
+      "loss": 0.7194,
+      "step": 2662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989609271921579e-05,
+      "loss": 0.726,
+      "step": 2663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298471911774901e-05,
+      "loss": 0.7458,
+      "step": 2664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979828179893175e-05,
+      "loss": 0.7582,
+      "step": 2665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974936459638272e-05,
+      "loss": 0.7682,
+      "step": 2666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297004395826869e-05,
+      "loss": 0.7048,
+      "step": 2667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965150677069031e-05,
+      "loss": 0.7041,
+      "step": 2668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960256617324105e-05,
+      "loss": 0.8049,
+      "step": 2669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955361780318916e-05,
+      "loss": 0.8071,
+      "step": 2670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295046616733868e-05,
+      "loss": 0.7847,
+      "step": 2671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294556977966881e-05,
+      "loss": 0.6638,
+      "step": 2672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940672618594934e-05,
+      "loss": 0.751,
+      "step": 2673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935774685402874e-05,
+      "loss": 0.7944,
+      "step": 2674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930875981378653e-05,
+      "loss": 0.7061,
+      "step": 2675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925976507808502e-05,
+      "loss": 0.7881,
+      "step": 2676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921076265978854e-05,
+      "loss": 0.699,
+      "step": 2677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291617525717634e-05,
+      "loss": 0.7566,
+      "step": 2678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911273482687789e-05,
+      "loss": 0.7736,
+      "step": 2679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906370943800246e-05,
+      "loss": 0.7448,
+      "step": 2680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901467641800945e-05,
+      "loss": 0.7208,
+      "step": 2681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2896563577977314e-05,
+      "loss": 0.7566,
+      "step": 2682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891658753617e-05,
+      "loss": 0.7477,
+      "step": 2683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288675317000783e-05,
+      "loss": 0.7446,
+      "step": 2684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881846828437847e-05,
+      "loss": 0.3371,
+      "step": 2685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876939730195278e-05,
+      "loss": 0.6998,
+      "step": 2686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872031876568562e-05,
+      "loss": 0.7866,
+      "step": 2687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867123268846328e-05,
+      "loss": 0.7505,
+      "step": 2688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862213908317405e-05,
+      "loss": 0.7224,
+      "step": 2689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857303796270821e-05,
+      "loss": 0.6875,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852392933995801e-05,
+      "loss": 0.7819,
+      "step": 2691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847481322781766e-05,
+      "loss": 0.8149,
+      "step": 2692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284256896391833e-05,
+      "loss": 0.71,
+      "step": 2693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837655858695313e-05,
+      "loss": 0.7732,
+      "step": 2694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2832742008402721e-05,
+      "loss": 0.8013,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827827414330762e-05,
+      "loss": 0.7461,
+      "step": 2696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822912077769839e-05,
+      "loss": 0.7534,
+      "step": 2697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817996000010546e-05,
+      "loss": 0.7644,
+      "step": 2698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813079182343673e-05,
+      "loss": 0.7856,
+      "step": 2699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808161626060205e-05,
+      "loss": 0.6692,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803243332451323e-05,
+      "loss": 0.7292,
+      "step": 2701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798324302808398e-05,
+      "loss": 0.7761,
+      "step": 2702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793404538422998e-05,
+      "loss": 0.3034,
+      "step": 2703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788484040586877e-05,
+      "loss": 0.7859,
+      "step": 2704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783562810591994e-05,
+      "loss": 0.7349,
+      "step": 2705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778640849730484e-05,
+      "loss": 0.7889,
+      "step": 2706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773718159294685e-05,
+      "loss": 0.7371,
+      "step": 2707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768794740577129e-05,
+      "loss": 0.8005,
+      "step": 2708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763870594870529e-05,
+      "loss": 0.791,
+      "step": 2709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758945723467795e-05,
+      "loss": 0.72,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275402012766203e-05,
+      "loss": 0.8186,
+      "step": 2711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274909380874652e-05,
+      "loss": 0.7102,
+      "step": 2712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2744166768014746e-05,
+      "loss": 0.7244,
+      "step": 2713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739239006760378e-05,
+      "loss": 0.7598,
+      "step": 2714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273431052627727e-05,
+      "loss": 0.679,
+      "step": 2715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729381327859476e-05,
+      "loss": 0.7123,
+      "step": 2716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724451412801232e-05,
+      "loss": 0.7557,
+      "step": 2717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719520782396955e-05,
+      "loss": 0.7673,
+      "step": 2718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714589437941257e-05,
+      "loss": 0.7244,
+      "step": 2719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709657380728943e-05,
+      "loss": 0.8101,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704724612054999e-05,
+      "loss": 0.7126,
+      "step": 2721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699791133214597e-05,
+      "loss": 0.761,
+      "step": 2722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694856945503096e-05,
+      "loss": 0.7188,
+      "step": 2723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689922050216042e-05,
+      "loss": 0.7803,
+      "step": 2724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684986448649164e-05,
+      "loss": 0.7456,
+      "step": 2725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680050142098382e-05,
+      "loss": 0.7803,
+      "step": 2726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675113131859798e-05,
+      "loss": 0.7673,
+      "step": 2727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26701754192297e-05,
+      "loss": 0.7529,
+      "step": 2728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665237005504554e-05,
+      "loss": 0.745,
+      "step": 2729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660297891981022e-05,
+      "loss": 0.6816,
+      "step": 2730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265535807995594e-05,
+      "loss": 0.7495,
+      "step": 2731
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2650417570726332e-05,
+      "loss": 0.8169,
+      "step": 2732
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26454763655894e-05,
+      "loss": 0.7734,
+      "step": 2733
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640534465842537e-05,
+      "loss": 0.7429,
+      "step": 2734
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263559187278331e-05,
+      "loss": 0.78,
+      "step": 2735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630648587709474e-05,
+      "loss": 0.792,
+      "step": 2736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625704611918964e-05,
+      "loss": 0.7281,
+      "step": 2737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620759946709893e-05,
+      "loss": 0.7233,
+      "step": 2738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261581459338056e-05,
+      "loss": 0.7512,
+      "step": 2739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610868553229443e-05,
+      "loss": 0.7412,
+      "step": 2740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605921827555196e-05,
+      "loss": 0.7844,
+      "step": 2741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600974417656661e-05,
+      "loss": 0.7622,
+      "step": 2742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596026324832853e-05,
+      "loss": 0.8069,
+      "step": 2743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259107755038297e-05,
+      "loss": 0.7288,
+      "step": 2744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586128095606389e-05,
+      "loss": 0.7864,
+      "step": 2745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581177961802664e-05,
+      "loss": 0.7635,
+      "step": 2746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576227150271524e-05,
+      "loss": 0.7202,
+      "step": 2747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571275662312886e-05,
+      "loss": 0.7803,
+      "step": 2748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566323499226831e-05,
+      "loss": 0.7522,
+      "step": 2749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.256137066231363e-05,
+      "loss": 0.7395,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2556417152873723e-05,
+      "loss": 0.8171,
+      "step": 2751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2551462972207731e-05,
+      "loss": 0.7697,
+      "step": 2752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546508121616445e-05,
+      "loss": 0.7191,
+      "step": 2753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541552602400839e-05,
+      "loss": 0.719,
+      "step": 2754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253659641586206e-05,
+      "loss": 0.7437,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253163956330143e-05,
+      "loss": 0.7246,
+      "step": 2756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526682046020446e-05,
+      "loss": 0.8047,
+      "step": 2757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521723865320784e-05,
+      "loss": 0.7781,
+      "step": 2758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2516765022504277e-05,
+      "loss": 0.7474,
+      "step": 2759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2511805518872952e-05,
+      "loss": 0.7786,
+      "step": 2760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506845355729003e-05,
+      "loss": 0.761,
+      "step": 2761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501884534374795e-05,
+      "loss": 0.7378,
+      "step": 2762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2496923056112863e-05,
+      "loss": 0.7532,
+      "step": 2763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491960922245928e-05,
+      "loss": 0.803,
+      "step": 2764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486998134076864e-05,
+      "loss": 0.7913,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482034692908726e-05,
+      "loss": 0.71,
+      "step": 2766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247707060004475e-05,
+      "loss": 0.7305,
+      "step": 2767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2472105856788326e-05,
+      "loss": 0.6675,
+      "step": 2768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467140464443026e-05,
+      "loss": 0.8074,
+      "step": 2769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462174424312586e-05,
+      "loss": 0.7391,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457207737700918e-05,
+      "loss": 0.7469,
+      "step": 2771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452240405912099e-05,
+      "loss": 0.7769,
+      "step": 2772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447272430250377e-05,
+      "loss": 0.8098,
+      "step": 2773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244230381202017e-05,
+      "loss": 0.7504,
+      "step": 2774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437334552526068e-05,
+      "loss": 0.813,
+      "step": 2775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432364653072822e-05,
+      "loss": 0.7581,
+      "step": 2776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427394114965352e-05,
+      "loss": 0.7637,
+      "step": 2777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422422939508752e-05,
+      "loss": 0.7209,
+      "step": 2778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417451128008276e-05,
+      "loss": 0.729,
+      "step": 2779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412478681769348e-05,
+      "loss": 0.7214,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240750560209757e-05,
+      "loss": 0.7369,
+      "step": 2781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402531890298686e-05,
+      "loss": 0.7571,
+      "step": 2782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2397557547678622e-05,
+      "loss": 0.7089,
+      "step": 2783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392582575543475e-05,
+      "loss": 0.7712,
+      "step": 2784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387606975199491e-05,
+      "loss": 0.7665,
+      "step": 2785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382630747953096e-05,
+      "loss": 0.7751,
+      "step": 2786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377653895110866e-05,
+      "loss": 0.7063,
+      "step": 2787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372676417979555e-05,
+      "loss": 0.7981,
+      "step": 2788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367698317866074e-05,
+      "loss": 0.7574,
+      "step": 2789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362719596077495e-05,
+      "loss": 0.7935,
+      "step": 2790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357740253921065e-05,
+      "loss": 0.7511,
+      "step": 2791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352760292704178e-05,
+      "loss": 0.7756,
+      "step": 2792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347779713734403e-05,
+      "loss": 0.7048,
+      "step": 2793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342798518319463e-05,
+      "loss": 0.7791,
+      "step": 2794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337816707767249e-05,
+      "loss": 0.7321,
+      "step": 2795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233283428338581e-05,
+      "loss": 0.7324,
+      "step": 2796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327851246483356e-05,
+      "loss": 0.7118,
+      "step": 2797
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232286759836826e-05,
+      "loss": 0.7228,
+      "step": 2798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231788334034906e-05,
+      "loss": 0.7756,
+      "step": 2799
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312898473734436e-05,
+      "loss": 0.7988,
+      "step": 2800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230791299983325e-05,
+      "loss": 0.7817,
+      "step": 2801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230292691995451e-05,
+      "loss": 0.8264,
+      "step": 2802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297940235407389e-05,
+      "loss": 0.7732,
+      "step": 2803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292952947501214e-05,
+      "loss": 0.7913,
+      "step": 2804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287965057545476e-05,
+      "loss": 0.7729,
+      "step": 2805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228297656684982e-05,
+      "loss": 0.7357,
+      "step": 2806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2277987476724047e-05,
+      "loss": 0.7355,
+      "step": 2807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2272997788478123e-05,
+      "loss": 0.7625,
+      "step": 2808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268007503422163e-05,
+      "loss": 0.8015,
+      "step": 2809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263016622866443e-05,
+      "loss": 0.6937,
+      "step": 2810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258025148121399e-05,
+      "loss": 0.7844,
+      "step": 2811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2253033080497611e-05,
+      "loss": 0.7314,
+      "step": 2812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248040421305822e-05,
+      "loss": 0.3055,
+      "step": 2813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243047171856937e-05,
+      "loss": 0.6753,
+      "step": 2814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238053333462006e-05,
+      "loss": 0.7222,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233058907432234e-05,
+      "loss": 0.8066,
+      "step": 2816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228063895078988e-05,
+      "loss": 0.7057,
+      "step": 2817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222306829771378e-05,
+      "loss": 0.804,
+      "step": 2818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218072116648286e-05,
+      "loss": 0.7678,
+      "step": 2819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221307535319432e-05,
+      "loss": 0.7397,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208078008663865e-05,
+      "loss": 0.7725,
+      "step": 2821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220308008436905e-05,
+      "loss": 0.7825,
+      "step": 2822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198081581622152e-05,
+      "loss": 0.8027,
+      "step": 2823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193082501735605e-05,
+      "loss": 0.6921,
+      "step": 2824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2188082846021993e-05,
+      "loss": 0.2978,
+      "step": 2825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183082615794053e-05,
+      "loss": 0.7185,
+      "step": 2826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178081812364669e-05,
+      "loss": 0.7098,
+      "step": 2827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217308043704688e-05,
+      "loss": 0.7383,
+      "step": 2828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168078491153873e-05,
+      "loss": 0.7576,
+      "step": 2829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216307597599898e-05,
+      "loss": 0.7522,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158072892895693e-05,
+      "loss": 0.7389,
+      "step": 2831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153069243157644e-05,
+      "loss": 0.7784,
+      "step": 2832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2148065028098621e-05,
+      "loss": 0.7456,
+      "step": 2833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143060249032551e-05,
+      "loss": 0.7666,
+      "step": 2834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213805490727352e-05,
+      "loss": 0.7908,
+      "step": 2835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133049004135753e-05,
+      "loss": 0.7407,
+      "step": 2836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128042540933622e-05,
+      "loss": 0.762,
+      "step": 2837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123035518981659e-05,
+      "loss": 0.7632,
+      "step": 2838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118027939594528e-05,
+      "loss": 0.2781,
+      "step": 2839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113019804087045e-05,
+      "loss": 0.7891,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108011113774174e-05,
+      "loss": 0.78,
+      "step": 2841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210300186997102e-05,
+      "loss": 0.761,
+      "step": 2842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097992073992839e-05,
+      "loss": 0.7965,
+      "step": 2843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092981727155024e-05,
+      "loss": 0.7498,
+      "step": 2844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087970830773123e-05,
+      "loss": 0.6998,
+      "step": 2845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082959386162823e-05,
+      "loss": 0.6851,
+      "step": 2846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077947394639947e-05,
+      "loss": 0.7069,
+      "step": 2847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072934857520478e-05,
+      "loss": 0.2676,
+      "step": 2848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067921776120529e-05,
+      "loss": 0.7195,
+      "step": 2849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206290815175636e-05,
+      "loss": 0.7446,
+      "step": 2850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057893985744378e-05,
+      "loss": 0.7565,
+      "step": 2851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052879279401128e-05,
+      "loss": 0.7397,
+      "step": 2852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047864034043293e-05,
+      "loss": 0.7716,
+      "step": 2853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042848250987703e-05,
+      "loss": 0.7773,
+      "step": 2854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037831931551331e-05,
+      "loss": 0.7598,
+      "step": 2855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032815077051286e-05,
+      "loss": 0.7371,
+      "step": 2856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027797688804816e-05,
+      "loss": 0.7048,
+      "step": 2857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202277976812932e-05,
+      "loss": 0.7485,
+      "step": 2858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201776131634232e-05,
+      "loss": 0.7297,
+      "step": 2859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012742334761494e-05,
+      "loss": 0.7722,
+      "step": 2860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007722824704646e-05,
+      "loss": 0.7227,
+      "step": 2861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002702787489728e-05,
+      "loss": 0.74,
+      "step": 2862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997682224434828e-05,
+      "loss": 0.7695,
+      "step": 2863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199266113685817e-05,
+      "loss": 0.7332,
+      "step": 2864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987639526078115e-05,
+      "loss": 0.7292,
+      "step": 2865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198261739341316e-05,
+      "loss": 0.7712,
+      "step": 2866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977594740181949e-05,
+      "loss": 0.7404,
+      "step": 2867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972571567703254e-05,
+      "loss": 0.7388,
+      "step": 2868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196754787729598e-05,
+      "loss": 0.8,
+      "step": 2869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196252367027918e-05,
+      "loss": 0.7517,
+      "step": 2870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957498947972032e-05,
+      "loss": 0.7917,
+      "step": 2871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952473711693852e-05,
+      "loss": 0.7839,
+      "step": 2872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1947447962764096e-05,
+      "loss": 0.7493,
+      "step": 2873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942421702502347e-05,
+      "loss": 0.7228,
+      "step": 2874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1937394932228328e-05,
+      "loss": 0.7467,
+      "step": 2875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1932367653261893e-05,
+      "loss": 0.7164,
+      "step": 2876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927339866923026e-05,
+      "loss": 0.7096,
+      "step": 2877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922311574531858e-05,
+      "loss": 0.8008,
+      "step": 2878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917282777408636e-05,
+      "loss": 0.7141,
+      "step": 2879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191225347687375e-05,
+      "loss": 0.7672,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907223674247714e-05,
+      "loss": 0.7024,
+      "step": 2881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902193370851189e-05,
+      "loss": 0.7991,
+      "step": 2882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189716256800495e-05,
+      "loss": 0.7825,
+      "step": 2883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892131267029909e-05,
+      "loss": 0.7688,
+      "step": 2884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887099469247114e-05,
+      "loss": 0.7056,
+      "step": 2885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882067175977743e-05,
+      "loss": 0.7861,
+      "step": 2886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877034388543094e-05,
+      "loss": 0.7542,
+      "step": 2887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872001108264606e-05,
+      "loss": 0.8003,
+      "step": 2888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866967336463841e-05,
+      "loss": 0.316,
+      "step": 2889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861933074462495e-05,
+      "loss": 0.6803,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856898323582384e-05,
+      "loss": 0.7632,
+      "step": 2891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851863085145462e-05,
+      "loss": 0.7698,
+      "step": 2892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846827360473814e-05,
+      "loss": 0.7629,
+      "step": 2893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841791150889632e-05,
+      "loss": 0.7854,
+      "step": 2894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183675445771526e-05,
+      "loss": 0.759,
+      "step": 2895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831717282273152e-05,
+      "loss": 0.7283,
+      "step": 2896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18266796258859e-05,
+      "loss": 0.7256,
+      "step": 2897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182164148987621e-05,
+      "loss": 0.7112,
+      "step": 2898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181660287556693e-05,
+      "loss": 0.7192,
+      "step": 2899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811563784281017e-05,
+      "loss": 0.7551,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806524217341565e-05,
+      "loss": 0.3125,
+      "step": 2901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801484176071785e-05,
+      "loss": 0.7814,
+      "step": 2902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1796443661795021e-05,
+      "loss": 0.7501,
+      "step": 2903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791402675834733e-05,
+      "loss": 0.7026,
+      "step": 2904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178636121951451e-05,
+      "loss": 0.7756,
+      "step": 2905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781319294158065e-05,
+      "loss": 0.8113,
+      "step": 2906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776276901089225e-05,
+      "loss": 0.7115,
+      "step": 2907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771234041631952e-05,
+      "loss": 0.7672,
+      "step": 2908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766190717110325e-05,
+      "loss": 0.723,
+      "step": 2909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761146928848543e-05,
+      "loss": 0.6643,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175610267817093e-05,
+      "loss": 0.7446,
+      "step": 2911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1751057966401933e-05,
+      "loss": 0.7714,
+      "step": 2912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746012794866118e-05,
+      "loss": 0.7957,
+      "step": 2913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1740967164888165e-05,
+      "loss": 0.301,
+      "step": 2914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735921077792885e-05,
+      "loss": 0.7261,
+      "step": 2915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730874534905203e-05,
+      "loss": 0.7781,
+      "step": 2916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725827537550172e-05,
+      "loss": 0.7747,
+      "step": 2917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720780087052947e-05,
+      "loss": 0.688,
+      "step": 2918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715732184738819e-05,
+      "loss": 0.767,
+      "step": 2919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710683831933192e-05,
+      "loss": 0.7197,
+      "step": 2920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705635029961582e-05,
+      "loss": 0.7745,
+      "step": 2921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700585780149635e-05,
+      "loss": 0.7766,
+      "step": 2922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695536083823109e-05,
+      "loss": 0.8281,
+      "step": 2923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169048594230787e-05,
+      "loss": 0.7871,
+      "step": 2924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685435356929915e-05,
+      "loss": 0.7158,
+      "step": 2925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680384329015351e-05,
+      "loss": 0.7135,
+      "step": 2926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675332859890403e-05,
+      "loss": 0.6505,
+      "step": 2927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670280950881405e-05,
+      "loss": 0.7401,
+      "step": 2928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1665228603314822e-05,
+      "loss": 0.7426,
+      "step": 2929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166017581851722e-05,
+      "loss": 0.7488,
+      "step": 2930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655122597815278e-05,
+      "loss": 0.2875,
+      "step": 2931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650068942535807e-05,
+      "loss": 0.7883,
+      "step": 2932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1645014854005712e-05,
+      "loss": 0.7289,
+      "step": 2933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1639960333552023e-05,
+      "loss": 0.7035,
+      "step": 2934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634905382501884e-05,
+      "loss": 0.3269,
+      "step": 2935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629850002182545e-05,
+      "loss": 0.7451,
+      "step": 2936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624794193921376e-05,
+      "loss": 0.2827,
+      "step": 2937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161973795904585e-05,
+      "loss": 0.7649,
+      "step": 2938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161468129888357e-05,
+      "loss": 0.7061,
+      "step": 2939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609624214762228e-05,
+      "loss": 0.7457,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160456670800964e-05,
+      "loss": 0.7913,
+      "step": 2941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1599508779953736e-05,
+      "loss": 0.738,
+      "step": 2942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594450431922548e-05,
+      "loss": 0.7305,
+      "step": 2943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589391665244223e-05,
+      "loss": 0.7568,
+      "step": 2944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584332481247012e-05,
+      "loss": 0.7539,
+      "step": 2945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157927288125929e-05,
+      "loss": 0.7355,
+      "step": 2946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574212866609525e-05,
+      "loss": 0.7535,
+      "step": 2947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15691524386263e-05,
+      "loss": 0.7371,
+      "step": 2948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156409159863831e-05,
+      "loss": 0.7714,
+      "step": 2949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559030347974356e-05,
+      "loss": 0.7617,
+      "step": 2950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553968687963344e-05,
+      "loss": 0.7683,
+      "step": 2951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154890661993429e-05,
+      "loss": 0.6927,
+      "step": 2952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543844145216316e-05,
+      "loss": 0.762,
+      "step": 2953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538781265138651e-05,
+      "loss": 0.8042,
+      "step": 2954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533717981030635e-05,
+      "loss": 0.6893,
+      "step": 2955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1528654294221706e-05,
+      "loss": 0.7042,
+      "step": 2956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523590206041413e-05,
+      "loss": 0.7466,
+      "step": 2957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518525717819411e-05,
+      "loss": 0.7479,
+      "step": 2958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513460830885455e-05,
+      "loss": 0.7908,
+      "step": 2959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508395546569408e-05,
+      "loss": 0.7579,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150332986620124e-05,
+      "loss": 0.8164,
+      "step": 2961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498263791111019e-05,
+      "loss": 0.6891,
+      "step": 2962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493197322628924e-05,
+      "loss": 0.6602,
+      "step": 2963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488130462085233e-05,
+      "loss": 0.7986,
+      "step": 2964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148306321081032e-05,
+      "loss": 0.7708,
+      "step": 2965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477995570134672e-05,
+      "loss": 0.741,
+      "step": 2966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472927541388881e-05,
+      "loss": 0.691,
+      "step": 2967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467859125903625e-05,
+      "loss": 0.6855,
+      "step": 2968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14627903250097e-05,
+      "loss": 0.6907,
+      "step": 2969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457721140037998e-05,
+      "loss": 0.7011,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452651572319504e-05,
+      "loss": 0.2755,
+      "step": 2971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447581623185316e-05,
+      "loss": 0.7878,
+      "step": 2972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144251129396662e-05,
+      "loss": 0.6609,
+      "step": 2973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437440585994712e-05,
+      "loss": 0.7446,
+      "step": 2974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143236950060098e-05,
+      "loss": 0.7131,
+      "step": 2975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1427298039116923e-05,
+      "loss": 0.679,
+      "step": 2976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142222620287412e-05,
+      "loss": 0.8262,
+      "step": 2977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141715399320426e-05,
+      "loss": 0.7058,
+      "step": 2978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412081411439132e-05,
+      "loss": 0.7261,
+      "step": 2979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140700845891062e-05,
+      "loss": 0.6587,
+      "step": 2980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1401935136950702e-05,
+      "loss": 0.7776,
+      "step": 2981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396861446891457e-05,
+      "loss": 0.7554,
+      "step": 2982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391787390065057e-05,
+      "loss": 0.6996,
+      "step": 2983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386712967803774e-05,
+      "loss": 0.7561,
+      "step": 2984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381638181439973e-05,
+      "loss": 0.3251,
+      "step": 2985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137656303230612e-05,
+      "loss": 0.7458,
+      "step": 2986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371487521734769e-05,
+      "loss": 0.7627,
+      "step": 2987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366411651058575e-05,
+      "loss": 0.7631,
+      "step": 2988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361335421610282e-05,
+      "loss": 0.7546,
+      "step": 2989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356258834722733e-05,
+      "loss": 0.6997,
+      "step": 2990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351181891728862e-05,
+      "loss": 0.7787,
+      "step": 2991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346104593961696e-05,
+      "loss": 0.7607,
+      "step": 2992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341026942754357e-05,
+      "loss": 0.7524,
+      "step": 2993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335948939440065e-05,
+      "loss": 0.7721,
+      "step": 2994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330870585352116e-05,
+      "loss": 0.2892,
+      "step": 2995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325791881823917e-05,
+      "loss": 0.7505,
+      "step": 2996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320712830188958e-05,
+      "loss": 0.7583,
+      "step": 2997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315633431780817e-05,
+      "loss": 0.7244,
+      "step": 2998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131055368793317e-05,
+      "loss": 0.7861,
+      "step": 2999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305473599979781e-05,
+      "loss": 0.7609,
+      "step": 3000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300393169254505e-05,
+      "loss": 0.8032,
+      "step": 3001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295312397091283e-05,
+      "loss": 0.7559,
+      "step": 3002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.129023128482415e-05,
+      "loss": 0.7717,
+      "step": 3003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128514983378723e-05,
+      "loss": 0.7251,
+      "step": 3004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280068045314734e-05,
+      "loss": 0.3136,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274985920740962e-05,
+      "loss": 0.7191,
+      "step": 3006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269903461400307e-05,
+      "loss": 0.7396,
+      "step": 3007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264820668627242e-05,
+      "loss": 0.781,
+      "step": 3008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1259737543756325e-05,
+      "loss": 0.769,
+      "step": 3009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254654088122221e-05,
+      "loss": 0.7749,
+      "step": 3010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249570303059665e-05,
+      "loss": 0.734,
+      "step": 3011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244486189903471e-05,
+      "loss": 0.7422,
+      "step": 3012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239401749988563e-05,
+      "loss": 0.797,
+      "step": 3013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234316984649932e-05,
+      "loss": 0.762,
+      "step": 3014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229231895222658e-05,
+      "loss": 0.7357,
+      "step": 3015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224146483041911e-05,
+      "loss": 0.6853,
+      "step": 3016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219060749442947e-05,
+      "loss": 0.7867,
+      "step": 3017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213974695761098e-05,
+      "loss": 0.7797,
+      "step": 3018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120888832333178e-05,
+      "loss": 0.7233,
+      "step": 3019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203801633490507e-05,
+      "loss": 0.6752,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198714627572862e-05,
+      "loss": 0.7646,
+      "step": 3021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193627306914512e-05,
+      "loss": 0.7581,
+      "step": 3022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188539672851218e-05,
+      "loss": 0.7505,
+      "step": 3023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183451726718808e-05,
+      "loss": 0.7666,
+      "step": 3024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178363469853201e-05,
+      "loss": 0.8093,
+      "step": 3025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11732749035904e-05,
+      "loss": 0.7834,
+      "step": 3026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168186029266483e-05,
+      "loss": 0.7471,
+      "step": 3027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163096848217612e-05,
+      "loss": 0.8005,
+      "step": 3028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158007361780026e-05,
+      "loss": 0.8115,
+      "step": 3029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152917571290048e-05,
+      "loss": 0.6901,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147827478084078e-05,
+      "loss": 0.7886,
+      "step": 3031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142737083498601e-05,
+      "loss": 0.7139,
+      "step": 3032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1137646388870176e-05,
+      "loss": 0.7755,
+      "step": 3033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132555395535441e-05,
+      "loss": 0.7664,
+      "step": 3034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127464104831115e-05,
+      "loss": 0.7604,
+      "step": 3035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122372518093991e-05,
+      "loss": 0.6893,
+      "step": 3036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117280636660944e-05,
+      "loss": 0.7705,
+      "step": 3037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1112188461868925e-05,
+      "loss": 0.7473,
+      "step": 3038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107095995054959e-05,
+      "loss": 0.7142,
+      "step": 3039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102003237556156e-05,
+      "loss": 0.707,
+      "step": 3040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096910190709693e-05,
+      "loss": 0.6682,
+      "step": 3041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091816855852823e-05,
+      "loss": 0.772,
+      "step": 3042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086723234322888e-05,
+      "loss": 0.7528,
+      "step": 3043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081629327457286e-05,
+      "loss": 0.759,
+      "step": 3044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076535136593504e-05,
+      "loss": 0.7634,
+      "step": 3045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071440663069097e-05,
+      "loss": 0.6779,
+      "step": 3046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066345908221704e-05,
+      "loss": 0.3155,
+      "step": 3047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1061250873389018e-05,
+      "loss": 0.707,
+      "step": 3048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056155559908822e-05,
+      "loss": 0.7764,
+      "step": 3049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1051059969118972e-05,
+      "loss": 0.6934,
+      "step": 3050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104596410235739e-05,
+      "loss": 0.7712,
+      "step": 3051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040867960962075e-05,
+      "loss": 0.717,
+      "step": 3052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103577154627109e-05,
+      "loss": 0.6873,
+      "step": 3053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030674859622582e-05,
+      "loss": 0.8295,
+      "step": 3054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1025577902354763e-05,
+      "loss": 0.6968,
+      "step": 3055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020480675805912e-05,
+      "loss": 0.7974,
+      "step": 3056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015383181314385e-05,
+      "loss": 0.7333,
+      "step": 3057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010285420218611e-05,
+      "loss": 0.7708,
+      "step": 3058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100518739385708e-05,
+      "loss": 0.7009,
+      "step": 3059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000089103568356e-05,
+      "loss": 0.7754,
+      "step": 3060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994990550691076e-05,
+      "loss": 0.7014,
+      "step": 3061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0989891736563939e-05,
+      "loss": 0.7638,
+      "step": 3062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984792662525714e-05,
+      "loss": 0.7141,
+      "step": 3063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979693329915247e-05,
+      "loss": 0.2852,
+      "step": 3064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097459374007144e-05,
+      "loss": 0.6576,
+      "step": 3065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969493894333266e-05,
+      "loss": 0.7437,
+      "step": 3066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964393794039771e-05,
+      "loss": 0.7354,
+      "step": 3067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959293440530063e-05,
+      "loss": 0.7902,
+      "step": 3068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0954192835143318e-05,
+      "loss": 0.2849,
+      "step": 3069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949091979218773e-05,
+      "loss": 0.7753,
+      "step": 3070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094399087409574e-05,
+      "loss": 0.7068,
+      "step": 3071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093888952111359e-05,
+      "loss": 0.76,
+      "step": 3072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933787921611753e-05,
+      "loss": 0.6973,
+      "step": 3073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928686076929744e-05,
+      "loss": 0.3105,
+      "step": 3074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923583988407125e-05,
+      "loss": 0.7006,
+      "step": 3075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918481657383523e-05,
+      "loss": 0.7233,
+      "step": 3076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913379085198635e-05,
+      "loss": 0.7515,
+      "step": 3077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090827627319222e-05,
+      "loss": 0.8088,
+      "step": 3078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903173222704093e-05,
+      "loss": 0.7078,
+      "step": 3079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898069935074145e-05,
+      "loss": 0.7305,
+      "step": 3080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892966411642318e-05,
+      "loss": 0.8101,
+      "step": 3081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088786265374862e-05,
+      "loss": 0.7542,
+      "step": 3082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088275866273312e-05,
+      "loss": 0.7671,
+      "step": 3083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877654439935948e-05,
+      "loss": 0.8008,
+      "step": 3084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872549986697293e-05,
+      "loss": 0.709,
+      "step": 3085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867445304357407e-05,
+      "loss": 0.7703,
+      "step": 3086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862340394256607e-05,
+      "loss": 0.8247,
+      "step": 3087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085723525773526e-05,
+      "loss": 0.7404,
+      "step": 3088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852129896133795e-05,
+      "loss": 0.7979,
+      "step": 3089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084702431079271e-05,
+      "loss": 0.7428,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841918503052545e-05,
+      "loss": 0.6906,
+      "step": 3091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836812474253912e-05,
+      "loss": 0.7056,
+      "step": 3092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831706225737473e-05,
+      "loss": 0.7996,
+      "step": 3093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826599758843959e-05,
+      "loss": 0.3147,
+      "step": 3094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082149307491414e-05,
+      "loss": 0.792,
+      "step": 3095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816386175288858e-05,
+      "loss": 0.7281,
+      "step": 3096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811279061309007e-05,
+      "loss": 0.7998,
+      "step": 3097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806171734315538e-05,
+      "loss": 0.7505,
+      "step": 3098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801064195649458e-05,
+      "loss": 0.7393,
+      "step": 3099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795956446651826e-05,
+      "loss": 0.7367,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790848488663756e-05,
+      "loss": 0.7788,
+      "step": 3101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785740323026428e-05,
+      "loss": 0.7659,
+      "step": 3102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078063195108106e-05,
+      "loss": 0.8064,
+      "step": 3103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775523374168939e-05,
+      "loss": 0.7532,
+      "step": 3104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770414593631397e-05,
+      "loss": 0.6967,
+      "step": 3105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765305610809823e-05,
+      "loss": 0.6459,
+      "step": 3106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760196427045655e-05,
+      "loss": 0.7784,
+      "step": 3107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755087043680387e-05,
+      "loss": 0.7336,
+      "step": 3108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0749977462055567e-05,
+      "loss": 0.7394,
+      "step": 3109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744867683512792e-05,
+      "loss": 0.7648,
+      "step": 3110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739757709393714e-05,
+      "loss": 0.7839,
+      "step": 3111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734647541040033e-05,
+      "loss": 0.8073,
+      "step": 3112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07295371797935e-05,
+      "loss": 0.7528,
+      "step": 3113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724426626995917e-05,
+      "loss": 0.6957,
+      "step": 3114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719315883989139e-05,
+      "loss": 0.7446,
+      "step": 3115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714204952115068e-05,
+      "loss": 0.7744,
+      "step": 3116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070909383271566e-05,
+      "loss": 0.6979,
+      "step": 3117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703982527132912e-05,
+      "loss": 0.7637,
+      "step": 3118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698871036708877e-05,
+      "loss": 0.7153,
+      "step": 3119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693759362785653e-05,
+      "loss": 0.8188,
+      "step": 3120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688647506705392e-05,
+      "loss": 0.7213,
+      "step": 3121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683535469810285e-05,
+      "loss": 0.7294,
+      "step": 3122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678423253442575e-05,
+      "loss": 0.7827,
+      "step": 3123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673310858944554e-05,
+      "loss": 0.6377,
+      "step": 3124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668198287658557e-05,
+      "loss": 0.7629,
+      "step": 3125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663085540926967e-05,
+      "loss": 0.6929,
+      "step": 3126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657972620092216e-05,
+      "loss": 0.7795,
+      "step": 3127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652859526496775e-05,
+      "loss": 0.762,
+      "step": 3128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0647746261483168e-05,
+      "loss": 0.7548,
+      "step": 3129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642632826393956e-05,
+      "loss": 0.7169,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637519222571752e-05,
+      "loss": 0.7729,
+      "step": 3131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063240545135921e-05,
+      "loss": 0.7279,
+      "step": 3132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627291514099025e-05,
+      "loss": 0.2977,
+      "step": 3133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622177412133943e-05,
+      "loss": 0.7649,
+      "step": 3134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061706314680675e-05,
+      "loss": 0.7739,
+      "step": 3135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611948719460267e-05,
+      "loss": 0.8071,
+      "step": 3136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606834131437371e-05,
+      "loss": 0.781,
+      "step": 3137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601719384080973e-05,
+      "loss": 0.7603,
+      "step": 3138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0596604478734025e-05,
+      "loss": 0.7698,
+      "step": 3139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591489416739524e-05,
+      "loss": 0.7454,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586374199440512e-05,
+      "loss": 0.7854,
+      "step": 3141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058125882818006e-05,
+      "loss": 0.7751,
+      "step": 3142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0576143304301287e-05,
+      "loss": 0.7892,
+      "step": 3143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571027629147357e-05,
+      "loss": 0.7649,
+      "step": 3144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565911804061463e-05,
+      "loss": 0.7186,
+      "step": 3145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560795830386846e-05,
+      "loss": 0.272,
+      "step": 3146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555679709466777e-05,
+      "loss": 0.7595,
+      "step": 3147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0550563442644577e-05,
+      "loss": 0.746,
+      "step": 3148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545447031263596e-05,
+      "loss": 0.7708,
+      "step": 3149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540330476667224e-05,
+      "loss": 0.2847,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535213780198896e-05,
+      "loss": 0.7673,
+      "step": 3151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530096943202073e-05,
+      "loss": 0.7414,
+      "step": 3152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052497996702026e-05,
+      "loss": 0.7274,
+      "step": 3153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519862852996996e-05,
+      "loss": 0.6991,
+      "step": 3154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514745602475858e-05,
+      "loss": 0.739,
+      "step": 3155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509628216800456e-05,
+      "loss": 0.7406,
+      "step": 3156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504510697314439e-05,
+      "loss": 0.7516,
+      "step": 3157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499393045361488e-05,
+      "loss": 0.7056,
+      "step": 3158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494275262285323e-05,
+      "loss": 0.7708,
+      "step": 3159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0489157349429688e-05,
+      "loss": 0.823,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484039308138379e-05,
+      "loss": 0.741,
+      "step": 3161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478921139755206e-05,
+      "loss": 0.7561,
+      "step": 3162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473802845624028e-05,
+      "loss": 0.738,
+      "step": 3163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468684427088726e-05,
+      "loss": 0.7898,
+      "step": 3164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463565885493227e-05,
+      "loss": 0.7996,
+      "step": 3165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458447222181474e-05,
+      "loss": 0.2933,
+      "step": 3166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453328438497446e-05,
+      "loss": 0.7473,
+      "step": 3167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448209535785167e-05,
+      "loss": 0.7493,
+      "step": 3168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443090515388678e-05,
+      "loss": 0.741,
+      "step": 3169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437971378652054e-05,
+      "loss": 0.681,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432852126919407e-05,
+      "loss": 0.7734,
+      "step": 3171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427732761534867e-05,
+      "loss": 0.7292,
+      "step": 3172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422613283842607e-05,
+      "loss": 0.7384,
+      "step": 3173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417493695186819e-05,
+      "loss": 0.7686,
+      "step": 3174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041237399691173e-05,
+      "loss": 0.7207,
+      "step": 3175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407254190361602e-05,
+      "loss": 0.3046,
+      "step": 3176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402134276880704e-05,
+      "loss": 0.7898,
+      "step": 3177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397014257813357e-05,
+      "loss": 0.6956,
+      "step": 3178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391894134503897e-05,
+      "loss": 0.7203,
+      "step": 3179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386773908296685e-05,
+      "loss": 0.7532,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381653580536123e-05,
+      "loss": 0.7861,
+      "step": 3181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037653315256663e-05,
+      "loss": 0.7407,
+      "step": 3182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371412625732642e-05,
+      "loss": 0.7124,
+      "step": 3183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036629200137864e-05,
+      "loss": 0.7581,
+      "step": 3184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361171280849121e-05,
+      "loss": 0.7637,
+      "step": 3185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0356050465488603e-05,
+      "loss": 0.7292,
+      "step": 3186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350929556641635e-05,
+      "loss": 0.7678,
+      "step": 3187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345808555652794e-05,
+      "loss": 0.7421,
+      "step": 3188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340687463866674e-05,
+      "loss": 0.7806,
+      "step": 3189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033556628262789e-05,
+      "loss": 0.7379,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330445013281094e-05,
+      "loss": 0.7355,
+      "step": 3191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0325323657170946e-05,
+      "loss": 0.77,
+      "step": 3192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320202215642144e-05,
+      "loss": 0.7925,
+      "step": 3193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315080690039391e-05,
+      "loss": 0.7202,
+      "step": 3194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309959081707428e-05,
+      "loss": 0.6619,
+      "step": 3195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304837391991005e-05,
+      "loss": 0.2898,
+      "step": 3196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299715622234903e-05,
+      "loss": 0.7324,
+      "step": 3197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029459377378392e-05,
+      "loss": 0.7218,
+      "step": 3198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289471847982877e-05,
+      "loss": 0.8081,
+      "step": 3199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284349846176609e-05,
+      "loss": 0.7561,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279227769709978e-05,
+      "loss": 0.7234,
+      "step": 3201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274105619927863e-05,
+      "loss": 0.7692,
+      "step": 3202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026898339817516e-05,
+      "loss": 0.7561,
+      "step": 3203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263861105796787e-05,
+      "loss": 0.6963,
+      "step": 3204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025873874413768e-05,
+      "loss": 0.7462,
+      "step": 3205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253616314542797e-05,
+      "loss": 0.7815,
+      "step": 3206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248493818357097e-05,
+      "loss": 0.2956,
+      "step": 3207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024337125692558e-05,
+      "loss": 0.7598,
+      "step": 3208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238248631593252e-05,
+      "loss": 0.7639,
+      "step": 3209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233125943705131e-05,
+      "loss": 0.7517,
+      "step": 3210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228003194606258e-05,
+      "loss": 0.7433,
+      "step": 3211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0222880385641692e-05,
+      "loss": 0.7279,
+      "step": 3212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02177575181565e-05,
+      "loss": 0.7865,
+      "step": 3213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0212634593495767e-05,
+      "loss": 0.7319,
+      "step": 3214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02075116130046e-05,
+      "loss": 0.6912,
+      "step": 3215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020238857802811e-05,
+      "loss": 0.7156,
+      "step": 3216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197265489911431e-05,
+      "loss": 0.729,
+      "step": 3217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019214234999971e-05,
+      "loss": 0.7909,
+      "step": 3218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187019159638098e-05,
+      "loss": 0.7413,
+      "step": 3219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181895920171772e-05,
+      "loss": 0.7666,
+      "step": 3220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176772632945911e-05,
+      "loss": 0.287,
+      "step": 3221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171649299305714e-05,
+      "loss": 0.7416,
+      "step": 3222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166525920596395e-05,
+      "loss": 0.7539,
+      "step": 3223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016140249816317e-05,
+      "loss": 0.7651,
+      "step": 3224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015627903335127e-05,
+      "loss": 0.7347,
+      "step": 3225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015115552750594e-05,
+      "loss": 0.793,
+      "step": 3226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146031981972435e-05,
+      "loss": 0.7371,
+      "step": 3227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140908398096019e-05,
+      "loss": 0.7603,
+      "step": 3228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135784777221969e-05,
+      "loss": 0.7788,
+      "step": 3229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130661120695564e-05,
+      "loss": 0.7208,
+      "step": 3230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0125537429862098e-05,
+      "loss": 0.7645,
+      "step": 3231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120413706066878e-05,
+      "loss": 0.7633,
+      "step": 3232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115289950655215e-05,
+      "loss": 0.6965,
+      "step": 3233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110166164972425e-05,
+      "loss": 0.7292,
+      "step": 3234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105042350363842e-05,
+      "loss": 0.7852,
+      "step": 3235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099918508174798e-05,
+      "loss": 0.7072,
+      "step": 3236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009479463975063e-05,
+      "loss": 0.6643,
+      "step": 3237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089670746436693e-05,
+      "loss": 0.7491,
+      "step": 3238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084546829578342e-05,
+      "loss": 0.7285,
+      "step": 3239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007942289052094e-05,
+      "loss": 0.7185,
+      "step": 3240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074298930609851e-05,
+      "loss": 0.7996,
+      "step": 3241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069174951190455e-05,
+      "loss": 0.7141,
+      "step": 3242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064050953608124e-05,
+      "loss": 0.8,
+      "step": 3243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058926939208244e-05,
+      "loss": 0.7094,
+      "step": 3244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053802909336204e-05,
+      "loss": 0.7441,
+      "step": 3245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048678865337393e-05,
+      "loss": 0.7617,
+      "step": 3246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004355480855721e-05,
+      "loss": 0.7637,
+      "step": 3247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038430740341049e-05,
+      "loss": 0.7386,
+      "step": 3248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033306662034317e-05,
+      "loss": 0.759,
+      "step": 3249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028182574982415e-05,
+      "loss": 0.6959,
+      "step": 3250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023058480530747e-05,
+      "loss": 0.7842,
+      "step": 3251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001793438002473e-05,
+      "loss": 0.724,
+      "step": 3252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001281027480977e-05,
+      "loss": 0.7361,
+      "step": 3253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0007686166231274e-05,
+      "loss": 0.7563,
+      "step": 3254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002562055634661e-05,
+      "loss": 0.7506,
+      "step": 3255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997437944365342e-06,
+      "loss": 0.7778,
+      "step": 3256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992313833768728e-06,
+      "loss": 0.7683,
+      "step": 3257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987189725190232e-06,
+      "loss": 0.7373,
+      "step": 3258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98206561997527e-06,
+      "loss": 0.7094,
+      "step": 3259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976941519469252e-06,
+      "loss": 0.7793,
+      "step": 3260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971817425017589e-06,
+      "loss": 0.7427,
+      "step": 3261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966693337965687e-06,
+      "loss": 0.7581,
+      "step": 3262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961569259658954e-06,
+      "loss": 0.7113,
+      "step": 3263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956445191442793e-06,
+      "loss": 0.31,
+      "step": 3264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951321134662609e-06,
+      "loss": 0.7814,
+      "step": 3265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946197090663798e-06,
+      "loss": 0.7366,
+      "step": 3266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941073060791758e-06,
+      "loss": 0.75,
+      "step": 3267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.935949046391878e-06,
+      "loss": 0.7578,
+      "step": 3268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930825048809548e-06,
+      "loss": 0.7581,
+      "step": 3269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92570106939015e-06,
+      "loss": 0.7761,
+      "step": 3270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920577109479064e-06,
+      "loss": 0.7299,
+      "step": 3271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91545317042166e-06,
+      "loss": 0.739,
+      "step": 3272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910329253563309e-06,
+      "loss": 0.715,
+      "step": 3273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905205360249375e-06,
+      "loss": 0.6936,
+      "step": 3274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900081491825208e-06,
+      "loss": 0.7543,
+      "step": 3275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894957649636163e-06,
+      "loss": 0.6802,
+      "step": 3276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889833835027577e-06,
+      "loss": 0.681,
+      "step": 3277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884710049344786e-06,
+      "loss": 0.7273,
+      "step": 3278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879586293933124e-06,
+      "loss": 0.7781,
+      "step": 3279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.874462570137905e-06,
+      "loss": 0.6694,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86933887930444e-06,
+      "loss": 0.7166,
+      "step": 3281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864215222778036e-06,
+      "loss": 0.6876,
+      "step": 3282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859091601903983e-06,
+      "loss": 0.7756,
+      "step": 3283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853968018027567e-06,
+      "loss": 0.7264,
+      "step": 3284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848844472494065e-06,
+      "loss": 0.8252,
+      "step": 3285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843720966648736e-06,
+      "loss": 0.7239,
+      "step": 3286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838597501836834e-06,
+      "loss": 0.8,
+      "step": 3287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833474079403608e-06,
+      "loss": 0.6686,
+      "step": 3288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828350700694288e-06,
+      "loss": 0.7063,
+      "step": 3289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823227367054094e-06,
+      "loss": 0.7668,
+      "step": 3290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818104079828232e-06,
+      "loss": 0.7427,
+      "step": 3291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812980840361905e-06,
+      "loss": 0.7617,
+      "step": 3292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807857650000294e-06,
+      "loss": 0.7681,
+      "step": 3293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80273451008857e-06,
+      "loss": 0.7072,
+      "step": 3294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79761142197189e-06,
+      "loss": 0.8015,
+      "step": 3295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792488386995402e-06,
+      "loss": 0.8049,
+      "step": 3296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787365406504237e-06,
+      "loss": 0.8098,
+      "step": 3297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782242481843505e-06,
+      "loss": 0.749,
+      "step": 3298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777119614358313e-06,
+      "loss": 0.7242,
+      "step": 3299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771996805393746e-06,
+      "loss": 0.7373,
+      "step": 3300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766874056294872e-06,
+      "loss": 0.7808,
+      "step": 3301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761751368406751e-06,
+      "loss": 0.6996,
+      "step": 3302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75662874307442e-06,
+      "loss": 0.7252,
+      "step": 3303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751506181642906e-06,
+      "loss": 0.7217,
+      "step": 3304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746383685457207e-06,
+      "loss": 0.7168,
+      "step": 3305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74126125586232e-06,
+      "loss": 0.3347,
+      "step": 3306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736138894203215e-06,
+      "loss": 0.7566,
+      "step": 3307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731016601824842e-06,
+      "loss": 0.7097,
+      "step": 3308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725894380072142e-06,
+      "loss": 0.7837,
+      "step": 3309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720772230290025e-06,
+      "loss": 0.7579,
+      "step": 3310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715650153823393e-06,
+      "loss": 0.8228,
+      "step": 3311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710528152017126e-06,
+      "loss": 0.7703,
+      "step": 3312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705406226216081e-06,
+      "loss": 0.7725,
+      "step": 3313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7002843777651e-06,
+      "loss": 0.7327,
+      "step": 3314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695162608008997e-06,
+      "loss": 0.7692,
+      "step": 3315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690040918292575e-06,
+      "loss": 0.7283,
+      "step": 3316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68491930996061e-06,
+      "loss": 0.7379,
+      "step": 3317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.67979778435786e-06,
+      "loss": 0.782,
+      "step": 3318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.674676342829052e-06,
+      "loss": 0.7959,
+      "step": 3319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669554986718908e-06,
+      "loss": 0.7197,
+      "step": 3320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664433717372112e-06,
+      "loss": 0.7742,
+      "step": 3321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659312536133331e-06,
+      "loss": 0.7693,
+      "step": 3322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65419144434721e-06,
+      "loss": 0.7275,
+      "step": 3323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649070443358368e-06,
+      "loss": 0.7476,
+      "step": 3324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6439495345114e-06,
+      "loss": 0.7146,
+      "step": 3325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638828719150884e-06,
+      "loss": 0.6992,
+      "step": 3326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633707998621362e-06,
+      "loss": 0.6846,
+      "step": 3327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628587374267361e-06,
+      "loss": 0.7222,
+      "step": 3328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623466847433374e-06,
+      "loss": 0.7949,
+      "step": 3329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618346419463877e-06,
+      "loss": 0.7008,
+      "step": 3330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613226091703315e-06,
+      "loss": 0.7462,
+      "step": 3331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608105865496105e-06,
+      "loss": 0.7878,
+      "step": 3332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602985742186648e-06,
+      "loss": 0.2985,
+      "step": 3333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597865723119302e-06,
+      "loss": 0.7238,
+      "step": 3334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592745809638405e-06,
+      "loss": 0.7654,
+      "step": 3335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587626003088273e-06,
+      "loss": 0.2864,
+      "step": 3336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582506304813185e-06,
+      "loss": 0.8154,
+      "step": 3337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577386716157396e-06,
+      "loss": 0.2927,
+      "step": 3338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572267238465136e-06,
+      "loss": 0.2734,
+      "step": 3339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567147873080598e-06,
+      "loss": 0.6818,
+      "step": 3340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562028621347948e-06,
+      "loss": 0.7941,
+      "step": 3341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556909484611323e-06,
+      "loss": 0.6832,
+      "step": 3342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551790464214835e-06,
+      "loss": 0.7194,
+      "step": 3343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546671561502556e-06,
+      "loss": 0.7622,
+      "step": 3344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541552777818533e-06,
+      "loss": 0.7747,
+      "step": 3345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536434114506778e-06,
+      "loss": 0.7402,
+      "step": 3346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531315572911275e-06,
+      "loss": 0.728,
+      "step": 3347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526197154375973e-06,
+      "loss": 0.7412,
+      "step": 3348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521078860244795e-06,
+      "loss": 0.8005,
+      "step": 3349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515960691861625e-06,
+      "loss": 0.7266,
+      "step": 3350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.510842650570315e-06,
+      "loss": 0.3092,
+      "step": 3351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50572473771468e-06,
+      "loss": 0.7273,
+      "step": 3352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500606954638514e-06,
+      "loss": 0.6776,
+      "step": 3353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495489302685563e-06,
+      "loss": 0.7852,
+      "step": 3354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490371783199545e-06,
+      "loss": 0.7823,
+      "step": 3355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485254397524146e-06,
+      "loss": 0.7347,
+      "step": 3356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480137147003009e-06,
+      "loss": 0.7395,
+      "step": 3357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475020032979742e-06,
+      "loss": 0.7106,
+      "step": 3358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46990305679793e-06,
+      "loss": 0.7402,
+      "step": 3359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.464786219801107e-06,
+      "loss": 0.7432,
+      "step": 3360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459669523332778e-06,
+      "loss": 0.743,
+      "step": 3361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454552968736408e-06,
+      "loss": 0.7968,
+      "step": 3362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449436557355426e-06,
+      "loss": 0.7438,
+      "step": 3363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444320290533224e-06,
+      "loss": 0.7297,
+      "step": 3364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439204169613157e-06,
+      "loss": 0.6978,
+      "step": 3365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434088195938537e-06,
+      "loss": 0.8213,
+      "step": 3366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428972370852643e-06,
+      "loss": 0.7611,
+      "step": 3367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423856695698715e-06,
+      "loss": 0.2921,
+      "step": 3368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418741171819944e-06,
+      "loss": 0.7269,
+      "step": 3369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413625800559493e-06,
+      "loss": 0.7522,
+      "step": 3370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408510583260478e-06,
+      "loss": 0.7241,
+      "step": 3371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403395521265977e-06,
+      "loss": 0.8193,
+      "step": 3372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.39828061591903e-06,
+      "loss": 0.7646,
+      "step": 3373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393165868562632e-06,
+      "loss": 0.7461,
+      "step": 3374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388051280539735e-06,
+      "loss": 0.7412,
+      "step": 3375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382936853193252e-06,
+      "loss": 0.6915,
+      "step": 3376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377822587866058e-06,
+      "loss": 0.7487,
+      "step": 3377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372708485900976e-06,
+      "loss": 0.6852,
+      "step": 3378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367594548640792e-06,
+      "loss": 0.2557,
+      "step": 3379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362480777428253e-06,
+      "loss": 0.75,
+      "step": 3380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35736717360605e-06,
+      "loss": 0.781,
+      "step": 3381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352253738516837e-06,
+      "loss": 0.7526,
+      "step": 3382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347140473503229e-06,
+      "loss": 0.757,
+      "step": 3383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342027379907787e-06,
+      "loss": 0.7747,
+      "step": 3384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336914459073035e-06,
+      "loss": 0.2954,
+      "step": 3385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331801712341445e-06,
+      "loss": 0.6766,
+      "step": 3386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326689141055448e-06,
+      "loss": 0.7793,
+      "step": 3387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321576746557426e-06,
+      "loss": 0.7407,
+      "step": 3388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316464530189715e-06,
+      "loss": 0.6549,
+      "step": 3389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311352493294608e-06,
+      "loss": 0.738,
+      "step": 3390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306240637214347e-06,
+      "loss": 0.7441,
+      "step": 3391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301128963291128e-06,
+      "loss": 0.814,
+      "step": 3392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296017472867092e-06,
+      "loss": 0.7379,
+      "step": 3393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290906167284346e-06,
+      "loss": 0.803,
+      "step": 3394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285795047884934e-06,
+      "loss": 0.7362,
+      "step": 3395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280684116010864e-06,
+      "loss": 0.7609,
+      "step": 3396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275573373004087e-06,
+      "loss": 0.7134,
+      "step": 3397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270462820206506e-06,
+      "loss": 0.7108,
+      "step": 3398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265352458959969e-06,
+      "loss": 0.7035,
+      "step": 3399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260242290606287e-06,
+      "loss": 0.7396,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255132316487208e-06,
+      "loss": 0.7561,
+      "step": 3401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.250022537944434e-06,
+      "loss": 0.7279,
+      "step": 3402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244912956319618e-06,
+      "loss": 0.6929,
+      "step": 3403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239803572954352e-06,
+      "loss": 0.8083,
+      "step": 3404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23469438919018e-06,
+      "loss": 0.6907,
+      "step": 3405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229585406368604e-06,
+      "loss": 0.7605,
+      "step": 3406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224476625831064e-06,
+      "loss": 0.7307,
+      "step": 3407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219368048918942e-06,
+      "loss": 0.7798,
+      "step": 3408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214259676973575e-06,
+      "loss": 0.7615,
+      "step": 3409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209151511336246e-06,
+      "loss": 0.7146,
+      "step": 3410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204043553348179e-06,
+      "loss": 0.7341,
+      "step": 3411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198935804350545e-06,
+      "loss": 0.7717,
+      "step": 3412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19382826568446e-06,
+      "loss": 0.7327,
+      "step": 3413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188720938690993e-06,
+      "loss": 0.6981,
+      "step": 3414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183613824711146e-06,
+      "loss": 0.7673,
+      "step": 3415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178506925085864e-06,
+      "loss": 0.8232,
+      "step": 3416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173400241156046e-06,
+      "loss": 0.7834,
+      "step": 3417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168293774262528e-06,
+      "loss": 0.738,
+      "step": 3418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16318752574609e-06,
+      "loss": 0.7693,
+      "step": 3419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158081496947457e-06,
+      "loss": 0.7518,
+      "step": 3420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152975689207294e-06,
+      "loss": 0.6843,
+      "step": 3421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147870103866206e-06,
+      "loss": 0.7437,
+      "step": 3422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142764742264741e-06,
+      "loss": 0.8367,
+      "step": 3423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137659605743395e-06,
+      "loss": 0.7133,
+      "step": 3424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132554695642593e-06,
+      "loss": 0.8184,
+      "step": 3425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12745001330271e-06,
+      "loss": 0.7002,
+      "step": 3426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122345560064057e-06,
+      "loss": 0.7627,
+      "step": 3427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117241337266885e-06,
+      "loss": 0.697,
+      "step": 3428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112137346251382e-06,
+      "loss": 0.7157,
+      "step": 3429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107033588357684e-06,
+      "loss": 0.7238,
+      "step": 3430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101930064925857e-06,
+      "loss": 0.7556,
+      "step": 3431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09682677729591e-06,
+      "loss": 0.7096,
+      "step": 3432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091723726807784e-06,
+      "loss": 0.7114,
+      "step": 3433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086620914801367e-06,
+      "loss": 0.8181,
+      "step": 3434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081518342616479e-06,
+      "loss": 0.7345,
+      "step": 3435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076416011592876e-06,
+      "loss": 0.7068,
+      "step": 3436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071313923070256e-06,
+      "loss": 0.6646,
+      "step": 3437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066212078388245e-06,
+      "loss": 0.7015,
+      "step": 3438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061110478886417e-06,
+      "loss": 0.7766,
+      "step": 3439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056009125904265e-06,
+      "loss": 0.7721,
+      "step": 3440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05090802078123e-06,
+      "loss": 0.7861,
+      "step": 3441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.045807164856684e-06,
+      "loss": 0.7246,
+      "step": 3442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040706559469938e-06,
+      "loss": 0.73,
+      "step": 3443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035606205960232e-06,
+      "loss": 0.7317,
+      "step": 3444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.030506105666736e-06,
+      "loss": 0.7625,
+      "step": 3445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025406259928563e-06,
+      "loss": 0.7463,
+      "step": 3446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020306670084755e-06,
+      "loss": 0.7366,
+      "step": 3447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015207337474286e-06,
+      "loss": 0.7605,
+      "step": 3448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010108263436063e-06,
+      "loss": 0.7917,
+      "step": 3449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005009449308924e-06,
+      "loss": 0.7416,
+      "step": 3450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999910896431647e-06,
+      "loss": 0.8257,
+      "step": 3451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994812606142922e-06,
+      "loss": 0.7437,
+      "step": 3452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98971457978139e-06,
+      "loss": 0.7769,
+      "step": 3453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984616818685617e-06,
+      "loss": 0.8015,
+      "step": 3454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979519324194093e-06,
+      "loss": 0.7745,
+      "step": 3455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974422097645242e-06,
+      "loss": 0.7239,
+      "step": 3456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969325140377421e-06,
+      "loss": 0.7708,
+      "step": 3457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964228453728912e-06,
+      "loss": 0.7296,
+      "step": 3458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95913203903793e-06,
+      "loss": 0.6759,
+      "step": 3459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95403589764261e-06,
+      "loss": 0.7659,
+      "step": 3460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948940030881028e-06,
+      "loss": 0.7468,
+      "step": 3461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94384444009118e-06,
+      "loss": 0.7014,
+      "step": 3462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.938749126610987e-06,
+      "loss": 0.748,
+      "step": 3463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933654091778303e-06,
+      "loss": 0.7631,
+      "step": 3464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928559336930905e-06,
+      "loss": 0.7577,
+      "step": 3465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923464863406497e-06,
+      "loss": 0.7522,
+      "step": 3466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918370672542717e-06,
+      "loss": 0.7551,
+      "step": 3467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913276765677116e-06,
+      "loss": 0.6951,
+      "step": 3468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908183144147178e-06,
+      "loss": 0.7493,
+      "step": 3469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903089809290309e-06,
+      "loss": 0.741,
+      "step": 3470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897996762443845e-06,
+      "loss": 0.7925,
+      "step": 3471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892904004945041e-06,
+      "loss": 0.7764,
+      "step": 3472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887811538131078e-06,
+      "loss": 0.7015,
+      "step": 3473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882719363339061e-06,
+      "loss": 0.7095,
+      "step": 3474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877627481906016e-06,
+      "loss": 0.7904,
+      "step": 3475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872535895168889e-06,
+      "loss": 0.684,
+      "step": 3476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867444604464562e-06,
+      "loss": 0.7136,
+      "step": 3477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862353611129828e-06,
+      "loss": 0.2899,
+      "step": 3478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857262916501402e-06,
+      "loss": 0.7158,
+      "step": 3479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852172521915923e-06,
+      "loss": 0.6897,
+      "step": 3480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847082428709956e-06,
+      "loss": 0.7303,
+      "step": 3481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841992638219977e-06,
+      "loss": 0.7078,
+      "step": 3482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836903151782392e-06,
+      "loss": 0.7173,
+      "step": 3483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831813970733516e-06,
+      "loss": 0.72,
+      "step": 3484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8267250964096e-06,
+      "loss": 0.7296,
+      "step": 3485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8216365301468e-06,
+      "loss": 0.7706,
+      "step": 3486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816548273281197e-06,
+      "loss": 0.7611,
+      "step": 3487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811460327148787e-06,
+      "loss": 0.6925,
+      "step": 3488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806372693085491e-06,
+      "loss": 0.7256,
+      "step": 3489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801285372427142e-06,
+      "loss": 0.8176,
+      "step": 3490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796198366509495e-06,
+      "loss": 0.7954,
+      "step": 3491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791111676668221e-06,
+      "loss": 0.7458,
+      "step": 3492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786025304238905e-06,
+      "loss": 0.7709,
+      "step": 3493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780939250557055e-06,
+      "loss": 0.7976,
+      "step": 3494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775853516958089e-06,
+      "loss": 0.7476,
+      "step": 3495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770768104777344e-06,
+      "loss": 0.6931,
+      "step": 3496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76568301535007e-06,
+      "loss": 0.73,
+      "step": 3497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760598250011442e-06,
+      "loss": 0.6621,
+      "step": 3498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75551381009653e-06,
+      "loss": 0.7522,
+      "step": 3499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75042969694034e-06,
+      "loss": 0.7628,
+      "step": 3500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74534591187778e-06,
+      "loss": 0.7666,
+      "step": 3501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.740262456243676e-06,
+      "loss": 0.7347,
+      "step": 3502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735179331372763e-06,
+      "loss": 0.7688,
+      "step": 3503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730096538599696e-06,
+      "loss": 0.7544,
+      "step": 3504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72501407925904e-06,
+      "loss": 0.7476,
+      "step": 3505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71993195468527e-06,
+      "loss": 0.7853,
+      "step": 3506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714850166212772e-06,
+      "loss": 0.7644,
+      "step": 3507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709768715175852e-06,
+      "loss": 0.6847,
+      "step": 3508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704687602908722e-06,
+      "loss": 0.7214,
+      "step": 3509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6996068307455e-06,
+      "loss": 0.7227,
+      "step": 3510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694526400020222e-06,
+      "loss": 0.7322,
+      "step": 3511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689446312066834e-06,
+      "loss": 0.6667,
+      "step": 3512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684366568219186e-06,
+      "loss": 0.6852,
+      "step": 3513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679287169811045e-06,
+      "loss": 0.6941,
+      "step": 3514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674208118176086e-06,
+      "loss": 0.7192,
+      "step": 3515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669129414647887e-06,
+      "loss": 0.6787,
+      "step": 3516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664051060559938e-06,
+      "loss": 0.7032,
+      "step": 3517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658973057245643e-06,
+      "loss": 0.6995,
+      "step": 3518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653895406038306e-06,
+      "loss": 0.7701,
+      "step": 3519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648818108271141e-06,
+      "loss": 0.7454,
+      "step": 3520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643741165277272e-06,
+      "loss": 0.6799,
+      "step": 3521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638664578389722e-06,
+      "loss": 0.7476,
+      "step": 3522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633588348941427e-06,
+      "loss": 0.7271,
+      "step": 3523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628512478265232e-06,
+      "loss": 0.7478,
+      "step": 3524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623436967693883e-06,
+      "loss": 0.7123,
+      "step": 3525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61836181856003e-06,
+      "loss": 0.7397,
+      "step": 3526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613287032196228e-06,
+      "loss": 0.7428,
+      "step": 3527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608212609934947e-06,
+      "loss": 0.8015,
+      "step": 3528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603138553108547e-06,
+      "loss": 0.6823,
+      "step": 3529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5980648630493e-06,
+      "loss": 0.7797,
+      "step": 3530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59299154108938e-06,
+      "loss": 0.7744,
+      "step": 3531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587918588560868e-06,
+      "loss": 0.7771,
+      "step": 3532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582846006795741e-06,
+      "loss": 0.7473,
+      "step": 3533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577773797125885e-06,
+      "loss": 0.7295,
+      "step": 3534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572701960883082e-06,
+      "loss": 0.6787,
+      "step": 3535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56763049939902e-06,
+      "loss": 0.7311,
+      "step": 3536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56255941400529e-06,
+      "loss": 0.7125,
+      "step": 3537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557488706033383e-06,
+      "loss": 0.8149,
+      "step": 3538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55241837681469e-06,
+      "loss": 0.7239,
+      "step": 3539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5473484276805e-06,
+      "loss": 0.7067,
+      "step": 3540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542278859962003e-06,
+      "loss": 0.6807,
+      "step": 3541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5372096749903e-06,
+      "loss": 0.7913,
+      "step": 3542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532140874096377e-06,
+      "loss": 0.7197,
+      "step": 3543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52707245861112e-06,
+      "loss": 0.7295,
+      "step": 3544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522004429865331e-06,
+      "loss": 0.7646,
+      "step": 3545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516936789189687e-06,
+      "loss": 0.7329,
+      "step": 3546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511869537914772e-06,
+      "loss": 0.7554,
+      "step": 3547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506802677371079e-06,
+      "loss": 0.6426,
+      "step": 3548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501736208888983e-06,
+      "loss": 0.6868,
+      "step": 3549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496670133798762e-06,
+      "loss": 0.7959,
+      "step": 3550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491604453430594e-06,
+      "loss": 0.7595,
+      "step": 3551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486539169114549e-06,
+      "loss": 0.7083,
+      "step": 3552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481474282180592e-06,
+      "loss": 0.748,
+      "step": 3553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476409793958586e-06,
+      "loss": 0.7439,
+      "step": 3554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471345705778295e-06,
+      "loss": 0.6996,
+      "step": 3555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466282018969367e-06,
+      "loss": 0.7209,
+      "step": 3556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461218734861352e-06,
+      "loss": 0.7184,
+      "step": 3557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456155854783689e-06,
+      "loss": 0.67,
+      "step": 3558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451093380065717e-06,
+      "loss": 0.7625,
+      "step": 3559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44603131203666e-06,
+      "loss": 0.7423,
+      "step": 3560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440969652025647e-06,
+      "loss": 0.7661,
+      "step": 3561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435908401361692e-06,
+      "loss": 0.3195,
+      "step": 3562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430847561373702e-06,
+      "loss": 0.8086,
+      "step": 3563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425787133390477e-06,
+      "loss": 0.2869,
+      "step": 3564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.420727118740712e-06,
+      "loss": 0.7665,
+      "step": 3565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415667518752988e-06,
+      "loss": 0.7522,
+      "step": 3566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410608334755782e-06,
+      "loss": 0.6934,
+      "step": 3567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405549568077457e-06,
+      "loss": 0.2921,
+      "step": 3568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.400491220046267e-06,
+      "loss": 0.7584,
+      "step": 3569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395433291990361e-06,
+      "loss": 0.728,
+      "step": 3570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390375785237774e-06,
+      "loss": 0.2899,
+      "step": 3571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385318701116433e-06,
+      "loss": 0.791,
+      "step": 3572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38026204095415e-06,
+      "loss": 0.7661,
+      "step": 3573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375205806078626e-06,
+      "loss": 0.7388,
+      "step": 3574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370149997817457e-06,
+      "loss": 0.762,
+      "step": 3575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365094617498119e-06,
+      "loss": 0.682,
+      "step": 3576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.360039666447979e-06,
+      "loss": 0.7261,
+      "step": 3577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35498514599429e-06,
+      "loss": 0.7676,
+      "step": 3578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349931057464195e-06,
+      "loss": 0.75,
+      "step": 3579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344877402184724e-06,
+      "loss": 0.7036,
+      "step": 3580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339824181482786e-06,
+      "loss": 0.7017,
+      "step": 3581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334771396685181e-06,
+      "loss": 0.7684,
+      "step": 3582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329719049118598e-06,
+      "loss": 0.7734,
+      "step": 3583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324667140109602e-06,
+      "loss": 0.7693,
+      "step": 3584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319615670984652e-06,
+      "loss": 0.7612,
+      "step": 3585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314564643070088e-06,
+      "loss": 0.2845,
+      "step": 3586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309514057692135e-06,
+      "loss": 0.2828,
+      "step": 3587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304463916176894e-06,
+      "loss": 0.7637,
+      "step": 3588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299414219850365e-06,
+      "loss": 0.75,
+      "step": 3589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294364970038418e-06,
+      "loss": 0.7649,
+      "step": 3590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28931616806681e-06,
+      "loss": 0.7634,
+      "step": 3591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284267815261186e-06,
+      "loss": 0.658,
+      "step": 3592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279219912947058e-06,
+      "loss": 0.7327,
+      "step": 3593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274172462449833e-06,
+      "loss": 0.735,
+      "step": 3594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269125465094799e-06,
+      "loss": 0.7667,
+      "step": 3595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264078922207119e-06,
+      "loss": 0.7022,
+      "step": 3596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259032835111837e-06,
+      "loss": 0.7551,
+      "step": 3597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253987205133887e-06,
+      "loss": 0.7032,
+      "step": 3598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.248942033598068e-06,
+      "loss": 0.7073,
+      "step": 3599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243897321829071e-06,
+      "loss": 0.7495,
+      "step": 3600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238853071151459e-06,
+      "loss": 0.7809,
+      "step": 3601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.233809282889677e-06,
+      "loss": 0.7103,
+      "step": 3602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22876595836805e-06,
+      "loss": 0.8018,
+      "step": 3603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22372309891078e-06,
+      "loss": 0.7771,
+      "step": 3604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218680705841942e-06,
+      "loss": 0.7034,
+      "step": 3605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213638780485493e-06,
+      "loss": 0.7659,
+      "step": 3606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208597324165269e-06,
+      "loss": 0.7086,
+      "step": 3607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20355633820498e-06,
+      "loss": 0.7009,
+      "step": 3608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198515823928217e-06,
+      "loss": 0.6975,
+      "step": 3609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19347578265844e-06,
+      "loss": 0.6951,
+      "step": 3610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188436215718984e-06,
+      "loss": 0.6982,
+      "step": 3611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183397124433072e-06,
+      "loss": 0.6531,
+      "step": 3612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178358510123791e-06,
+      "loss": 0.6975,
+      "step": 3613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173320374114104e-06,
+      "loss": 0.733,
+      "step": 3614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168282717726848e-06,
+      "loss": 0.7656,
+      "step": 3615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163245542284745e-06,
+      "loss": 0.7339,
+      "step": 3616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158208849110371e-06,
+      "loss": 0.7495,
+      "step": 3617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153172639526191e-06,
+      "loss": 0.7041,
+      "step": 3618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14813691485454e-06,
+      "loss": 0.7581,
+      "step": 3619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14310167641762e-06,
+      "loss": 0.7266,
+      "step": 3620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138066925537509e-06,
+      "loss": 0.6987,
+      "step": 3621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133032663536162e-06,
+      "loss": 0.6973,
+      "step": 3622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127998891735398e-06,
+      "loss": 0.7949,
+      "step": 3623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122965611456909e-06,
+      "loss": 0.7096,
+      "step": 3624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117932824022259e-06,
+      "loss": 0.7706,
+      "step": 3625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112900530752886e-06,
+      "loss": 0.2769,
+      "step": 3626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107868732970095e-06,
+      "loss": 0.731,
+      "step": 3627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102837431995056e-06,
+      "loss": 0.7886,
+      "step": 3628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097806629148816e-06,
+      "loss": 0.7427,
+      "step": 3629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092776325752287e-06,
+      "loss": 0.7405,
+      "step": 3630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087746523126254e-06,
+      "loss": 0.7214,
+      "step": 3631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082717222591366e-06,
+      "loss": 0.7166,
+      "step": 3632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077688425468144e-06,
+      "loss": 0.7412,
+      "step": 3633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072660133076976e-06,
+      "loss": 0.7925,
+      "step": 3634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067632346738109e-06,
+      "loss": 0.7207,
+      "step": 3635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062605067771674e-06,
+      "loss": 0.7446,
+      "step": 3636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057578297497655e-06,
+      "loss": 0.7949,
+      "step": 3637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052552037235906e-06,
+      "loss": 0.7203,
+      "step": 3638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04752628830615e-06,
+      "loss": 0.7847,
+      "step": 3639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042501052027973e-06,
+      "loss": 0.6727,
+      "step": 3640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037476329720822e-06,
+      "loss": 0.7817,
+      "step": 3641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032452122704022e-06,
+      "loss": 0.7288,
+      "step": 3642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027428432296751e-06,
+      "loss": 0.7207,
+      "step": 3643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022405259818054e-06,
+      "loss": 0.7179,
+      "step": 3644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01738260658684e-06,
+      "loss": 0.7751,
+      "step": 3645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01236047392189e-06,
+      "loss": 0.6804,
+      "step": 3646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007338863141834e-06,
+      "loss": 0.7534,
+      "step": 3647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002317775565172e-06,
+      "loss": 0.7598,
+      "step": 3648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99729721251027e-06,
+      "loss": 0.7858,
+      "step": 3649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992277175295354e-06,
+      "loss": 0.7264,
+      "step": 3650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98725766523851e-06,
+      "loss": 0.7327,
+      "step": 3651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.982238683657683e-06,
+      "loss": 0.7474,
+      "step": 3652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977220231870685e-06,
+      "loss": 0.2692,
+      "step": 3653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972202311195185e-06,
+      "loss": 0.7286,
+      "step": 3654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967184922948717e-06,
+      "loss": 0.7781,
+      "step": 3655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962168068448672e-06,
+      "loss": 0.7622,
+      "step": 3656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957151749012299e-06,
+      "loss": 0.702,
+      "step": 3657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952135965956709e-06,
+      "loss": 0.6796,
+      "step": 3658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947120720598875e-06,
+      "loss": 0.7761,
+      "step": 3659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942106014255623e-06,
+      "loss": 0.7395,
+      "step": 3660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937091848243641e-06,
+      "loss": 0.7463,
+      "step": 3661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932078223879473e-06,
+      "loss": 0.769,
+      "step": 3662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927065142479529e-06,
+      "loss": 0.7157,
+      "step": 3663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922052605360055e-06,
+      "loss": 0.7183,
+      "step": 3664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917040613837182e-06,
+      "loss": 0.7007,
+      "step": 3665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91202916922688e-06,
+      "loss": 0.7495,
+      "step": 3666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907018272844979e-06,
+      "loss": 0.7693,
+      "step": 3667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902007926007164e-06,
+      "loss": 0.7444,
+      "step": 3668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896998130028982e-06,
+      "loss": 0.7917,
+      "step": 3669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891988886225828e-06,
+      "loss": 0.701,
+      "step": 3670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886980195912957e-06,
+      "loss": 0.7544,
+      "step": 3671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881972060405472e-06,
+      "loss": 0.7327,
+      "step": 3672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876964481018343e-06,
+      "loss": 0.7075,
+      "step": 3673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871957459066378e-06,
+      "loss": 0.7346,
+      "step": 3674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866950995864254e-06,
+      "loss": 0.8245,
+      "step": 3675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861945092726485e-06,
+      "loss": 0.73,
+      "step": 3676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856939750967454e-06,
+      "loss": 0.7649,
+      "step": 3677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851934971901382e-06,
+      "loss": 0.7463,
+      "step": 3678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846930756842358e-06,
+      "loss": 0.6406,
+      "step": 3679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841927107104308e-06,
+      "loss": 0.7842,
+      "step": 3680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836924024001023e-06,
+      "loss": 0.7822,
+      "step": 3681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83192150884613e-06,
+      "loss": 0.7322,
+      "step": 3682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826919562953121e-06,
+      "loss": 0.7112,
+      "step": 3683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821918187635331e-06,
+      "loss": 0.7787,
+      "step": 3684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816917384205948e-06,
+      "loss": 0.8103,
+      "step": 3685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81191715397801e-06,
+      "loss": 0.8027,
+      "step": 3686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8069174982644e-06,
+      "loss": 0.7379,
+      "step": 3687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801918418377852e-06,
+      "loss": 0.688,
+      "step": 3688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796919915630954e-06,
+      "loss": 0.7668,
+      "step": 3689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791921991336138e-06,
+      "loss": 0.7058,
+      "step": 3690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786924646805684e-06,
+      "loss": 0.7603,
+      "step": 3691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781927883351717e-06,
+      "loss": 0.7632,
+      "step": 3692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.776931702286221e-06,
+      "loss": 0.7952,
+      "step": 3693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771936104921015e-06,
+      "loss": 0.7019,
+      "step": 3694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76694109256777e-06,
+      "loss": 0.7319,
+      "step": 3695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761946666537995e-06,
+      "loss": 0.7168,
+      "step": 3696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756952828143065e-06,
+      "loss": 0.736,
+      "step": 3697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751959578694181e-06,
+      "loss": 0.8074,
+      "step": 3698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.746966919502395e-06,
+      "loss": 0.7654,
+      "step": 3699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741974851878606e-06,
+      "loss": 0.7495,
+      "step": 3700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736983377133558e-06,
+      "loss": 0.7947,
+      "step": 3701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731992496577839e-06,
+      "loss": 0.7206,
+      "step": 3702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727002211521879e-06,
+      "loss": 0.7356,
+      "step": 3703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722012523275956e-06,
+      "loss": 0.783,
+      "step": 3704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717023433150183e-06,
+      "loss": 0.7152,
+      "step": 3705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712034942454525e-06,
+      "loss": 0.7198,
+      "step": 3706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707047052498788e-06,
+      "loss": 0.3264,
+      "step": 3707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702059764592614e-06,
+      "loss": 0.7513,
+      "step": 3708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69707308004549e-06,
+      "loss": 0.7166,
+      "step": 3709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692087000166755e-06,
+      "loss": 0.697,
+      "step": 3710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687101526265567e-06,
+      "loss": 0.6964,
+      "step": 3711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682116659650947e-06,
+      "loss": 0.6915,
+      "step": 3712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677132401631742e-06,
+      "loss": 0.709,
+      "step": 3713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672148753516648e-06,
+      "loss": 0.7476,
+      "step": 3714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667165716614193e-06,
+      "loss": 0.6785,
+      "step": 3715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662183292232755e-06,
+      "loss": 0.7528,
+      "step": 3716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65720148168054e-06,
+      "loss": 0.7664,
+      "step": 3717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652220286265602e-06,
+      "loss": 0.6561,
+      "step": 3718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647239707295823e-06,
+      "loss": 0.6343,
+      "step": 3719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642259746078937e-06,
+      "loss": 0.7522,
+      "step": 3720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637280403922505e-06,
+      "loss": 0.7037,
+      "step": 3721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632301682133931e-06,
+      "loss": 0.7859,
+      "step": 3722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627323582020451e-06,
+      "loss": 0.774,
+      "step": 3723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622346104889139e-06,
+      "loss": 0.6943,
+      "step": 3724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617369252046909e-06,
+      "loss": 0.7805,
+      "step": 3725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612393024800511e-06,
+      "loss": 0.7751,
+      "step": 3726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6074174244565295e-06,
+      "loss": 0.7365,
+      "step": 3727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602442452321379e-06,
+      "loss": 0.7922,
+      "step": 3728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597468109701317e-06,
+      "loss": 0.7209,
+      "step": 3729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592494397902434e-06,
+      "loss": 0.687,
+      "step": 3730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587521318230651e-06,
+      "loss": 0.7341,
+      "step": 3731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582548871991726e-06,
+      "loss": 0.3177,
+      "step": 3732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57757706049125e-06,
+      "loss": 0.7279,
+      "step": 3733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572605885034654e-06,
+      "loss": 0.699,
+      "step": 3734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567635346927182e-06,
+      "loss": 0.2868,
+      "step": 3735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5626654474739346e-06,
+      "loss": 0.733,
+      "step": 3736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.557696187979832e-06,
+      "loss": 0.7213,
+      "step": 3737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552727569749626e-06,
+      "loss": 0.7308,
+      "step": 3738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547759594087905e-06,
+      "loss": 0.7103,
+      "step": 3739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542792262299086e-06,
+      "loss": 0.8035,
+      "step": 3740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5378255756874166e-06,
+      "loss": 0.7437,
+      "step": 3741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532859535556978e-06,
+      "loss": 0.7216,
+      "step": 3742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527894143211674e-06,
+      "loss": 0.7532,
+      "step": 3743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522929399955251e-06,
+      "loss": 0.7788,
+      "step": 3744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517965307091275e-06,
+      "loss": 0.7054,
+      "step": 3745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513001865923141e-06,
+      "loss": 0.737,
+      "step": 3746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508039077754077e-06,
+      "loss": 0.7205,
+      "step": 3747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.503076943887138e-06,
+      "loss": 0.7391,
+      "step": 3748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4981154656252085e-06,
+      "loss": 0.772,
+      "step": 3749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4931546442710005e-06,
+      "loss": 0.7524,
+      "step": 3750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.488194481127051e-06,
+      "loss": 0.71,
+      "step": 3751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.483234977495726e-06,
+      "loss": 0.7557,
+      "step": 3752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478276134679221e-06,
+      "loss": 0.6982,
+      "step": 3753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473317953979553e-06,
+      "loss": 0.7546,
+      "step": 3754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46836043669857e-06,
+      "loss": 0.7021,
+      "step": 3755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463403584137938e-06,
+      "loss": 0.6882,
+      "step": 3756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458447397599164e-06,
+      "loss": 0.7678,
+      "step": 3757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4534918783835605e-06,
+      "loss": 0.715,
+      "step": 3758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4485370277922745e-06,
+      "loss": 0.7748,
+      "step": 3759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44358284712628e-06,
+      "loss": 0.7546,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.438629337686374e-06,
+      "loss": 0.7246,
+      "step": 3761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43367650077317e-06,
+      "loss": 0.3071,
+      "step": 3762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428724337687117e-06,
+      "loss": 0.8018,
+      "step": 3763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423772849728476e-06,
+      "loss": 0.7849,
+      "step": 3764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418822038197339e-06,
+      "loss": 0.6974,
+      "step": 3765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41387190439361e-06,
+      "loss": 0.7466,
+      "step": 3766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408922449617029e-06,
+      "loss": 0.7123,
+      "step": 3767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403973675167148e-06,
+      "loss": 0.6697,
+      "step": 3768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399025582343344e-06,
+      "loss": 0.6943,
+      "step": 3769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3940781724448084e-06,
+      "loss": 0.7883,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389131446770563e-06,
+      "loss": 0.7134,
+      "step": 3771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384185406619443e-06,
+      "loss": 0.646,
+      "step": 3772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3792400532901096e-06,
+      "loss": 0.7322,
+      "step": 3773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3742953880810395e-06,
+      "loss": 0.3005,
+      "step": 3774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3693514122905285e-06,
+      "loss": 0.7361,
+      "step": 3775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364408127216691e-06,
+      "loss": 0.6897,
+      "step": 3776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.359465534157465e-06,
+      "loss": 0.7039,
+      "step": 3777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354523634410602e-06,
+      "loss": 0.6896,
+      "step": 3778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349582429273672e-06,
+      "loss": 0.7175,
+      "step": 3779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34464192004406e-06,
+      "loss": 0.7131,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339702108018983e-06,
+      "loss": 0.7529,
+      "step": 3781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334762994495448e-06,
+      "loss": 0.2865,
+      "step": 3782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329824580770305e-06,
+      "loss": 0.2943,
+      "step": 3783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324886868140206e-06,
+      "loss": 0.7886,
+      "step": 3784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3199498579016224e-06,
+      "loss": 0.7546,
+      "step": 3785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315013551350839e-06,
+      "loss": 0.6858,
+      "step": 3786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310077949783962e-06,
+      "loss": 0.7539,
+      "step": 3787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305143054496907e-06,
+      "loss": 0.7481,
+      "step": 3788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300208866785405e-06,
+      "loss": 0.6597,
+      "step": 3789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295275387945e-06,
+      "loss": 0.7802,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290342619271056e-06,
+      "loss": 0.7249,
+      "step": 3791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285410562058746e-06,
+      "loss": 0.7539,
+      "step": 3792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280479217603051e-06,
+      "loss": 0.7168,
+      "step": 3793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275548587198775e-06,
+      "loss": 0.7872,
+      "step": 3794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270618672140528e-06,
+      "loss": 0.6426,
+      "step": 3795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265689473722731e-06,
+      "loss": 0.2967,
+      "step": 3796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260760993239626e-06,
+      "loss": 0.7603,
+      "step": 3797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255833231985258e-06,
+      "loss": 0.7319,
+      "step": 3798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2509061912534835e-06,
+      "loss": 0.7151,
+      "step": 3799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245979872337972e-06,
+      "loss": 0.7446,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241054276532204e-06,
+      "loss": 0.7583,
+      "step": 3801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236129405129473e-06,
+      "loss": 0.7467,
+      "step": 3802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231205259422871e-06,
+      "loss": 0.7097,
+      "step": 3803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226281840705318e-06,
+      "loss": 0.7522,
+      "step": 3804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221359150269523e-06,
+      "loss": 0.7651,
+      "step": 3805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216437189408011e-06,
+      "loss": 0.7783,
+      "step": 3806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211515959413125e-06,
+      "loss": 0.7668,
+      "step": 3807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206595461577007e-06,
+      "loss": 0.7783,
+      "step": 3808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201675697191603e-06,
+      "loss": 0.7402,
+      "step": 3809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196756667548679e-06,
+      "loss": 0.7754,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191838373939797e-06,
+      "loss": 0.7344,
+      "step": 3811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186920817656331e-06,
+      "loss": 0.7507,
+      "step": 3812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182003999989455e-06,
+      "loss": 0.702,
+      "step": 3813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177087922230161e-06,
+      "loss": 0.6774,
+      "step": 3814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172172585669238e-06,
+      "loss": 0.7722,
+      "step": 3815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1672579915972825e-06,
+      "loss": 0.7622,
+      "step": 3816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162344141304692e-06,
+      "loss": 0.7426,
+      "step": 3817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157431036081675e-06,
+      "loss": 0.6771,
+      "step": 3818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1525186772182385e-06,
+      "loss": 0.7744,
+      "step": 3819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147607066004202e-06,
+      "loss": 0.7842,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142696203729181e-06,
+      "loss": 0.7186,
+      "step": 3821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137786091682597e-06,
+      "loss": 0.798,
+      "step": 3822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1328767311536724e-06,
+      "loss": 0.6835,
+      "step": 3823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1279681234314395e-06,
+      "loss": 0.7256,
+      "step": 3824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123060269804722e-06,
+      "loss": 0.7329,
+      "step": 3825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118153171562157e-06,
+      "loss": 0.7561,
+      "step": 3826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113246829992169e-06,
+      "loss": 0.708,
+      "step": 3827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1083412463830055e-06,
+      "loss": 0.7273,
+      "step": 3828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.103436422022687e-06,
+      "loss": 0.765,
+      "step": 3829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098532358199059e-06,
+      "loss": 0.3041,
+      "step": 3830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0936290561997555e-06,
+      "loss": 0.6841,
+      "step": 3831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088726517312212e-06,
+      "loss": 0.7784,
+      "step": 3832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083824742823664e-06,
+      "loss": 0.7188,
+      "step": 3833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07892373402115e-06,
+      "loss": 0.7377,
+      "step": 3834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074023492191499e-06,
+      "loss": 0.7454,
+      "step": 3835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06912401862135e-06,
+      "loss": 0.7531,
+      "step": 3836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064225314597129e-06,
+      "loss": 0.7371,
+      "step": 3837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059327381405065e-06,
+      "loss": 0.6816,
+      "step": 3838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0544302203311896e-06,
+      "loss": 0.7319,
+      "step": 3839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049533832661326e-06,
+      "loss": 0.7292,
+      "step": 3840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044638219681088e-06,
+      "loss": 0.7411,
+      "step": 3841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0397433826759e-06,
+      "loss": 0.7843,
+      "step": 3842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03484932293097e-06,
+      "loss": 0.2507,
+      "step": 3843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029956041731313e-06,
+      "loss": 0.6815,
+      "step": 3844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0250635403617315e-06,
+      "loss": 0.7078,
+      "step": 3845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020171820106827e-06,
+      "loss": 0.7533,
+      "step": 3846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015280882250992e-06,
+      "loss": 0.7529,
+      "step": 3847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010390728078423e-06,
+      "loss": 0.76,
+      "step": 3848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005501358873097e-06,
+      "loss": 0.7732,
+      "step": 3849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000612775918796e-06,
+      "loss": 0.7549,
+      "step": 3850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995724980499092e-06,
+      "loss": 0.7795,
+      "step": 3851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990837973897346e-06,
+      "loss": 0.6522,
+      "step": 3852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985951757396716e-06,
+      "loss": 0.6882,
+      "step": 3853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9810663322801555e-06,
+      "loss": 0.7046,
+      "step": 3854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9761816998304065e-06,
+      "loss": 0.7563,
+      "step": 3855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971297861330002e-06,
+      "loss": 0.7407,
+      "step": 3856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966414818061267e-06,
+      "loss": 0.8093,
+      "step": 3857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961532571306321e-06,
+      "loss": 0.2752,
+      "step": 3858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956651122347071e-06,
+      "loss": 0.2761,
+      "step": 3859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951770472465214e-06,
+      "loss": 0.7466,
+      "step": 3860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94689062294224e-06,
+      "loss": 0.7888,
+      "step": 3861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942011575059429e-06,
+      "loss": 0.7288,
+      "step": 3862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937133330097847e-06,
+      "loss": 0.7332,
+      "step": 3863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932255889338351e-06,
+      "loss": 0.759,
+      "step": 3864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927379254061588e-06,
+      "loss": 0.6635,
+      "step": 3865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922503425547988e-06,
+      "loss": 0.7751,
+      "step": 3866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9176284050777806e-06,
+      "loss": 0.7228,
+      "step": 3867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912754193930975e-06,
+      "loss": 0.3268,
+      "step": 3868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.907880793387366e-06,
+      "loss": 0.7305,
+      "step": 3869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903008204726539e-06,
+      "loss": 0.7167,
+      "step": 3870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89813642922787e-06,
+      "loss": 0.7183,
+      "step": 3871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.893265468170514e-06,
+      "loss": 0.7115,
+      "step": 3872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8883953228334175e-06,
+      "loss": 0.7344,
+      "step": 3873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883525994495308e-06,
+      "loss": 0.7556,
+      "step": 3874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878657484434709e-06,
+      "loss": 0.7395,
+      "step": 3875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8737897939299124e-06,
+      "loss": 0.7449,
+      "step": 3876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868922924259009e-06,
+      "loss": 0.7251,
+      "step": 3877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864056876699869e-06,
+      "loss": 0.7269,
+      "step": 3878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859191652530146e-06,
+      "loss": 0.7246,
+      "step": 3879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8543272530272756e-06,
+      "loss": 0.7227,
+      "step": 3880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849463679468487e-06,
+      "loss": 0.6744,
+      "step": 3881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844600933130778e-06,
+      "loss": 0.7085,
+      "step": 3882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839739015290942e-06,
+      "loss": 0.7545,
+      "step": 3883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8348779272255425e-06,
+      "loss": 0.7404,
+      "step": 3884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83001767021094e-06,
+      "loss": 0.7053,
+      "step": 3885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825158245523263e-06,
+      "loss": 0.7415,
+      "step": 3886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820299654438431e-06,
+      "loss": 0.2744,
+      "step": 3887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815441898232138e-06,
+      "loss": 0.7274,
+      "step": 3888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810584978179861e-06,
+      "loss": 0.7339,
+      "step": 3889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805728895556856e-06,
+      "loss": 0.7224,
+      "step": 3890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800873651638167e-06,
+      "loss": 0.7378,
+      "step": 3891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796019247698611e-06,
+      "loss": 0.7432,
+      "step": 3892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791165685012782e-06,
+      "loss": 0.7375,
+      "step": 3893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786312964855055e-06,
+      "loss": 0.7478,
+      "step": 3894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78146108849959e-06,
+      "loss": 0.7427,
+      "step": 3895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776610057220321e-06,
+      "loss": 0.8,
+      "step": 3896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771759872290957e-06,
+      "loss": 0.717,
+      "step": 3897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7669105349849865e-06,
+      "loss": 0.6713,
+      "step": 3898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762062046575684e-06,
+      "loss": 0.6855,
+      "step": 3899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7572144083360815e-06,
+      "loss": 0.7236,
+      "step": 3900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752367621539009e-06,
+      "loss": 0.767,
+      "step": 3901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.747521687457061e-06,
+      "loss": 0.7212,
+      "step": 3902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7426766073626125e-06,
+      "loss": 0.7301,
+      "step": 3903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737832382527809e-06,
+      "loss": 0.7207,
+      "step": 3904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732989014224579e-06,
+      "loss": 0.7671,
+      "step": 3905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728146503724622e-06,
+      "loss": 0.7317,
+      "step": 3906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723304852299408e-06,
+      "loss": 0.741,
+      "step": 3907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718464061220193e-06,
+      "loss": 0.6932,
+      "step": 3908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713624131757996e-06,
+      "loss": 0.7495,
+      "step": 3909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708785065183616e-06,
+      "loss": 0.6538,
+      "step": 3910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703946862767622e-06,
+      "loss": 0.7566,
+      "step": 3911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699109525780356e-06,
+      "loss": 0.7111,
+      "step": 3912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694273055491935e-06,
+      "loss": 0.7534,
+      "step": 3913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689437453172246e-06,
+      "loss": 0.2639,
+      "step": 3914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684602720090954e-06,
+      "loss": 0.709,
+      "step": 3915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6797688575174905e-06,
+      "loss": 0.6785,
+      "step": 3916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674935866721054e-06,
+      "loss": 0.7594,
+      "step": 3917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670103748970626e-06,
+      "loss": 0.7427,
+      "step": 3918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665272505534951e-06,
+      "loss": 0.7822,
+      "step": 3919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660442137682543e-06,
+      "loss": 0.7695,
+      "step": 3920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6556126466816885e-06,
+      "loss": 0.7764,
+      "step": 3921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65078403380045e-06,
+      "loss": 0.7302,
+      "step": 3922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645956300306642e-06,
+      "loss": 0.7522,
+      "step": 3923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641129447467867e-06,
+      "loss": 0.7263,
+      "step": 3924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.636303476551485e-06,
+      "loss": 0.8127,
+      "step": 3925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6314783888246305e-06,
+      "loss": 0.7722,
+      "step": 3926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626654185554199e-06,
+      "loss": 0.8096,
+      "step": 3927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621830868006863e-06,
+      "loss": 0.7095,
+      "step": 3928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6170084374490566e-06,
+      "loss": 0.6808,
+      "step": 3929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.61218689514698e-06,
+      "loss": 0.6677,
+      "step": 3930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6073662423666016e-06,
+      "loss": 0.7773,
+      "step": 3931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.602546480373662e-06,
+      "loss": 0.813,
+      "step": 3932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597727610433659e-06,
+      "loss": 0.3029,
+      "step": 3933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592909633811862e-06,
+      "loss": 0.7034,
+      "step": 3934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5880925517733015e-06,
+      "loss": 0.7303,
+      "step": 3935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5832763655827755e-06,
+      "loss": 0.7205,
+      "step": 3936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578461076504845e-06,
+      "loss": 0.7487,
+      "step": 3937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573646685803843e-06,
+      "loss": 0.8064,
+      "step": 3938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568833194743858e-06,
+      "loss": 0.7512,
+      "step": 3939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564020604588742e-06,
+      "loss": 0.7042,
+      "step": 3940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5592089166021155e-06,
+      "loss": 0.7062,
+      "step": 3941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5543981320473626e-06,
+      "loss": 0.6772,
+      "step": 3942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549588252187626e-06,
+      "loss": 0.2611,
+      "step": 3943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544779278285811e-06,
+      "loss": 0.7502,
+      "step": 3944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5399712116045875e-06,
+      "loss": 0.6898,
+      "step": 3945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535164053406393e-06,
+      "loss": 0.7245,
+      "step": 3946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530357804953407e-06,
+      "loss": 0.7451,
+      "step": 3947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.525552467507592e-06,
+      "loss": 0.6798,
+      "step": 3948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520748042330659e-06,
+      "loss": 0.7817,
+      "step": 3949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515944530684084e-06,
+      "loss": 0.7266,
+      "step": 3950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5111419338290985e-06,
+      "loss": 0.6853,
+      "step": 3951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506340253026702e-06,
+      "loss": 0.7524,
+      "step": 3952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501539489537647e-06,
+      "loss": 0.7668,
+      "step": 3953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496739644622445e-06,
+      "loss": 0.6897,
+      "step": 3954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.491940719541369e-06,
+      "loss": 0.7,
+      "step": 3955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487142715554451e-06,
+      "loss": 0.7651,
+      "step": 3956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482345633921478e-06,
+      "loss": 0.7174,
+      "step": 3957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477549475902e-06,
+      "loss": 0.7374,
+      "step": 3958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4727542427553144e-06,
+      "loss": 0.7511,
+      "step": 3959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467959935740488e-06,
+      "loss": 0.7529,
+      "step": 3960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463166556116334e-06,
+      "loss": 0.7238,
+      "step": 3961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458374105141432e-06,
+      "loss": 0.733,
+      "step": 3962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453582584074109e-06,
+      "loss": 0.7391,
+      "step": 3963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448791994172452e-06,
+      "loss": 0.7485,
+      "step": 3964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444002336694306e-06,
+      "loss": 0.7067,
+      "step": 3965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439213612897266e-06,
+      "loss": 0.7098,
+      "step": 3966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434425824038683e-06,
+      "loss": 0.741,
+      "step": 3967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4296389713756645e-06,
+      "loss": 0.7495,
+      "step": 3968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424853056165077e-06,
+      "loss": 0.749,
+      "step": 3969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420068079663523e-06,
+      "loss": 0.7554,
+      "step": 3970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4152840431273786e-06,
+      "loss": 0.7162,
+      "step": 3971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410500947812765e-06,
+      "loss": 0.7332,
+      "step": 3972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405718794975555e-06,
+      "loss": 0.7336,
+      "step": 3973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400937585871374e-06,
+      "loss": 0.679,
+      "step": 3974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396157321755603e-06,
+      "loss": 0.7114,
+      "step": 3975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391378003883373e-06,
+      "loss": 0.7241,
+      "step": 3976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386599633509565e-06,
+      "loss": 0.7869,
+      "step": 3977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381822211888813e-06,
+      "loss": 0.7665,
+      "step": 3978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377045740275501e-06,
+      "loss": 0.7198,
+      "step": 3979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372270219923766e-06,
+      "loss": 0.7327,
+      "step": 3980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3674956520874955e-06,
+      "loss": 0.7448,
+      "step": 3981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362722038020317e-06,
+      "loss": 0.7111,
+      "step": 3982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357949378975621e-06,
+      "loss": 0.8069,
+      "step": 3983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3531776762065375e-06,
+      "loss": 0.6901,
+      "step": 3984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348406930965953e-06,
+      "loss": 0.6907,
+      "step": 3985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3436371445065e-06,
+      "loss": 0.6649,
+      "step": 3986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3388683180805546e-06,
+      "loss": 0.7478,
+      "step": 3987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334100452940244e-06,
+      "loss": 0.2605,
+      "step": 3988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329333550337449e-06,
+      "loss": 0.7632,
+      "step": 3989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3245676115237864e-06,
+      "loss": 0.7629,
+      "step": 3990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31980263775063e-06,
+      "loss": 0.7391,
+      "step": 3991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31503863026909e-06,
+      "loss": 0.6847,
+      "step": 3992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310275590330041e-06,
+      "loss": 0.6882,
+      "step": 3993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305513519184074e-06,
+      "loss": 0.7262,
+      "step": 3994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300752418081557e-06,
+      "loss": 0.6981,
+      "step": 3995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295992288272583e-06,
+      "loss": 0.7736,
+      "step": 3996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291233131006997e-06,
+      "loss": 0.6339,
+      "step": 3997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2864749475343886e-06,
+      "loss": 0.708,
+      "step": 3998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2817177391040915e-06,
+      "loss": 0.6512,
+      "step": 3999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276961506965182e-06,
+      "loss": 0.7979,
+      "step": 4000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272206252366482e-06,
+      "loss": 0.7404,
+      "step": 4001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267451976556553e-06,
+      "loss": 0.7422,
+      "step": 4002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262698680783705e-06,
+      "loss": 0.7267,
+      "step": 4003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2579463662959886e-06,
+      "loss": 0.7218,
+      "step": 4004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.253195034341196e-06,
+      "loss": 0.7246,
+      "step": 4005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248444686166858e-06,
+      "loss": 0.7554,
+      "step": 4006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243695323020253e-06,
+      "loss": 0.7146,
+      "step": 4007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238946946148395e-06,
+      "loss": 0.7205,
+      "step": 4008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2341995567980465e-06,
+      "loss": 0.736,
+      "step": 4009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2294531562157056e-06,
+      "loss": 0.78,
+      "step": 4010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224707745647612e-06,
+      "loss": 0.7466,
+      "step": 4011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219963326339742e-06,
+      "loss": 0.7866,
+      "step": 4012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215219899537821e-06,
+      "loss": 0.6884,
+      "step": 4013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.210477466487302e-06,
+      "loss": 0.7654,
+      "step": 4014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205736028433383e-06,
+      "loss": 0.72,
+      "step": 4015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200995586621009e-06,
+      "loss": 0.6648,
+      "step": 4016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196256142294843e-06,
+      "loss": 0.7137,
+      "step": 4017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1915176966993025e-06,
+      "loss": 0.2668,
+      "step": 4018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186780251078539e-06,
+      "loss": 0.7108,
+      "step": 4019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.182043806676442e-06,
+      "loss": 0.7351,
+      "step": 4020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177308364736633e-06,
+      "loss": 0.7529,
+      "step": 4021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.172573926502477e-06,
+      "loss": 0.7207,
+      "step": 4022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167840493217074e-06,
+      "loss": 0.7505,
+      "step": 4023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163108066123256e-06,
+      "loss": 0.6742,
+      "step": 4024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158376646463591e-06,
+      "loss": 0.7617,
+      "step": 4025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1536462354803904e-06,
+      "loss": 0.7211,
+      "step": 4026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148916834415693e-06,
+      "loss": 0.7114,
+      "step": 4027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144188444511276e-06,
+      "loss": 0.7332,
+      "step": 4028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139461067008647e-06,
+      "loss": 0.7101,
+      "step": 4029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1347347031490525e-06,
+      "loss": 0.7418,
+      "step": 4030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130009354173469e-06,
+      "loss": 0.7576,
+      "step": 4031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125285021322612e-06,
+      "loss": 0.7896,
+      "step": 4032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120561705836925e-06,
+      "loss": 0.761,
+      "step": 4033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115839408956585e-06,
+      "loss": 0.7338,
+      "step": 4034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111118131921503e-06,
+      "loss": 0.6663,
+      "step": 4035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1063978759713246e-06,
+      "loss": 0.7224,
+      "step": 4036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101678642345423e-06,
+      "loss": 0.7001,
+      "step": 4037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096960432282904e-06,
+      "loss": 0.7209,
+      "step": 4038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092243247022602e-06,
+      "loss": 0.7708,
+      "step": 4039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087527087803096e-06,
+      "loss": 0.6956,
+      "step": 4040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082811955862671e-06,
+      "loss": 0.6914,
+      "step": 4041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078097852439367e-06,
+      "loss": 0.3215,
+      "step": 4042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073384778770937e-06,
+      "loss": 0.739,
+      "step": 4043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068672736094874e-06,
+      "loss": 0.7494,
+      "step": 4044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0639617256483926e-06,
+      "loss": 0.683,
+      "step": 4045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059251748668444e-06,
+      "loss": 0.652,
+      "step": 4046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0545428063917035e-06,
+      "loss": 0.7273,
+      "step": 4047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049834900054574e-06,
+      "loss": 0.7278,
+      "step": 4048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045128030893186e-06,
+      "loss": 0.7042,
+      "step": 4049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040422200143403e-06,
+      "loss": 0.769,
+      "step": 4050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03571740904081e-06,
+      "loss": 0.7266,
+      "step": 4051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031013658820724e-06,
+      "loss": 0.6619,
+      "step": 4052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026310950718182e-06,
+      "loss": 0.7361,
+      "step": 4053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021609285967951e-06,
+      "loss": 0.7324,
+      "step": 4054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016908665804526e-06,
+      "loss": 0.6616,
+      "step": 4055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012209091462129e-06,
+      "loss": 0.7494,
+      "step": 4056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007510564174701e-06,
+      "loss": 0.7559,
+      "step": 4057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002813085175913e-06,
+      "loss": 0.7512,
+      "step": 4058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9981166556991556e-06,
+      "loss": 0.7175,
+      "step": 4059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993421276977554e-06,
+      "loss": 0.7759,
+      "step": 4060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9887269502439485e-06,
+      "loss": 0.7476,
+      "step": 4061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.984033676730902e-06,
+      "loss": 0.7603,
+      "step": 4062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979341457670709e-06,
+      "loss": 0.6823,
+      "step": 4063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9746502942953875e-06,
+      "loss": 0.7477,
+      "step": 4064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969960187836661e-06,
+      "loss": 0.7695,
+      "step": 4065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9652711395259955e-06,
+      "loss": 0.7061,
+      "step": 4066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9605831505945725e-06,
+      "loss": 0.7015,
+      "step": 4067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955896222273292e-06,
+      "loss": 0.7285,
+      "step": 4068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951210355792778e-06,
+      "loss": 0.6581,
+      "step": 4069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946525552383377e-06,
+      "loss": 0.7402,
+      "step": 4070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941841813275157e-06,
+      "loss": 0.2634,
+      "step": 4071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9371591396979e-06,
+      "loss": 0.7834,
+      "step": 4072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932477532881118e-06,
+      "loss": 0.7693,
+      "step": 4073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927796994054036e-06,
+      "loss": 0.7091,
+      "step": 4074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923117524445601e-06,
+      "loss": 0.7104,
+      "step": 4075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918439125284476e-06,
+      "loss": 0.7432,
+      "step": 4076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.913761797799049e-06,
+      "loss": 0.7866,
+      "step": 4077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909085543217418e-06,
+      "loss": 0.7607,
+      "step": 4078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90441036276741e-06,
+      "loss": 0.6655,
+      "step": 4079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8997362576765635e-06,
+      "loss": 0.686,
+      "step": 4080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895063229172136e-06,
+      "loss": 0.7148,
+      "step": 4081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8903912784810955e-06,
+      "loss": 0.6677,
+      "step": 4082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885720406830143e-06,
+      "loss": 0.7206,
+      "step": 4083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881050615445683e-06,
+      "loss": 0.7285,
+      "step": 4084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8763819055538385e-06,
+      "loss": 0.712,
+      "step": 4085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8717142783804495e-06,
+      "loss": 0.7129,
+      "step": 4086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86704773515108e-06,
+      "loss": 0.7091,
+      "step": 4087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86238227709099e-06,
+      "loss": 0.7251,
+      "step": 4088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857717905425173e-06,
+      "loss": 0.6947,
+      "step": 4089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853054621378329e-06,
+      "loss": 0.7715,
+      "step": 4090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848392426174876e-06,
+      "loss": 0.7092,
+      "step": 4091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843731321038939e-06,
+      "loss": 0.7329,
+      "step": 4092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8390713071943686e-06,
+      "loss": 0.7644,
+      "step": 4093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834412385864716e-06,
+      "loss": 0.7013,
+      "step": 4094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829754558273257e-06,
+      "loss": 0.7557,
+      "step": 4095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.825097825642971e-06,
+      "loss": 0.7485,
+      "step": 4096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8204421891965555e-06,
+      "loss": 0.689,
+      "step": 4097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8157876501564124e-06,
+      "loss": 0.7271,
+      "step": 4098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811134209744677e-06,
+      "loss": 0.6821,
+      "step": 4099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8064818691831645e-06,
+      "loss": 0.7128,
+      "step": 4100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801830629693425e-06,
+      "loss": 0.7198,
+      "step": 4101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797180492496711e-06,
+      "loss": 0.7126,
+      "step": 4102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7925314588139855e-06,
+      "loss": 0.6918,
+      "step": 4103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787883529865918e-06,
+      "loss": 0.7155,
+      "step": 4104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783236706872905e-06,
+      "loss": 0.7261,
+      "step": 4105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778590991055031e-06,
+      "loss": 0.7556,
+      "step": 4106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773946383632102e-06,
+      "loss": 0.7148,
+      "step": 4107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76930288582363e-06,
+      "loss": 0.7286,
+      "step": 4108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764660498848835e-06,
+      "loss": 0.7522,
+      "step": 4109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760019223926647e-06,
+      "loss": 0.7542,
+      "step": 4110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755379062275702e-06,
+      "loss": 0.7118,
+      "step": 4111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750740015114347e-06,
+      "loss": 0.7683,
+      "step": 4112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74610208366063e-06,
+      "loss": 0.7736,
+      "step": 4113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741465269132315e-06,
+      "loss": 0.7554,
+      "step": 4114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7368295727468645e-06,
+      "loss": 0.7449,
+      "step": 4115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7321949957214515e-06,
+      "loss": 0.6481,
+      "step": 4116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727561539272957e-06,
+      "loss": 0.7207,
+      "step": 4117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722929204617958e-06,
+      "loss": 0.7236,
+      "step": 4118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718297992972754e-06,
+      "loss": 0.2934,
+      "step": 4119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713667905553336e-06,
+      "loss": 0.7654,
+      "step": 4120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709038943575402e-06,
+      "loss": 0.2787,
+      "step": 4121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704411108254358e-06,
+      "loss": 0.6847,
+      "step": 4122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699784400805316e-06,
+      "loss": 0.7065,
+      "step": 4123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695158822443075e-06,
+      "loss": 0.7684,
+      "step": 4124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690534374382164e-06,
+      "loss": 0.7021,
+      "step": 4125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685911057836797e-06,
+      "loss": 0.7096,
+      "step": 4126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.681288874020896e-06,
+      "loss": 0.7634,
+      "step": 4127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6766678241480855e-06,
+      "loss": 0.6444,
+      "step": 4128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672047909431694e-06,
+      "loss": 0.7472,
+      "step": 4129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667429131084747e-06,
+      "loss": 0.6394,
+      "step": 4130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662811490319974e-06,
+      "loss": 0.7739,
+      "step": 4131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658194988349812e-06,
+      "loss": 0.7264,
+      "step": 4132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653579626386391e-06,
+      "loss": 0.687,
+      "step": 4133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648965405641546e-06,
+      "loss": 0.7476,
+      "step": 4134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644352327326807e-06,
+      "loss": 0.7491,
+      "step": 4135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639740392653409e-06,
+      "loss": 0.6986,
+      "step": 4136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635129602832281e-06,
+      "loss": 0.7219,
+      "step": 4137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6305199590740635e-06,
+      "loss": 0.7388,
+      "step": 4138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.625911462589086e-06,
+      "loss": 0.7506,
+      "step": 4139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621304114587378e-06,
+      "loss": 0.7478,
+      "step": 4140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616697916278669e-06,
+      "loss": 0.7639,
+      "step": 4141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612092868872387e-06,
+      "loss": 0.7576,
+      "step": 4142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607488973577655e-06,
+      "loss": 0.7275,
+      "step": 4143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602886231603297e-06,
+      "loss": 0.7726,
+      "step": 4144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598284644157828e-06,
+      "loss": 0.7473,
+      "step": 4145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593684212449477e-06,
+      "loss": 0.7302,
+      "step": 4146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589084937686145e-06,
+      "loss": 0.7675,
+      "step": 4147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584486821075447e-06,
+      "loss": 0.769,
+      "step": 4148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579889863824685e-06,
+      "loss": 0.7034,
+      "step": 4149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575294067140862e-06,
+      "loss": 0.7533,
+      "step": 4150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570699432230669e-06,
+      "loss": 0.7764,
+      "step": 4151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566105960300505e-06,
+      "loss": 0.6934,
+      "step": 4152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5615136525564515e-06,
+      "loss": 0.7485,
+      "step": 4153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556922510204292e-06,
+      "loss": 0.683,
+      "step": 4154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552332534449496e-06,
+      "loss": 0.6873,
+      "step": 4155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5477437264972344e-06,
+      "loss": 0.7495,
+      "step": 4156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543156087552366e-06,
+      "loss": 0.7217,
+      "step": 4157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.538569618819446e-06,
+      "loss": 0.7622,
+      "step": 4158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533984321502721e-06,
+      "loss": 0.7235,
+      "step": 4159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529400196806132e-06,
+      "loss": 0.7012,
+      "step": 4160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524817245933307e-06,
+      "loss": 0.7085,
+      "step": 4161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5202354700875724e-06,
+      "loss": 0.7732,
+      "step": 4162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5156548704719426e-06,
+      "loss": 0.8086,
+      "step": 4163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.511075448289121e-06,
+      "loss": 0.686,
+      "step": 4164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506497204741503e-06,
+      "loss": 0.6962,
+      "step": 4165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5019201410311826e-06,
+      "loss": 0.7032,
+      "step": 4166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497344258359933e-06,
+      "loss": 0.2615,
+      "step": 4167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492769557929224e-06,
+      "loss": 0.7769,
+      "step": 4168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48819604094021e-06,
+      "loss": 0.6368,
+      "step": 4169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483623708593742e-06,
+      "loss": 0.6686,
+      "step": 4170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479052562090342e-06,
+      "loss": 0.677,
+      "step": 4171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474482602630248e-06,
+      "loss": 0.6665,
+      "step": 4172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469913831413367e-06,
+      "loss": 0.7131,
+      "step": 4173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4653462496392985e-06,
+      "loss": 0.7471,
+      "step": 4174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460779858507333e-06,
+      "loss": 0.7397,
+      "step": 4175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456214659216441e-06,
+      "loss": 0.6292,
+      "step": 4176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4516506529652906e-06,
+      "loss": 0.7394,
+      "step": 4177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447087840952221e-06,
+      "loss": 0.7422,
+      "step": 4178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442526224375281e-06,
+      "loss": 0.6925,
+      "step": 4179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4379658044321845e-06,
+      "loss": 0.6659,
+      "step": 4180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433406582320341e-06,
+      "loss": 0.7715,
+      "step": 4181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428848559236846e-06,
+      "loss": 0.7576,
+      "step": 4182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424291736378471e-06,
+      "loss": 0.7256,
+      "step": 4183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419736114941676e-06,
+      "loss": 0.2783,
+      "step": 4184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41518169612262e-06,
+      "loss": 0.7764,
+      "step": 4185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410628481117131e-06,
+      "loss": 0.7463,
+      "step": 4186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4060764711207205e-06,
+      "loss": 0.7336,
+      "step": 4187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401525667328591e-06,
+      "loss": 0.6849,
+      "step": 4188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396976070935624e-06,
+      "loss": 0.6533,
+      "step": 4189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392427683136385e-06,
+      "loss": 0.7585,
+      "step": 4190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3878805051251225e-06,
+      "loss": 0.7716,
+      "step": 4191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383334538095763e-06,
+      "loss": 0.7622,
+      "step": 4192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378789783241932e-06,
+      "loss": 0.7771,
+      "step": 4193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374246241756911e-06,
+      "loss": 0.7017,
+      "step": 4194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369703914833679e-06,
+      "loss": 0.7384,
+      "step": 4195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365162803664892e-06,
+      "loss": 0.7634,
+      "step": 4196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36062290944289e-06,
+      "loss": 0.709,
+      "step": 4197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356084233359686e-06,
+      "loss": 0.7369,
+      "step": 4198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351546776606986e-06,
+      "loss": 0.7129,
+      "step": 4199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347010540376165e-06,
+      "loss": 0.7238,
+      "step": 4200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342475525858279e-06,
+      "loss": 0.7041,
+      "step": 4201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337941734244068e-06,
+      "loss": 0.7166,
+      "step": 4202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333409166723943e-06,
+      "loss": 0.707,
+      "step": 4203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328877824488001e-06,
+      "loss": 0.7249,
+      "step": 4204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3243477087260155e-06,
+      "loss": 0.7722,
+      "step": 4205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319818820627436e-06,
+      "loss": 0.7183,
+      "step": 4206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3152911613813895e-06,
+      "loss": 0.6953,
+      "step": 4207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310764732176683e-06,
+      "loss": 0.6949,
+      "step": 4208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3062395342018e-06,
+      "loss": 0.6716,
+      "step": 4209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301715568644899e-06,
+      "loss": 0.7163,
+      "step": 4210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297192836693814e-06,
+      "loss": 0.7207,
+      "step": 4211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2926713395360555e-06,
+      "loss": 0.7549,
+      "step": 4212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288151078358819e-06,
+      "loss": 0.7634,
+      "step": 4213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283632054348961e-06,
+      "loss": 0.718,
+      "step": 4214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279114268693022e-06,
+      "loss": 0.7507,
+      "step": 4215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274597722577215e-06,
+      "loss": 0.7109,
+      "step": 4216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2700824171874296e-06,
+      "loss": 0.7327,
+      "step": 4217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265568353709218e-06,
+      "loss": 0.6247,
+      "step": 4218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261055533327828e-06,
+      "loss": 0.7205,
+      "step": 4219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256543957228164e-06,
+      "loss": 0.2914,
+      "step": 4220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25203362659481e-06,
+      "loss": 0.7041,
+      "step": 4221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24752454261202e-06,
+      "loss": 0.7,
+      "step": 4222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243016706463724e-06,
+      "loss": 0.7761,
+      "step": 4223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23851011933352e-06,
+      "loss": 0.7168,
+      "step": 4224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23400478240468e-06,
+      "loss": 0.7129,
+      "step": 4225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.229500696860155e-06,
+      "loss": 0.2642,
+      "step": 4226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224997863882557e-06,
+      "loss": 0.7275,
+      "step": 4227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220496284654173e-06,
+      "loss": 0.6628,
+      "step": 4228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215995960356965e-06,
+      "loss": 0.6954,
+      "step": 4229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211496892172554e-06,
+      "loss": 0.7363,
+      "step": 4230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206999081282238e-06,
+      "loss": 0.7222,
+      "step": 4231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202502528866993e-06,
+      "loss": 0.7778,
+      "step": 4232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198007236107454e-06,
+      "loss": 0.7068,
+      "step": 4233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193513204183928e-06,
+      "loss": 0.7764,
+      "step": 4234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18902043427639e-06,
+      "loss": 0.7217,
+      "step": 4235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184528927564487e-06,
+      "loss": 0.7144,
+      "step": 4236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180038685227531e-06,
+      "loss": 0.7615,
+      "step": 4237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175549708444503e-06,
+      "loss": 0.707,
+      "step": 4238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171061998394048e-06,
+      "loss": 0.7864,
+      "step": 4239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166575556254494e-06,
+      "loss": 0.6997,
+      "step": 4240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1620903832038135e-06,
+      "loss": 0.7429,
+      "step": 4241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157606480419661e-06,
+      "loss": 0.7466,
+      "step": 4242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153123849079351e-06,
+      "loss": 0.7017,
+      "step": 4243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148642490359868e-06,
+      "loss": 0.7058,
+      "step": 4244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144162405437857e-06,
+      "loss": 0.6766,
+      "step": 4245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139683595489639e-06,
+      "loss": 0.7133,
+      "step": 4246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135206061691189e-06,
+      "loss": 0.7266,
+      "step": 4247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1307298052181545e-06,
+      "loss": 0.7762,
+      "step": 4248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12625482724584e-06,
+      "loss": 0.7688,
+      "step": 4249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121781128949224e-06,
+      "loss": 0.6895,
+      "step": 4250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117308711502939e-06,
+      "loss": 0.7203,
+      "step": 4251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112837576081288e-06,
+      "loss": 0.7466,
+      "step": 4252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108367723858237e-06,
+      "loss": 0.2563,
+      "step": 4253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10389915600741e-06,
+      "loss": 0.7573,
+      "step": 4254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099431873702101e-06,
+      "loss": 0.2366,
+      "step": 4255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09496587811526e-06,
+      "loss": 0.7299,
+      "step": 4256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090501170419503e-06,
+      "loss": 0.2554,
+      "step": 4257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086037751787105e-06,
+      "loss": 0.2723,
+      "step": 4258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0815756233900036e-06,
+      "loss": 0.6721,
+      "step": 4259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077114786399804e-06,
+      "loss": 0.7076,
+      "step": 4260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072655241987763e-06,
+      "loss": 0.7111,
+      "step": 4261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0681969913248005e-06,
+      "loss": 0.6602,
+      "step": 4262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0637400355815e-06,
+      "loss": 0.7169,
+      "step": 4263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0592843759281065e-06,
+      "loss": 0.6792,
+      "step": 4264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054830013534506e-06,
+      "loss": 0.7557,
+      "step": 4265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050376949570275e-06,
+      "loss": 0.7383,
+      "step": 4266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045925185204625e-06,
+      "loss": 0.7014,
+      "step": 4267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041474721606438e-06,
+      "loss": 0.6877,
+      "step": 4268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037025559944249e-06,
+      "loss": 0.7208,
+      "step": 4269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032577701386252e-06,
+      "loss": 0.7261,
+      "step": 4270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0281311471003e-06,
+      "loss": 0.7476,
+      "step": 4271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023685898253905e-06,
+      "loss": 0.8018,
+      "step": 4272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019241956014229e-06,
+      "loss": 0.7053,
+      "step": 4273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014799321548105e-06,
+      "loss": 0.7493,
+      "step": 4274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0103579960220085e-06,
+      "loss": 0.3043,
+      "step": 4275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005917980602083e-06,
+      "loss": 0.7407,
+      "step": 4276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001479276454113e-06,
+      "loss": 0.6849,
+      "step": 4277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997041884743551e-06,
+      "loss": 0.6827,
+      "step": 4278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992605806635499e-06,
+      "loss": 0.7024,
+      "step": 4279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988171043294723e-06,
+      "loss": 0.7324,
+      "step": 4280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9837375958856334e-06,
+      "loss": 0.3031,
+      "step": 4281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.979305465572301e-06,
+      "loss": 0.7314,
+      "step": 4282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974874653518447e-06,
+      "loss": 0.7332,
+      "step": 4283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9704451608874485e-06,
+      "loss": 0.7241,
+      "step": 4284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966016988842335e-06,
+      "loss": 0.7388,
+      "step": 4285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961590138545789e-06,
+      "loss": 0.6718,
+      "step": 4286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957164611160152e-06,
+      "loss": 0.6802,
+      "step": 4287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952740407847416e-06,
+      "loss": 0.6565,
+      "step": 4288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948317529769214e-06,
+      "loss": 0.6559,
+      "step": 4289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.943895978086843e-06,
+      "loss": 0.7207,
+      "step": 4290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939475753961248e-06,
+      "loss": 0.7535,
+      "step": 4291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935056858553023e-06,
+      "loss": 0.6466,
+      "step": 4292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9306392930224235e-06,
+      "loss": 0.745,
+      "step": 4293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926223058529344e-06,
+      "loss": 0.7507,
+      "step": 4294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921808156233334e-06,
+      "loss": 0.7769,
+      "step": 4295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917394587293595e-06,
+      "loss": 0.7301,
+      "step": 4296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9129823528689725e-06,
+      "loss": 0.7189,
+      "step": 4297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908571454117969e-06,
+      "loss": 0.681,
+      "step": 4298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904161892198731e-06,
+      "loss": 0.7479,
+      "step": 4299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899753668269056e-06,
+      "loss": 0.707,
+      "step": 4300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895346783486391e-06,
+      "loss": 0.725,
+      "step": 4301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89094123900783e-06,
+      "loss": 0.6998,
+      "step": 4302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886537035990116e-06,
+      "loss": 0.7104,
+      "step": 4303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882134175589637e-06,
+      "loss": 0.7329,
+      "step": 4304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877732658962433e-06,
+      "loss": 0.6539,
+      "step": 4305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8733324872641856e-06,
+      "loss": 0.7612,
+      "step": 4306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868933661650232e-06,
+      "loss": 0.7274,
+      "step": 4307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864536183275548e-06,
+      "loss": 0.6816,
+      "step": 4308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860140053294759e-06,
+      "loss": 0.769,
+      "step": 4309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855745272862134e-06,
+      "loss": 0.6729,
+      "step": 4310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851351843131594e-06,
+      "loss": 0.7489,
+      "step": 4311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846959765256689e-06,
+      "loss": 0.7097,
+      "step": 4312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.842569040390637e-06,
+      "loss": 0.7581,
+      "step": 4313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838179669686285e-06,
+      "loss": 0.7737,
+      "step": 4314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833791654296129e-06,
+      "loss": 0.7031,
+      "step": 4315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829404995372311e-06,
+      "loss": 0.7327,
+      "step": 4316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825019694066612e-06,
+      "loss": 0.6404,
+      "step": 4317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820635751530461e-06,
+      "loss": 0.755,
+      "step": 4318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81625316891493e-06,
+      "loss": 0.7498,
+      "step": 4319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811871947370724e-06,
+      "loss": 0.6632,
+      "step": 4320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807492088048211e-06,
+      "loss": 0.7098,
+      "step": 4321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803113592097383e-06,
+      "loss": 0.7028,
+      "step": 4322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798736460667888e-06,
+      "loss": 0.7595,
+      "step": 4323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794360694908996e-06,
+      "loss": 0.6753,
+      "step": 4324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789986295969637e-06,
+      "loss": 0.3075,
+      "step": 4325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785613264998373e-06,
+      "loss": 0.6976,
+      "step": 4326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781241603143415e-06,
+      "loss": 0.708,
+      "step": 4327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776871311552607e-06,
+      "loss": 0.3072,
+      "step": 4328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772502391373433e-06,
+      "loss": 0.7239,
+      "step": 4329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768134843753022e-06,
+      "loss": 0.7637,
+      "step": 4330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763768669838137e-06,
+      "loss": 0.7576,
+      "step": 4331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7594038707751855e-06,
+      "loss": 0.6775,
+      "step": 4332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7550404477102076e-06,
+      "loss": 0.74,
+      "step": 4333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750678401788893e-06,
+      "loss": 0.6537,
+      "step": 4334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746317734156564e-06,
+      "loss": 0.703,
+      "step": 4335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74195844595817e-06,
+      "loss": 0.7361,
+      "step": 4336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737600538338315e-06,
+      "loss": 0.6772,
+      "step": 4337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733244012441233e-06,
+      "loss": 0.7213,
+      "step": 4338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7288888694107906e-06,
+      "loss": 0.7144,
+      "step": 4339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724535110390505e-06,
+      "loss": 0.7316,
+      "step": 4340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720182736523517e-06,
+      "loss": 0.7561,
+      "step": 4341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7158317489526116e-06,
+      "loss": 0.7698,
+      "step": 4342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711482148820204e-06,
+      "loss": 0.8018,
+      "step": 4343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707133937268348e-06,
+      "loss": 0.7106,
+      "step": 4344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702787115438732e-06,
+      "loss": 0.6882,
+      "step": 4345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69844168447268e-06,
+      "loss": 0.7096,
+      "step": 4346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694097645511152e-06,
+      "loss": 0.7583,
+      "step": 4347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68975499969474e-06,
+      "loss": 0.7515,
+      "step": 4348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.685413748163673e-06,
+      "loss": 0.7761,
+      "step": 4349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681073892057811e-06,
+      "loss": 0.7415,
+      "step": 4350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676735432516649e-06,
+      "loss": 0.6936,
+      "step": 4351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672398370679317e-06,
+      "loss": 0.7151,
+      "step": 4352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668062707684571e-06,
+      "loss": 0.7285,
+      "step": 4353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663728444670813e-06,
+      "loss": 0.7094,
+      "step": 4354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659395582776066e-06,
+      "loss": 0.268,
+      "step": 4355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655064123137988e-06,
+      "loss": 0.7345,
+      "step": 4356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6507340668938695e-06,
+      "loss": 0.6936,
+      "step": 4357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646405415180637e-06,
+      "loss": 0.759,
+      "step": 4358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6420781691348314e-06,
+      "loss": 0.6758,
+      "step": 4359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637752329892648e-06,
+      "loss": 0.7549,
+      "step": 4360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6334278985899e-06,
+      "loss": 0.7087,
+      "step": 4361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629104876362029e-06,
+      "loss": 0.7324,
+      "step": 4362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624783264344113e-06,
+      "loss": 0.7552,
+      "step": 4363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620463063670854e-06,
+      "loss": 0.738,
+      "step": 4364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616144275476588e-06,
+      "loss": 0.7548,
+      "step": 4365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611826900895278e-06,
+      "loss": 0.7422,
+      "step": 4366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607510941060511e-06,
+      "loss": 0.7585,
+      "step": 4367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603196397105518e-06,
+      "loss": 0.7069,
+      "step": 4368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598883270163143e-06,
+      "loss": 0.7347,
+      "step": 4369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594571561365866e-06,
+      "loss": 0.7661,
+      "step": 4370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590261271845784e-06,
+      "loss": 0.6907,
+      "step": 4371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585952402734632e-06,
+      "loss": 0.7708,
+      "step": 4372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581644955163768e-06,
+      "loss": 0.7502,
+      "step": 4373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5773389302641826e-06,
+      "loss": 0.6978,
+      "step": 4374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573034329166484e-06,
+      "loss": 0.6816,
+      "step": 4375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568731153000913e-06,
+      "loss": 0.6379,
+      "step": 4376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56442940289733e-06,
+      "loss": 0.7576,
+      "step": 4377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560129079985228e-06,
+      "loss": 0.7516,
+      "step": 4378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5558301853937214e-06,
+      "loss": 0.7177,
+      "step": 4379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5515327202515456e-06,
+      "loss": 0.7017,
+      "step": 4380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547236685687073e-06,
+      "loss": 0.7451,
+      "step": 4381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5429420828282925e-06,
+      "loss": 0.7104,
+      "step": 4382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538648912802809e-06,
+      "loss": 0.7432,
+      "step": 4383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534357176737865e-06,
+      "loss": 0.7244,
+      "step": 4384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530066875760318e-06,
+      "loss": 0.2659,
+      "step": 4385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525778010996652e-06,
+      "loss": 0.6935,
+      "step": 4386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.521490583572977e-06,
+      "loss": 0.7454,
+      "step": 4387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517204594615021e-06,
+      "loss": 0.6807,
+      "step": 4388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5129200452481334e-06,
+      "loss": 0.7158,
+      "step": 4389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508636936597289e-06,
+      "loss": 0.7014,
+      "step": 4390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504355269787083e-06,
+      "loss": 0.7356,
+      "step": 4391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500075045941731e-06,
+      "loss": 0.2786,
+      "step": 4392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495796266185071e-06,
+      "loss": 0.7312,
+      "step": 4393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491518931640561e-06,
+      "loss": 0.6796,
+      "step": 4394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487243043431282e-06,
+      "loss": 0.7098,
+      "step": 4395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48296860267993e-06,
+      "loss": 0.7028,
+      "step": 4396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478695610508828e-06,
+      "loss": 0.7451,
+      "step": 4397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474424068039912e-06,
+      "loss": 0.7393,
+      "step": 4398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470153976394741e-06,
+      "loss": 0.718,
+      "step": 4399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46588533669449e-06,
+      "loss": 0.3066,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4616181500599595e-06,
+      "loss": 0.708,
+      "step": 4401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457352417611565e-06,
+      "loss": 0.2886,
+      "step": 4402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453088140469334e-06,
+      "loss": 0.7219,
+      "step": 4403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4488253197529205e-06,
+      "loss": 0.2656,
+      "step": 4404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444563956581591e-06,
+      "loss": 0.7473,
+      "step": 4405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440304052074233e-06,
+      "loss": 0.7722,
+      "step": 4406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436045607349347e-06,
+      "loss": 0.7391,
+      "step": 4407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431788623525054e-06,
+      "loss": 0.7092,
+      "step": 4408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427533101719088e-06,
+      "loss": 0.2893,
+      "step": 4409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423279043048802e-06,
+      "loss": 0.7566,
+      "step": 4410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419026448631162e-06,
+      "loss": 0.6677,
+      "step": 4411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414775319582752e-06,
+      "loss": 0.7247,
+      "step": 4412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410525657019772e-06,
+      "loss": 0.7095,
+      "step": 4413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40627746205803e-06,
+      "loss": 0.6361,
+      "step": 4414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402030735812961e-06,
+      "loss": 0.7483,
+      "step": 4415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397785479399605e-06,
+      "loss": 0.7173,
+      "step": 4416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.393541693932621e-06,
+      "loss": 0.75,
+      "step": 4417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389299380526273e-06,
+      "loss": 0.6646,
+      "step": 4418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385058540294447e-06,
+      "loss": 0.79,
+      "step": 4419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380819174350637e-06,
+      "loss": 0.7411,
+      "step": 4420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37658128380796e-06,
+      "loss": 0.6507,
+      "step": 4421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372344869779136e-06,
+      "loss": 0.707,
+      "step": 4422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368109933376498e-06,
+      "loss": 0.6964,
+      "step": 4423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363876475711992e-06,
+      "loss": 0.6621,
+      "step": 4424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3596444978971795e-06,
+      "loss": 0.6901,
+      "step": 4425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355414001043228e-06,
+      "loss": 0.6917,
+      "step": 4426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35118498626092e-06,
+      "loss": 0.7539,
+      "step": 4427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346957454660643e-06,
+      "loss": 0.7292,
+      "step": 4428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342731407352411e-06,
+      "loss": 0.6919,
+      "step": 4429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338506845445824e-06,
+      "loss": 0.6987,
+      "step": 4430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334283770050112e-06,
+      "loss": 0.7275,
+      "step": 4431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330062182274104e-06,
+      "loss": 0.2915,
+      "step": 4432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325842083226244e-06,
+      "loss": 0.7212,
+      "step": 4433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321623474014579e-06,
+      "loss": 0.7108,
+      "step": 4434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3174063557467746e-06,
+      "loss": 0.6923,
+      "step": 4435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313190729530098e-06,
+      "loss": 0.273,
+      "step": 4436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308976596471424e-06,
+      "loss": 0.7266,
+      "step": 4437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304763957677237e-06,
+      "loss": 0.72,
+      "step": 4438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30055281425363e-06,
+      "loss": 0.6855,
+      "step": 4439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2963431673063035e-06,
+      "loss": 0.7722,
+      "step": 4440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292135017940562e-06,
+      "loss": 0.6969,
+      "step": 4441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28792836726132e-06,
+      "loss": 0.7003,
+      "step": 4442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283723216373098e-06,
+      "loss": 0.7502,
+      "step": 4443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279519566380021e-06,
+      "loss": 0.7605,
+      "step": 4444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275317418385821e-06,
+      "loss": 0.6908,
+      "step": 4445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2711167734938365e-06,
+      "loss": 0.2802,
+      "step": 4446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266917632807006e-06,
+      "loss": 0.6846,
+      "step": 4447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262719997427887e-06,
+      "loss": 0.7068,
+      "step": 4448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258523868458626e-06,
+      "loss": 0.7037,
+      "step": 4449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254329247000981e-06,
+      "loss": 0.7322,
+      "step": 4450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250136134156314e-06,
+      "loss": 0.27,
+      "step": 4451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245944531025591e-06,
+      "loss": 0.7139,
+      "step": 4452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241754438709379e-06,
+      "loss": 0.6838,
+      "step": 4453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237565858307853e-06,
+      "loss": 0.6863,
+      "step": 4454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233378790920786e-06,
+      "loss": 0.7754,
+      "step": 4455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229193237647555e-06,
+      "loss": 0.2785,
+      "step": 4456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225009199587145e-06,
+      "loss": 0.6849,
+      "step": 4457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220826677838134e-06,
+      "loss": 0.7407,
+      "step": 4458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21664567349871e-06,
+      "loss": 0.6826,
+      "step": 4459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212466187666655e-06,
+      "loss": 0.6517,
+      "step": 4460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208288221439357e-06,
+      "loss": 0.7231,
+      "step": 4461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20411177591381e-06,
+      "loss": 0.733,
+      "step": 4462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199936852186598e-06,
+      "loss": 0.6963,
+      "step": 4463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195763451353913e-06,
+      "loss": 0.7135,
+      "step": 4464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191591574511547e-06,
+      "loss": 0.7369,
+      "step": 4465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187421222754882e-06,
+      "loss": 0.7063,
+      "step": 4466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183252397178907e-06,
+      "loss": 0.7028,
+      "step": 4467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1790850988782185e-06,
+      "loss": 0.7266,
+      "step": 4468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174919328947e-06,
+      "loss": 0.6932,
+      "step": 4469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170755088479038e-06,
+      "loss": 0.7095,
+      "step": 4470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1665923785677155e-06,
+      "loss": 0.6948,
+      "step": 4471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162431200306017e-06,
+      "loss": 0.7532,
+      "step": 4472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1582715547865205e-06,
+      "loss": 0.7286,
+      "step": 4473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154113443101406e-06,
+      "loss": 0.761,
+      "step": 4474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149956866342445e-06,
+      "loss": 0.6447,
+      "step": 4475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14580182560102e-06,
+      "loss": 0.7128,
+      "step": 4476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141648321968089e-06,
+      "loss": 0.7329,
+      "step": 4477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137496356534222e-06,
+      "loss": 0.7593,
+      "step": 4478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1333459303895795e-06,
+      "loss": 0.673,
+      "step": 4479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129197044623919e-06,
+      "loss": 0.7162,
+      "step": 4480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1250497003265886e-06,
+      "loss": 0.7649,
+      "step": 4481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120903898586547e-06,
+      "loss": 0.7214,
+      "step": 4482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1167596404923305e-06,
+      "loss": 0.7023,
+      "step": 4483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112616927132078e-06,
+      "loss": 0.7148,
+      "step": 4484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108475759593522e-06,
+      "loss": 0.7102,
+      "step": 4485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104336138963988e-06,
+      "loss": 0.6119,
+      "step": 4486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100198066330398e-06,
+      "loss": 0.7401,
+      "step": 4487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096061542779263e-06,
+      "loss": 0.6987,
+      "step": 4488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091926569396692e-06,
+      "loss": 0.7668,
+      "step": 4489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087793147268383e-06,
+      "loss": 0.7146,
+      "step": 4490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083661277479631e-06,
+      "loss": 0.7865,
+      "step": 4491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07953096111532e-06,
+      "loss": 0.653,
+      "step": 4492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075402199259928e-06,
+      "loss": 0.7291,
+      "step": 4493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07127499299752e-06,
+      "loss": 0.7036,
+      "step": 4494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067149343411762e-06,
+      "loss": 0.7424,
+      "step": 4495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063025251585904e-06,
+      "loss": 0.2986,
+      "step": 4496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058902718602789e-06,
+      "loss": 0.7092,
+      "step": 4497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05478174554485e-06,
+      "loss": 0.7705,
+      "step": 4498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050662333494111e-06,
+      "loss": 0.6838,
+      "step": 4499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046544483532186e-06,
+      "loss": 0.7681,
+      "step": 4500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042428196740279e-06,
+      "loss": 0.7416,
+      "step": 4501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038313474199184e-06,
+      "loss": 0.6833,
+      "step": 4502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034200316989284e-06,
+      "loss": 0.733,
+      "step": 4503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03008872619055e-06,
+      "loss": 0.6956,
+      "step": 4504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0259787028825415e-06,
+      "loss": 0.7329,
+      "step": 4505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021870248144409e-06,
+      "loss": 0.7079,
+      "step": 4506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01776336305489e-06,
+      "loss": 0.7739,
+      "step": 4507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013658048692305e-06,
+      "loss": 0.7141,
+      "step": 4508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009554306134575e-06,
+      "loss": 0.717,
+      "step": 4509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005452136459193e-06,
+      "loss": 0.731,
+      "step": 4510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001351540743249e-06,
+      "loss": 0.7983,
+      "step": 4511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997252520063419e-06,
+      "loss": 0.761,
+      "step": 4512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993155075495955e-06,
+      "loss": 0.7339,
+      "step": 4513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989059208116705e-06,
+      "loss": 0.6892,
+      "step": 4514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984964919001107e-06,
+      "loss": 0.7338,
+      "step": 4515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9808722092241755e-06,
+      "loss": 0.7161,
+      "step": 4516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976781079860515e-06,
+      "loss": 0.7249,
+      "step": 4517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972691531984311e-06,
+      "loss": 0.7206,
+      "step": 4518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9686035666693365e-06,
+      "loss": 0.7129,
+      "step": 4519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964517184988951e-06,
+      "loss": 0.7274,
+      "step": 4520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960432388016093e-06,
+      "loss": 0.7605,
+      "step": 4521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9563491768232874e-06,
+      "loss": 0.7141,
+      "step": 4522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.952267552482652e-06,
+      "loss": 0.7417,
+      "step": 4523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94818751606587e-06,
+      "loss": 0.704,
+      "step": 4524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944109068644217e-06,
+      "loss": 0.73,
+      "step": 4525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940032211288556e-06,
+      "loss": 0.7231,
+      "step": 4526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935956945069324e-06,
+      "loss": 0.7096,
+      "step": 4527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9318832710565415e-06,
+      "loss": 0.7727,
+      "step": 4528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927811190319821e-06,
+      "loss": 0.7202,
+      "step": 4529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923740703928345e-06,
+      "loss": 0.8062,
+      "step": 4530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9196718129508815e-06,
+      "loss": 0.7393,
+      "step": 4531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915604518455779e-06,
+      "loss": 0.7264,
+      "step": 4532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911538821510968e-06,
+      "loss": 0.6547,
+      "step": 4533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90747472318396e-06,
+      "loss": 0.738,
+      "step": 4534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903412224541844e-06,
+      "loss": 0.7788,
+      "step": 4535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8993513266512905e-06,
+      "loss": 0.2986,
+      "step": 4536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8952920305785505e-06,
+      "loss": 0.7311,
+      "step": 4537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8912343373894535e-06,
+      "loss": 0.7812,
+      "step": 4538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887178248149408e-06,
+      "loss": 0.7366,
+      "step": 4539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883123763923403e-06,
+      "loss": 0.691,
+      "step": 4540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879070885776e-06,
+      "loss": 0.7114,
+      "step": 4541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875019614771352e-06,
+      "loss": 0.7202,
+      "step": 4542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8709699519731755e-06,
+      "loss": 0.7556,
+      "step": 4543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866921898444774e-06,
+      "loss": 0.7544,
+      "step": 4544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862875455249023e-06,
+      "loss": 0.2778,
+      "step": 4545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8588306234483795e-06,
+      "loss": 0.6997,
+      "step": 4546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854787404104874e-06,
+      "loss": 0.72,
+      "step": 4547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850745798280115e-06,
+      "loss": 0.6676,
+      "step": 4548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846705807035288e-06,
+      "loss": 0.739,
+      "step": 4549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842667431431154e-06,
+      "loss": 0.6947,
+      "step": 4550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83863067252805e-06,
+      "loss": 0.7316,
+      "step": 4551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8345955313858875e-06,
+      "loss": 0.7126,
+      "step": 4552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830562009064153e-06,
+      "loss": 0.2562,
+      "step": 4553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826530106621913e-06,
+      "loss": 0.739,
+      "step": 4554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822499825117796e-06,
+      "loss": 0.6866,
+      "step": 4555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818471165610024e-06,
+      "loss": 0.7296,
+      "step": 4556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8144441291563773e-06,
+      "loss": 0.6855,
+      "step": 4557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810418716814216e-06,
+      "loss": 0.7429,
+      "step": 4558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806394929640478e-06,
+      "loss": 0.7805,
+      "step": 4559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80237276869166e-06,
+      "loss": 0.7808,
+      "step": 4560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798352235023843e-06,
+      "loss": 0.7102,
+      "step": 4561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7943333296926865e-06,
+      "loss": 0.7695,
+      "step": 4562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79031605375341e-06,
+      "loss": 0.7886,
+      "step": 4563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.786300408260811e-06,
+      "loss": 0.7181,
+      "step": 4564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782286394269259e-06,
+      "loss": 0.6918,
+      "step": 4565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778274012832693e-06,
+      "loss": 0.7356,
+      "step": 4566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7742632650046253e-06,
+      "loss": 0.7592,
+      "step": 4567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7702541518381374e-06,
+      "loss": 0.6558,
+      "step": 4568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7662466743858804e-06,
+      "loss": 0.677,
+      "step": 4569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762240833700086e-06,
+      "loss": 0.7166,
+      "step": 4570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7582366308325457e-06,
+      "loss": 0.7097,
+      "step": 4571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7542340668346198e-06,
+      "loss": 0.7212,
+      "step": 4572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7502331427572435e-06,
+      "loss": 0.7893,
+      "step": 4573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7462338596509206e-06,
+      "loss": 0.7875,
+      "step": 4574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7422362185657213e-06,
+      "loss": 0.7126,
+      "step": 4575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7382402205512912e-06,
+      "loss": 0.7144,
+      "step": 4576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7342458666568392e-06,
+      "loss": 0.7184,
+      "step": 4577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730253157931143e-06,
+      "loss": 0.7463,
+      "step": 4578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726262095422548e-06,
+      "loss": 0.7085,
+      "step": 4579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722272680178968e-06,
+      "loss": 0.7217,
+      "step": 4580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718284913247886e-06,
+      "loss": 0.651,
+      "step": 4581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7142987956763487e-06,
+      "loss": 0.7205,
+      "step": 4582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7103143285109732e-06,
+      "loss": 0.7402,
+      "step": 4583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7063315127979414e-06,
+      "loss": 0.7128,
+      "step": 4584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7023503495830015e-06,
+      "loss": 0.6958,
+      "step": 4585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698370839911469e-06,
+      "loss": 0.2877,
+      "step": 4586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6943929848282234e-06,
+      "loss": 0.7472,
+      "step": 4587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690416785377713e-06,
+      "loss": 0.7183,
+      "step": 4588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6864422426039426e-06,
+      "loss": 0.7263,
+      "step": 4589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.682469357550499e-06,
+      "loss": 0.7287,
+      "step": 4590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6784981312605174e-06,
+      "loss": 0.7271,
+      "step": 4591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.674528564776704e-06,
+      "loss": 0.7495,
+      "step": 4592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67056065914133e-06,
+      "loss": 0.7018,
+      "step": 4593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666594415396231e-06,
+      "loss": 0.6906,
+      "step": 4594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662629834582795e-06,
+      "loss": 0.76,
+      "step": 4595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658666917741993e-06,
+      "loss": 0.7043,
+      "step": 4596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6547056659143454e-06,
+      "loss": 0.7351,
+      "step": 4597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6507460801399386e-06,
+      "loss": 0.7437,
+      "step": 4598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6467881614584232e-06,
+      "loss": 0.7115,
+      "step": 4599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6428319109090095e-06,
+      "loss": 0.7817,
+      "step": 4600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6388773295304703e-06,
+      "loss": 0.7505,
+      "step": 4601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6349244183611397e-06,
+      "loss": 0.7468,
+      "step": 4602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630973178438919e-06,
+      "loss": 0.7808,
+      "step": 4603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627023610801265e-06,
+      "loss": 0.6786,
+      "step": 4604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623075716485195e-06,
+      "loss": 0.762,
+      "step": 4605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619129496527293e-06,
+      "loss": 0.7732,
+      "step": 4606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6151849519636904e-06,
+      "loss": 0.3033,
+      "step": 4607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6112420838300886e-06,
+      "loss": 0.718,
+      "step": 4608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6073008931617536e-06,
+      "loss": 0.6473,
+      "step": 4609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603361380993502e-06,
+      "loss": 0.7301,
+      "step": 4610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5994235483597117e-06,
+      "loss": 0.7488,
+      "step": 4611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5954873962943217e-06,
+      "loss": 0.7107,
+      "step": 4612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591552925830827e-06,
+      "loss": 0.7449,
+      "step": 4613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5876201380022825e-06,
+      "loss": 0.6913,
+      "step": 4614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5836890338413024e-06,
+      "loss": 0.7314,
+      "step": 4615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.579759614380053e-06,
+      "loss": 0.7471,
+      "step": 4616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57583188065027e-06,
+      "loss": 0.6692,
+      "step": 4617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57190583368324e-06,
+      "loss": 0.7551,
+      "step": 4618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5679814745097994e-06,
+      "loss": 0.7549,
+      "step": 4619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5640588041603518e-06,
+      "loss": 0.74,
+      "step": 4620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560137823664853e-06,
+      "loss": 0.74,
+      "step": 4621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5562185340528134e-06,
+      "loss": 0.7035,
+      "step": 4622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.552300936353308e-06,
+      "loss": 0.6388,
+      "step": 4623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5483850315949574e-06,
+      "loss": 0.6831,
+      "step": 4624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.544470820805943e-06,
+      "loss": 0.6987,
+      "step": 4625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540558305014e-06,
+      "loss": 0.7325,
+      "step": 4626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5366474852464196e-06,
+      "loss": 0.7073,
+      "step": 4627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5327383625300437e-06,
+      "loss": 0.6849,
+      "step": 4628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5288309378912755e-06,
+      "loss": 0.6436,
+      "step": 4629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5249252123560663e-06,
+      "loss": 0.6662,
+      "step": 4630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5210211869499244e-06,
+      "loss": 0.6705,
+      "step": 4631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5171188626979105e-06,
+      "loss": 0.7986,
+      "step": 4632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5132182406246383e-06,
+      "loss": 0.7396,
+      "step": 4633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5093193217542766e-06,
+      "loss": 0.7141,
+      "step": 4634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5054221071105456e-06,
+      "loss": 0.2639,
+      "step": 4635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5015265977167145e-06,
+      "loss": 0.7134,
+      "step": 4636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4976327945956144e-06,
+      "loss": 0.7439,
+      "step": 4637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4937406987696198e-06,
+      "loss": 0.7595,
+      "step": 4638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4898503112606584e-06,
+      "loss": 0.6926,
+      "step": 4639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485961633090212e-06,
+      "loss": 0.7717,
+      "step": 4640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4820746652793146e-06,
+      "loss": 0.7322,
+      "step": 4641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478189408848539e-06,
+      "loss": 0.7179,
+      "step": 4642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4743058648180263e-06,
+      "loss": 0.728,
+      "step": 4643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.470424034207459e-06,
+      "loss": 0.6995,
+      "step": 4644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4665439180360692e-06,
+      "loss": 0.2764,
+      "step": 4645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4626655173226396e-06,
+      "loss": 0.7615,
+      "step": 4646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4587888330855057e-06,
+      "loss": 0.7515,
+      "step": 4647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4549138663425476e-06,
+      "loss": 0.6809,
+      "step": 4648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4510406181111932e-06,
+      "loss": 0.6946,
+      "step": 4649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4471690894084298e-06,
+      "loss": 0.6886,
+      "step": 4650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4432992812507827e-06,
+      "loss": 0.774,
+      "step": 4651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439431194654329e-06,
+      "loss": 0.7283,
+      "step": 4652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4355648306346956e-06,
+      "loss": 0.7131,
+      "step": 4653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.431700190207049e-06,
+      "loss": 0.7205,
+      "step": 4654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4278372743861097e-06,
+      "loss": 0.7596,
+      "step": 4655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4239760841861514e-06,
+      "loss": 0.7108,
+      "step": 4656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4201166206209847e-06,
+      "loss": 0.7058,
+      "step": 4657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.41625888470397e-06,
+      "loss": 0.6823,
+      "step": 4658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4124028774480134e-06,
+      "loss": 0.7318,
+      "step": 4659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4085485998655697e-06,
+      "loss": 0.2723,
+      "step": 4660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404696052968638e-06,
+      "loss": 0.7483,
+      "step": 4661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4008452377687605e-06,
+      "loss": 0.725,
+      "step": 4662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396996155277027e-06,
+      "loss": 0.7283,
+      "step": 4663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3931488065040773e-06,
+      "loss": 0.7262,
+      "step": 4664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389303192460092e-06,
+      "loss": 0.7983,
+      "step": 4665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3854593141547877e-06,
+      "loss": 0.7346,
+      "step": 4666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3816171725974356e-06,
+      "loss": 0.697,
+      "step": 4667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.377776768796851e-06,
+      "loss": 0.7236,
+      "step": 4668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373938103761385e-06,
+      "loss": 0.6591,
+      "step": 4669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3701011784989435e-06,
+      "loss": 0.7485,
+      "step": 4670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3662659940169673e-06,
+      "loss": 0.7373,
+      "step": 4671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362432551322441e-06,
+      "loss": 0.7153,
+      "step": 4672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.358600851421895e-06,
+      "loss": 0.7517,
+      "step": 4673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3547708953213974e-06,
+      "loss": 0.733,
+      "step": 4674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3509426840265644e-06,
+      "loss": 0.7336,
+      "step": 4675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3471162185425486e-06,
+      "loss": 0.2803,
+      "step": 4676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3432914998740474e-06,
+      "loss": 0.6523,
+      "step": 4677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3394685290252992e-06,
+      "loss": 0.7085,
+      "step": 4678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335647307000082e-06,
+      "loss": 0.7253,
+      "step": 4679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3318278348017154e-06,
+      "loss": 0.7283,
+      "step": 4680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328010113433061e-06,
+      "loss": 0.7488,
+      "step": 4681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3241941438965186e-06,
+      "loss": 0.7544,
+      "step": 4682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3203799271940253e-06,
+      "loss": 0.7057,
+      "step": 4683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3165674643270685e-06,
+      "loss": 0.6964,
+      "step": 4684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3127567562966658e-06,
+      "loss": 0.2692,
+      "step": 4685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308947804103374e-06,
+      "loss": 0.7205,
+      "step": 4686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.305140608747294e-06,
+      "loss": 0.7622,
+      "step": 4687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3013351712280605e-06,
+      "loss": 0.6846,
+      "step": 4688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2975314925448498e-06,
+      "loss": 0.698,
+      "step": 4689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2937295736963758e-06,
+      "loss": 0.7903,
+      "step": 4690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2899294156808893e-06,
+      "loss": 0.6614,
+      "step": 4691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2861310194961793e-06,
+      "loss": 0.6875,
+      "step": 4692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2823343861395727e-06,
+      "loss": 0.724,
+      "step": 4693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2785395166079327e-06,
+      "loss": 0.7266,
+      "step": 4694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27474641189766e-06,
+      "loss": 0.7217,
+      "step": 4695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2709550730046893e-06,
+      "loss": 0.7148,
+      "step": 4696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2671655009244996e-06,
+      "loss": 0.6714,
+      "step": 4697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263377696652098e-06,
+      "loss": 0.2755,
+      "step": 4698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259591661182029e-06,
+      "loss": 0.7079,
+      "step": 4699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255807395508378e-06,
+      "loss": 0.7052,
+      "step": 4700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2520249006247538e-06,
+      "loss": 0.6748,
+      "step": 4701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.248244177524308e-06,
+      "loss": 0.7228,
+      "step": 4702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2444652271997344e-06,
+      "loss": 0.7437,
+      "step": 4703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2406880506432493e-06,
+      "loss": 0.7712,
+      "step": 4704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2369126488466073e-06,
+      "loss": 0.6984,
+      "step": 4705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2331390228010982e-06,
+      "loss": 0.7803,
+      "step": 4706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2293671734975452e-06,
+      "loss": 0.6863,
+      "step": 4707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2255971019263043e-06,
+      "loss": 0.7178,
+      "step": 4708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2218288090772642e-06,
+      "loss": 0.7377,
+      "step": 4709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2180622959398443e-06,
+      "loss": 0.7412,
+      "step": 4710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214297563503006e-06,
+      "loss": 0.7294,
+      "step": 4711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2105346127552372e-06,
+      "loss": 0.7,
+      "step": 4712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2067734446845523e-06,
+      "loss": 0.7275,
+      "step": 4713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2030140602785043e-06,
+      "loss": 0.6166,
+      "step": 4714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1992564605241783e-06,
+      "loss": 0.6667,
+      "step": 4715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.195500646408184e-06,
+      "loss": 0.2722,
+      "step": 4716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1917466189166758e-06,
+      "loss": 0.7456,
+      "step": 4717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.187994379035326e-06,
+      "loss": 0.7269,
+      "step": 4718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.184243927749343e-06,
+      "loss": 0.7217,
+      "step": 4719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.180495266043463e-06,
+      "loss": 0.7272,
+      "step": 4720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1767483949019563e-06,
+      "loss": 0.7449,
+      "step": 4721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1730033153086184e-06,
+      "loss": 0.7175,
+      "step": 4722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1692600282467777e-06,
+      "loss": 0.7015,
+      "step": 4723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1655185346992923e-06,
+      "loss": 0.6713,
+      "step": 4724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161778835648547e-06,
+      "loss": 0.6791,
+      "step": 4725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.158040932076455e-06,
+      "loss": 0.6887,
+      "step": 4726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1543048249644615e-06,
+      "loss": 0.7183,
+      "step": 4727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1505705152935374e-06,
+      "loss": 0.6406,
+      "step": 4728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1468380040441824e-06,
+      "loss": 0.73,
+      "step": 4729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1431072921964212e-06,
+      "loss": 0.7473,
+      "step": 4730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139378380729815e-06,
+      "loss": 0.6511,
+      "step": 4731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135651270623442e-06,
+      "loss": 0.7195,
+      "step": 4732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131925962855913e-06,
+      "loss": 0.6952,
+      "step": 4733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128202458405362e-06,
+      "loss": 0.7539,
+      "step": 4734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.124480758249453e-06,
+      "loss": 0.6918,
+      "step": 4735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1207608633653743e-06,
+      "loss": 0.7118,
+      "step": 4736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1170427747298403e-06,
+      "loss": 0.7183,
+      "step": 4737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113326493319093e-06,
+      "loss": 0.6742,
+      "step": 4738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1096120201088973e-06,
+      "loss": 0.6919,
+      "step": 4739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1058993560745444e-06,
+      "loss": 0.6868,
+      "step": 4740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1021885021908515e-06,
+      "loss": 0.7344,
+      "step": 4741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0984794594321577e-06,
+      "loss": 0.7067,
+      "step": 4742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.09477222877233e-06,
+      "loss": 0.7334,
+      "step": 4743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091066811184755e-06,
+      "loss": 0.7245,
+      "step": 4744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087363207642352e-06,
+      "loss": 0.7297,
+      "step": 4745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083661419117555e-06,
+      "loss": 0.7053,
+      "step": 4746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.079961446582328e-06,
+      "loss": 0.7017,
+      "step": 4747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0762632910081493e-06,
+      "loss": 0.7268,
+      "step": 4748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0725669533660274e-06,
+      "loss": 0.7019,
+      "step": 4749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.06887243462649e-06,
+      "loss": 0.7141,
+      "step": 4750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0651797357595935e-06,
+      "loss": 0.7817,
+      "step": 4751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.061488857734911e-06,
+      "loss": 0.7731,
+      "step": 4752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0577998015215383e-06,
+      "loss": 0.6804,
+      "step": 4753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0541125680880914e-06,
+      "loss": 0.6885,
+      "step": 4754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.05042715840271e-06,
+      "loss": 0.731,
+      "step": 4755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0467435734330553e-06,
+      "loss": 0.2598,
+      "step": 4756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.043061814146304e-06,
+      "loss": 0.7185,
+      "step": 4757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0393818815091645e-06,
+      "loss": 0.6864,
+      "step": 4758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0357037764878594e-06,
+      "loss": 0.7578,
+      "step": 4759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0320275000481248e-06,
+      "loss": 0.679,
+      "step": 4760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.028353053155225e-06,
+      "loss": 0.6812,
+      "step": 4761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0246804367739414e-06,
+      "loss": 0.2775,
+      "step": 4762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0210096518685738e-06,
+      "loss": 0.698,
+      "step": 4763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0173406994029474e-06,
+      "loss": 0.6699,
+      "step": 4764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0136735803403983e-06,
+      "loss": 0.7195,
+      "step": 4765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010008295643785e-06,
+      "loss": 0.714,
+      "step": 4766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006344846275482e-06,
+      "loss": 0.7379,
+      "step": 4767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0026832331973855e-06,
+      "loss": 0.7468,
+      "step": 4768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9990234573709054e-06,
+      "loss": 0.6985,
+      "step": 4769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995365519756973e-06,
+      "loss": 0.709,
+      "step": 4770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9917094213160346e-06,
+      "loss": 0.7856,
+      "step": 4771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9880551630080543e-06,
+      "loss": 0.6746,
+      "step": 4772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9844027457925127e-06,
+      "loss": 0.7162,
+      "step": 4773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.980752170628408e-06,
+      "loss": 0.7325,
+      "step": 4774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9771034384742536e-06,
+      "loss": 0.707,
+      "step": 4775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9734565502880807e-06,
+      "loss": 0.7174,
+      "step": 4776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.96981150702743e-06,
+      "loss": 0.7231,
+      "step": 4777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.96616830964937e-06,
+      "loss": 0.7332,
+      "step": 4778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9625269591104744e-06,
+      "loss": 0.7231,
+      "step": 4779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9588874563668347e-06,
+      "loss": 0.7435,
+      "step": 4780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9552498023740583e-06,
+      "loss": 0.286,
+      "step": 4781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.951613998087266e-06,
+      "loss": 0.7234,
+      "step": 4782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947980044461093e-06,
+      "loss": 0.6553,
+      "step": 4783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944347942449689e-06,
+      "loss": 0.7174,
+      "step": 4784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940717693006717e-06,
+      "loss": 0.6689,
+      "step": 4785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.937089297085355e-06,
+      "loss": 0.7275,
+      "step": 4786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933462755638293e-06,
+      "loss": 0.691,
+      "step": 4787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9298380696177354e-06,
+      "loss": 0.7109,
+      "step": 4788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926215239975396e-06,
+      "loss": 0.7125,
+      "step": 4789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9225942676625062e-06,
+      "loss": 0.6318,
+      "step": 4790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9189751536298028e-06,
+      "loss": 0.708,
+      "step": 4791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9153578988275453e-06,
+      "loss": 0.6455,
+      "step": 4792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9117425042054957e-06,
+      "loss": 0.7183,
+      "step": 4793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.908128970712931e-06,
+      "loss": 0.7417,
+      "step": 4794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9045172992986427e-06,
+      "loss": 0.7712,
+      "step": 4795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9009074909109237e-06,
+      "loss": 0.6771,
+      "step": 4796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897299546497582e-06,
+      "loss": 0.7156,
+      "step": 4797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893693467005947e-06,
+      "loss": 0.6401,
+      "step": 4798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890089253382844e-06,
+      "loss": 0.7463,
+      "step": 4799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.886486906574616e-06,
+      "loss": 0.2861,
+      "step": 4800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882886427527113e-06,
+      "loss": 0.7798,
+      "step": 4801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.879287817185694e-06,
+      "loss": 0.7151,
+      "step": 4802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.875691076495232e-06,
+      "loss": 0.6609,
+      "step": 4803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.872096206400099e-06,
+      "loss": 0.7245,
+      "step": 4804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868503207844191e-06,
+      "loss": 0.7604,
+      "step": 4805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864912081770904e-06,
+      "loss": 0.7134,
+      "step": 4806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8613228291231367e-06,
+      "loss": 0.6991,
+      "step": 4807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.857735450843303e-06,
+      "loss": 0.7417,
+      "step": 4808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.854149947873326e-06,
+      "loss": 0.7047,
+      "step": 4809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8505663211546297e-06,
+      "loss": 0.7668,
+      "step": 4810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846984571628155e-06,
+      "loss": 0.7158,
+      "step": 4811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8434047002343425e-06,
+      "loss": 0.7231,
+      "step": 4812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8398267079131412e-06,
+      "loss": 0.7262,
+      "step": 4813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.836250595604009e-06,
+      "loss": 0.7198,
+      "step": 4814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8326763642459053e-06,
+      "loss": 0.7096,
+      "step": 4815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8291040147773006e-06,
+      "loss": 0.2725,
+      "step": 4816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825533548136169e-06,
+      "loss": 0.8069,
+      "step": 4817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8219649652599914e-06,
+      "loss": 0.7488,
+      "step": 4818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8183982670857525e-06,
+      "loss": 0.7681,
+      "step": 4819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8148334545499435e-06,
+      "loss": 0.7032,
+      "step": 4820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811270528588559e-06,
+      "loss": 0.7461,
+      "step": 4821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8077094901371005e-06,
+      "loss": 0.7876,
+      "step": 4822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8041503401305714e-06,
+      "loss": 0.7418,
+      "step": 4823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8005930795034784e-06,
+      "loss": 0.7289,
+      "step": 4824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7970377091898404e-06,
+      "loss": 0.723,
+      "step": 4825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7934842301231702e-06,
+      "loss": 0.7,
+      "step": 4826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7899326432364883e-06,
+      "loss": 0.7261,
+      "step": 4827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786382949462316e-06,
+      "loss": 0.6639,
+      "step": 4828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782835149732682e-06,
+      "loss": 0.7922,
+      "step": 4829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779289244979112e-06,
+      "loss": 0.6709,
+      "step": 4830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77574523613264e-06,
+      "loss": 0.2607,
+      "step": 4831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7722031241237958e-06,
+      "loss": 0.7942,
+      "step": 4832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768662909882618e-06,
+      "loss": 0.821,
+      "step": 4833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765124594338641e-06,
+      "loss": 0.6953,
+      "step": 4834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7615881784209043e-06,
+      "loss": 0.6866,
+      "step": 4835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.758053663057948e-06,
+      "loss": 0.7539,
+      "step": 4836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.754521049177811e-06,
+      "loss": 0.2911,
+      "step": 4837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.750990337708034e-06,
+      "loss": 0.73,
+      "step": 4838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7474615295756623e-06,
+      "loss": 0.6654,
+      "step": 4839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7439346257072376e-06,
+      "loss": 0.712,
+      "step": 4840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7404096270288007e-06,
+      "loss": 0.7113,
+      "step": 4841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7368865344658957e-06,
+      "loss": 0.6992,
+      "step": 4842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7333653489435587e-06,
+      "loss": 0.7004,
+      "step": 4843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7298460713863305e-06,
+      "loss": 0.6888,
+      "step": 4844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726328702718256e-06,
+      "loss": 0.3039,
+      "step": 4845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722813243862872e-06,
+      "loss": 0.7527,
+      "step": 4846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719299695743215e-06,
+      "loss": 0.6914,
+      "step": 4847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7157880592818196e-06,
+      "loss": 0.7207,
+      "step": 4848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.71227833540072e-06,
+      "loss": 0.6738,
+      "step": 4849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708770525021448e-06,
+      "loss": 0.7114,
+      "step": 4850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7052646290650276e-06,
+      "loss": 0.6738,
+      "step": 4851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701760648451992e-06,
+      "loss": 0.7595,
+      "step": 4852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.698258584102361e-06,
+      "loss": 0.6993,
+      "step": 4853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6947584369356593e-06,
+      "loss": 0.7183,
+      "step": 4854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.691260207870896e-06,
+      "loss": 0.7703,
+      "step": 4855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6877638978265874e-06,
+      "loss": 0.7129,
+      "step": 4856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6842695077207394e-06,
+      "loss": 0.6968,
+      "step": 4857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680777038470864e-06,
+      "loss": 0.7025,
+      "step": 4858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677286490993959e-06,
+      "loss": 0.2761,
+      "step": 4859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.673797866206519e-06,
+      "loss": 0.7471,
+      "step": 4860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.670311165024537e-06,
+      "loss": 0.7117,
+      "step": 4861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.666826388363499e-06,
+      "loss": 0.7368,
+      "step": 4862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6633435371383855e-06,
+      "loss": 0.7119,
+      "step": 4863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.659862612263673e-06,
+      "loss": 0.7028,
+      "step": 4864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6563836146533285e-06,
+      "loss": 0.7153,
+      "step": 4865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652906545220818e-06,
+      "loss": 0.6931,
+      "step": 4866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6494314048790982e-06,
+      "loss": 0.7039,
+      "step": 4867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.64595819454062e-06,
+      "loss": 0.7362,
+      "step": 4868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642486915117327e-06,
+      "loss": 0.7408,
+      "step": 4869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6390175675206563e-06,
+      "loss": 0.7311,
+      "step": 4870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6355501526615347e-06,
+      "loss": 0.7493,
+      "step": 4871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6320846714503913e-06,
+      "loss": 0.7252,
+      "step": 4872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.628621124797136e-06,
+      "loss": 0.7358,
+      "step": 4873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6251595136111773e-06,
+      "loss": 0.7291,
+      "step": 4874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6216998388014125e-06,
+      "loss": 0.7083,
+      "step": 4875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.618242101276234e-06,
+      "loss": 0.7603,
+      "step": 4876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6147863019435194e-06,
+      "loss": 0.7239,
+      "step": 4877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6113324417106447e-06,
+      "loss": 0.7756,
+      "step": 4878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6078805214844707e-06,
+      "loss": 0.7417,
+      "step": 4879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6044305421713523e-06,
+      "loss": 0.6637,
+      "step": 4880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6009825046771353e-06,
+      "loss": 0.7351,
+      "step": 4881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597536409907152e-06,
+      "loss": 0.7407,
+      "step": 4882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594092258766229e-06,
+      "loss": 0.6879,
+      "step": 4883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5906500521586786e-06,
+      "loss": 0.7432,
+      "step": 4884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.587209790988302e-06,
+      "loss": 0.7023,
+      "step": 4885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5837714761583977e-06,
+      "loss": 0.7351,
+      "step": 4886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5803351085717455e-06,
+      "loss": 0.6978,
+      "step": 4887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5769006891306134e-06,
+      "loss": 0.7163,
+      "step": 4888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5734682187367667e-06,
+      "loss": 0.6617,
+      "step": 4889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5700376982914434e-06,
+      "loss": 0.6567,
+      "step": 4890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.56660912869538e-06,
+      "loss": 0.702,
+      "step": 4891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5631825108488064e-06,
+      "loss": 0.7373,
+      "step": 4892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5597578456514284e-06,
+      "loss": 0.7423,
+      "step": 4893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5563351340024443e-06,
+      "loss": 0.7205,
+      "step": 4894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5529143768005395e-06,
+      "loss": 0.6814,
+      "step": 4895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5494955749438845e-06,
+      "loss": 0.7158,
+      "step": 4896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546078729330139e-06,
+      "loss": 0.7783,
+      "step": 4897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5426638408564464e-06,
+      "loss": 0.7012,
+      "step": 4898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5392509104194363e-06,
+      "loss": 0.7354,
+      "step": 4899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.535839938915229e-06,
+      "loss": 0.7649,
+      "step": 4900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5324309272394276e-06,
+      "loss": 0.6534,
+      "step": 4901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5290238762871146e-06,
+      "loss": 0.6803,
+      "step": 4902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.525618786952865e-06,
+      "loss": 0.2552,
+      "step": 4903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.522215660130737e-06,
+      "loss": 0.7222,
+      "step": 4904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5188144967142703e-06,
+      "loss": 0.6894,
+      "step": 4905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5154152975964975e-06,
+      "loss": 0.709,
+      "step": 4906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5120180636699275e-06,
+      "loss": 0.7246,
+      "step": 4907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508622795826554e-06,
+      "loss": 0.6979,
+      "step": 4908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505229494957858e-06,
+      "loss": 0.7773,
+      "step": 4909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5018381619548015e-06,
+      "loss": 0.7123,
+      "step": 4910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.498448797707831e-06,
+      "loss": 0.7249,
+      "step": 4911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.495061403106871e-06,
+      "loss": 0.769,
+      "step": 4912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4916759790413424e-06,
+      "loss": 0.7197,
+      "step": 4913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488292526400132e-06,
+      "loss": 0.6741,
+      "step": 4914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4849110460716185e-06,
+      "loss": 0.7291,
+      "step": 4915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4815315389436613e-06,
+      "loss": 0.7617,
+      "step": 4916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4781540059036014e-06,
+      "loss": 0.7734,
+      "step": 4917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4747784478382575e-06,
+      "loss": 0.7004,
+      "step": 4918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4714048656339406e-06,
+      "loss": 0.2717,
+      "step": 4919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4680332601764313e-06,
+      "loss": 0.6902,
+      "step": 4920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4646636323509974e-06,
+      "loss": 0.6963,
+      "step": 4921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4612959830423855e-06,
+      "loss": 0.7532,
+      "step": 4922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.457930313134823e-06,
+      "loss": 0.6809,
+      "step": 4923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4545666235120157e-06,
+      "loss": 0.2554,
+      "step": 4924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451204915057155e-06,
+      "loss": 0.6755,
+      "step": 4925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.447845188652904e-06,
+      "loss": 0.6894,
+      "step": 4926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444487445181414e-06,
+      "loss": 0.6927,
+      "step": 4927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441131685524308e-06,
+      "loss": 0.6829,
+      "step": 4928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437777910562694e-06,
+      "loss": 0.6609,
+      "step": 4929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434426121177156e-06,
+      "loss": 0.6766,
+      "step": 4930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.431076318247756e-06,
+      "loss": 0.7163,
+      "step": 4931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.427728502654033e-06,
+      "loss": 0.7402,
+      "step": 4932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4243826752750134e-06,
+      "loss": 0.6345,
+      "step": 4933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4210388369891914e-06,
+      "loss": 0.7054,
+      "step": 4934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4176969886745426e-06,
+      "loss": 0.7283,
+      "step": 4935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4143571312085223e-06,
+      "loss": 0.7546,
+      "step": 4936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4110192654680566e-06,
+      "loss": 0.7097,
+      "step": 4937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4076833923295506e-06,
+      "loss": 0.7197,
+      "step": 4938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404349512668895e-06,
+      "loss": 0.6556,
+      "step": 4939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401017627361448e-06,
+      "loss": 0.6821,
+      "step": 4940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3976877372820474e-06,
+      "loss": 0.7308,
+      "step": 4941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3943598433050043e-06,
+      "loss": 0.652,
+      "step": 4942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3910339463041087e-06,
+      "loss": 0.7146,
+      "step": 4943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387710047152626e-06,
+      "loss": 0.7468,
+      "step": 4944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3843881467232945e-06,
+      "loss": 0.6938,
+      "step": 4945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381068245888328e-06,
+      "loss": 0.6995,
+      "step": 4946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3777503455194227e-06,
+      "loss": 0.7939,
+      "step": 4947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3744344464877435e-06,
+      "loss": 0.749,
+      "step": 4948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3711205496639234e-06,
+      "loss": 0.7053,
+      "step": 4949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.36780865591808e-06,
+      "loss": 0.7627,
+      "step": 4950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3644987661198005e-06,
+      "loss": 0.769,
+      "step": 4951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3611908811381445e-06,
+      "loss": 0.6788,
+      "step": 4952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3578850018416535e-06,
+      "loss": 0.7515,
+      "step": 4953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3545811290983324e-06,
+      "loss": 0.2595,
+      "step": 4954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351279263775662e-06,
+      "loss": 0.7207,
+      "step": 4955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3479794067406e-06,
+      "loss": 0.719,
+      "step": 4956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3446815588595716e-06,
+      "loss": 0.7047,
+      "step": 4957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3413857209984782e-06,
+      "loss": 0.6741,
+      "step": 4958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3380918940226883e-06,
+      "loss": 0.6737,
+      "step": 4959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3348000787970548e-06,
+      "loss": 0.6759,
+      "step": 4960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3315102761858852e-06,
+      "loss": 0.7739,
+      "step": 4961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.328222487052969e-06,
+      "loss": 0.731,
+      "step": 4962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3249367122615673e-06,
+      "loss": 0.7549,
+      "step": 4963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3216529526744082e-06,
+      "loss": 0.6543,
+      "step": 4964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3183712091536905e-06,
+      "loss": 0.7314,
+      "step": 4965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3150914825610914e-06,
+      "loss": 0.683,
+      "step": 4966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3118137737577497e-06,
+      "loss": 0.6989,
+      "step": 4967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308538083604278e-06,
+      "loss": 0.7146,
+      "step": 4968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3052644129607584e-06,
+      "loss": 0.7136,
+      "step": 4969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3019927626867412e-06,
+      "loss": 0.7539,
+      "step": 4970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2987231336412497e-06,
+      "loss": 0.7335,
+      "step": 4971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295455526682774e-06,
+      "loss": 0.2838,
+      "step": 4972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292189942669274e-06,
+      "loss": 0.6912,
+      "step": 4973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2889263824581775e-06,
+      "loss": 0.6642,
+      "step": 4974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2856648469063834e-06,
+      "loss": 0.7018,
+      "step": 4975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2824053368702546e-06,
+      "loss": 0.7452,
+      "step": 4976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2791478532056276e-06,
+      "loss": 0.7341,
+      "step": 4977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2758923967678036e-06,
+      "loss": 0.7067,
+      "step": 4978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272638968411548e-06,
+      "loss": 0.78,
+      "step": 4979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269387568991104e-06,
+      "loss": 0.7368,
+      "step": 4980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.266138199360174e-06,
+      "loss": 0.7059,
+      "step": 4981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262890860371927e-06,
+      "loss": 0.72,
+      "step": 4982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.259645552879005e-06,
+      "loss": 0.7485,
+      "step": 4983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2564022777335083e-06,
+      "loss": 0.6895,
+      "step": 4984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.253161035787007e-06,
+      "loss": 0.7239,
+      "step": 4985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2499218278905433e-06,
+      "loss": 0.7214,
+      "step": 4986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2466846548946185e-06,
+      "loss": 0.7214,
+      "step": 4987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243449517649201e-06,
+      "loss": 0.714,
+      "step": 4988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2402164170037256e-06,
+      "loss": 0.688,
+      "step": 4989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.236985353807092e-06,
+      "loss": 0.6604,
+      "step": 4990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2337563289076637e-06,
+      "loss": 0.7301,
+      "step": 4991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2305293431532705e-06,
+      "loss": 0.7338,
+      "step": 4992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2273043973912047e-06,
+      "loss": 0.7057,
+      "step": 4993
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2240814924682286e-06,
+      "loss": 0.634,
+      "step": 4994
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2208606292305658e-06,
+      "loss": 0.7502,
+      "step": 4995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217641808523898e-06,
+      "loss": 0.7083,
+      "step": 4996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214425031193377e-06,
+      "loss": 0.2921,
+      "step": 4997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2112102980836157e-06,
+      "loss": 0.729,
+      "step": 4998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2079976100386915e-06,
+      "loss": 0.7015,
+      "step": 4999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.204786967902146e-06,
+      "loss": 0.6909,
+      "step": 5000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2015783725169816e-06,
+      "loss": 0.634,
+      "step": 5001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1983718247256623e-06,
+      "loss": 0.75,
+      "step": 5002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1951673253701167e-06,
+      "loss": 0.7401,
+      "step": 5003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1919648752917356e-06,
+      "loss": 0.2781,
+      "step": 5004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1887644753313687e-06,
+      "loss": 0.265,
+      "step": 5005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185566126329328e-06,
+      "loss": 0.7915,
+      "step": 5006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.182369829125397e-06,
+      "loss": 0.7266,
+      "step": 5007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1791755845588035e-06,
+      "loss": 0.6829,
+      "step": 5008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175983393468246e-06,
+      "loss": 0.701,
+      "step": 5009
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1727932566918864e-06,
+      "loss": 0.6627,
+      "step": 5010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1696051750673397e-06,
+      "loss": 0.7418,
+      "step": 5011
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1664191494316844e-06,
+      "loss": 0.6802,
+      "step": 5012
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1632351806214647e-06,
+      "loss": 0.6735,
+      "step": 5013
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160053269472677e-06,
+      "loss": 0.7603,
+      "step": 5014
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1568734168207806e-06,
+      "loss": 0.6917,
+      "step": 5015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1536956235006936e-06,
+      "loss": 0.7561,
+      "step": 5016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150519890346795e-06,
+      "loss": 0.7196,
+      "step": 5017
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1473462181929206e-06,
+      "loss": 0.6947,
+      "step": 5018
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1441746078723656e-06,
+      "loss": 0.7095,
+      "step": 5019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1410050602178866e-06,
+      "loss": 0.697,
+      "step": 5020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137837576061694e-06,
+      "loss": 0.7544,
+      "step": 5021
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1346721562354598e-06,
+      "loss": 0.7209,
+      "step": 5022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1315088015703135e-06,
+      "loss": 0.7549,
+      "step": 5023
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128347512896841e-06,
+      "loss": 0.708,
+      "step": 5024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125188291045087e-06,
+      "loss": 0.6688,
+      "step": 5025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122031136844551e-06,
+      "loss": 0.2545,
+      "step": 5026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1188760511241956e-06,
+      "loss": 0.7537,
+      "step": 5027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1157230347124346e-06,
+      "loss": 0.7222,
+      "step": 5028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1125720884371404e-06,
+      "loss": 0.7267,
+      "step": 5029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1094232131256433e-06,
+      "loss": 0.7417,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106276409604725e-06,
+      "loss": 0.7024,
+      "step": 5031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103131678700624e-06,
+      "loss": 0.2771,
+      "step": 5032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099989021239045e-06,
+      "loss": 0.7715,
+      "step": 5033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0968484380451347e-06,
+      "loss": 0.7007,
+      "step": 5034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093709929943504e-06,
+      "loss": 0.7139,
+      "step": 5035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0905734977582138e-06,
+      "loss": 0.6984,
+      "step": 5036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0874391423127826e-06,
+      "loss": 0.7141,
+      "step": 5037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0843068644301822e-06,
+      "loss": 0.2756,
+      "step": 5038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08117666493284e-06,
+      "loss": 0.6868,
+      "step": 5039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0780485446426356e-06,
+      "loss": 0.7277,
+      "step": 5040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0749225043809096e-06,
+      "loss": 0.7323,
+      "step": 5041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0717985449684496e-06,
+      "loss": 0.2731,
+      "step": 5042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0686766672254954e-06,
+      "loss": 0.6829,
+      "step": 5043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0655568719717447e-06,
+      "loss": 0.7158,
+      "step": 5044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0624391600263462e-06,
+      "loss": 0.7213,
+      "step": 5045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.059323532207901e-06,
+      "loss": 0.7036,
+      "step": 5046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05620998933447e-06,
+      "loss": 0.7471,
+      "step": 5047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0530985322235587e-06,
+      "loss": 0.7527,
+      "step": 5048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0499891616921253e-06,
+      "loss": 0.663,
+      "step": 5049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046881878556585e-06,
+      "loss": 0.7004,
+      "step": 5050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0437766836327997e-06,
+      "loss": 0.7115,
+      "step": 5051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0406735777360864e-06,
+      "loss": 0.6968,
+      "step": 5052
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0375725616812124e-06,
+      "loss": 0.7218,
+      "step": 5053
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034473636282397e-06,
+      "loss": 0.7173,
+      "step": 5054
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03137680235331e-06,
+      "loss": 0.7229,
+      "step": 5055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.028282060707072e-06,
+      "loss": 0.6915,
+      "step": 5056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0251894121562533e-06,
+      "loss": 0.7144,
+      "step": 5057
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0220988575128764e-06,
+      "loss": 0.7477,
+      "step": 5058
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0190103975884147e-06,
+      "loss": 0.6644,
+      "step": 5059
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0159240331937845e-06,
+      "loss": 0.7253,
+      "step": 5060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012839765139366e-06,
+      "loss": 0.7881,
+      "step": 5061
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0097575942349758e-06,
+      "loss": 0.6908,
+      "step": 5062
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006677521289886e-06,
+      "loss": 0.7016,
+      "step": 5063
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0035995471128145e-06,
+      "loss": 0.7832,
+      "step": 5064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0005236725119324e-06,
+      "loss": 0.7795,
+      "step": 5065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9974498982948565e-06,
+      "loss": 0.7288,
+      "step": 5066
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9943782252686518e-06,
+      "loss": 0.7518,
+      "step": 5067
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991308654239833e-06,
+      "loss": 0.73,
+      "step": 5068
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9882411860143635e-06,
+      "loss": 0.6582,
+      "step": 5069
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9851758213976522e-06,
+      "loss": 0.7139,
+      "step": 5070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982112561194559e-06,
+      "loss": 0.7283,
+      "step": 5071
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979051406209387e-06,
+      "loss": 0.6259,
+      "step": 5072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9759923572458885e-06,
+      "loss": 0.7515,
+      "step": 5073
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9729354151072666e-06,
+      "loss": 0.7123,
+      "step": 5074
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9698805805961653e-06,
+      "loss": 0.6938,
+      "step": 5075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9668278545146792e-06,
+      "loss": 0.7277,
+      "step": 5076
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9637772376643462e-06,
+      "loss": 0.7018,
+      "step": 5077
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9607287308461554e-06,
+      "loss": 0.7302,
+      "step": 5078
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9576823348605313e-06,
+      "loss": 0.7288,
+      "step": 5079
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954638050507358e-06,
+      "loss": 0.6835,
+      "step": 5080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951595878585957e-06,
+      "loss": 0.7126,
+      "step": 5081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948555819895095e-06,
+      "loss": 0.6793,
+      "step": 5082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9455178752329874e-06,
+      "loss": 0.7437,
+      "step": 5083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9424820453972915e-06,
+      "loss": 0.7024,
+      "step": 5084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9394483311851107e-06,
+      "loss": 0.7583,
+      "step": 5085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936416733392993e-06,
+      "loss": 0.6509,
+      "step": 5086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.933387252816927e-06,
+      "loss": 0.7134,
+      "step": 5087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9303598902523546e-06,
+      "loss": 0.7133,
+      "step": 5088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9273346464941555e-06,
+      "loss": 0.6792,
+      "step": 5089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.924311522336648e-06,
+      "loss": 0.7178,
+      "step": 5090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.921290518573603e-06,
+      "loss": 0.717,
+      "step": 5091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.918271635998229e-06,
+      "loss": 0.749,
+      "step": 5092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9152548754031784e-06,
+      "loss": 0.7163,
+      "step": 5093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9122402375805516e-06,
+      "loss": 0.7253,
+      "step": 5094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909227723321885e-06,
+      "loss": 0.7114,
+      "step": 5095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.90621733341816e-06,
+      "loss": 0.6833,
+      "step": 5096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9032090686597992e-06,
+      "loss": 0.754,
+      "step": 5097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.900202929836671e-06,
+      "loss": 0.7197,
+      "step": 5098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8971989177380802e-06,
+      "loss": 0.7017,
+      "step": 5099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8941970331527759e-06,
+      "loss": 0.7327,
+      "step": 5100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8911972768689491e-06,
+      "loss": 0.699,
+      "step": 5101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8881996496742305e-06,
+      "loss": 0.686,
+      "step": 5102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8852041523556941e-06,
+      "loss": 0.6648,
+      "step": 5103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8822107856998517e-06,
+      "loss": 0.7172,
+      "step": 5104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.879219550492657e-06,
+      "loss": 0.7267,
+      "step": 5105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8762304475195058e-06,
+      "loss": 0.7258,
+      "step": 5106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8732434775652276e-06,
+      "loss": 0.724,
+      "step": 5107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8702586414141044e-06,
+      "loss": 0.7114,
+      "step": 5108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8672759398498453e-06,
+      "loss": 0.7397,
+      "step": 5109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8642953736556047e-06,
+      "loss": 0.7452,
+      "step": 5110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8613169436139755e-06,
+      "loss": 0.2838,
+      "step": 5111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.85834065050699e-06,
+      "loss": 0.7214,
+      "step": 5112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8553664951161189e-06,
+      "loss": 0.6532,
+      "step": 5113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8523944782222703e-06,
+      "loss": 0.6879,
+      "step": 5114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8494246006057947e-06,
+      "loss": 0.7178,
+      "step": 5115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8464568630464775e-06,
+      "loss": 0.2845,
+      "step": 5116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8434912663235427e-06,
+      "loss": 0.6963,
+      "step": 5117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8405278112156533e-06,
+      "loss": 0.6903,
+      "step": 5118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837566498500909e-06,
+      "loss": 0.6987,
+      "step": 5119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8346073289568456e-06,
+      "loss": 0.7542,
+      "step": 5120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8316503033604415e-06,
+      "loss": 0.7412,
+      "step": 5121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8286954224881081e-06,
+      "loss": 0.7065,
+      "step": 5122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8257426871156925e-06,
+      "loss": 0.7003,
+      "step": 5123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8227920980184798e-06,
+      "loss": 0.7502,
+      "step": 5124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8198436559711963e-06,
+      "loss": 0.2913,
+      "step": 5125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8168973617479913e-06,
+      "loss": 0.6927,
+      "step": 5126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813953216122467e-06,
+      "loss": 0.75,
+      "step": 5127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.811011219867651e-06,
+      "loss": 0.7271,
+      "step": 5128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8080713737560086e-06,
+      "loss": 0.7157,
+      "step": 5129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.805133678559442e-06,
+      "loss": 0.6769,
+      "step": 5130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8021981350492857e-06,
+      "loss": 0.6727,
+      "step": 5131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799264743996313e-06,
+      "loss": 0.7061,
+      "step": 5132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.79633350617073e-06,
+      "loss": 0.7136,
+      "step": 5133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7934044223421742e-06,
+      "loss": 0.6896,
+      "step": 5134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7904774932797263e-06,
+      "loss": 0.7307,
+      "step": 5135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7875527197518937e-06,
+      "loss": 0.7495,
+      "step": 5136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7846301025266222e-06,
+      "loss": 0.7615,
+      "step": 5137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7817096423712842e-06,
+      "loss": 0.6827,
+      "step": 5138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7787913400526945e-06,
+      "loss": 0.688,
+      "step": 5139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7758751963370935e-06,
+      "loss": 0.6864,
+      "step": 5140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7729612119901652e-06,
+      "loss": 0.6997,
+      "step": 5141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7700493877770165e-06,
+      "loss": 0.7014,
+      "step": 5142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7671397244621923e-06,
+      "loss": 0.7101,
+      "step": 5143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7642322228096686e-06,
+      "loss": 0.6735,
+      "step": 5144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7613268835828535e-06,
+      "loss": 0.6864,
+      "step": 5145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7584237075445877e-06,
+      "loss": 0.7021,
+      "step": 5146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7555226954571448e-06,
+      "loss": 0.7415,
+      "step": 5147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.75262384808223e-06,
+      "loss": 0.6562,
+      "step": 5148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7497271661809777e-06,
+      "loss": 0.7195,
+      "step": 5149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7468326505139576e-06,
+      "loss": 0.7166,
+      "step": 5150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7439403018411682e-06,
+      "loss": 0.6771,
+      "step": 5151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7410501209220388e-06,
+      "loss": 0.2925,
+      "step": 5152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7381621085154298e-06,
+      "loss": 0.7106,
+      "step": 5153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7352762653796318e-06,
+      "loss": 0.7196,
+      "step": 5154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7323925922723706e-06,
+      "loss": 0.7412,
+      "step": 5155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7295110899507972e-06,
+      "loss": 0.7041,
+      "step": 5156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.726631759171491e-06,
+      "loss": 0.7712,
+      "step": 5157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7237546006904659e-06,
+      "loss": 0.7859,
+      "step": 5158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7208796152631636e-06,
+      "loss": 0.729,
+      "step": 5159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7180068036444552e-06,
+      "loss": 0.6775,
+      "step": 5160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.715136166588639e-06,
+      "loss": 0.7678,
+      "step": 5161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7122677048494462e-06,
+      "loss": 0.729,
+      "step": 5162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7094014191800356e-06,
+      "loss": 0.707,
+      "step": 5163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065373103329919e-06,
+      "loss": 0.6494,
+      "step": 5164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70367537906033e-06,
+      "loss": 0.7059,
+      "step": 5165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7008156261134957e-06,
+      "loss": 0.7524,
+      "step": 5166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697958052243357e-06,
+      "loss": 0.6904,
+      "step": 5167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951026582002185e-06,
+      "loss": 0.72,
+      "step": 5168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6922494447338034e-06,
+      "loss": 0.759,
+      "step": 5169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6893984125932672e-06,
+      "loss": 0.6985,
+      "step": 5170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.686549562527191e-06,
+      "loss": 0.7161,
+      "step": 5171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6837028952835865e-06,
+      "loss": 0.6768,
+      "step": 5172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6808584116098848e-06,
+      "loss": 0.6849,
+      "step": 5173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.678016112252948e-06,
+      "loss": 0.6982,
+      "step": 5174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.67517599795907e-06,
+      "loss": 0.7834,
+      "step": 5175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6723380694739621e-06,
+      "loss": 0.6874,
+      "step": 5176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6695023275427668e-06,
+      "loss": 0.6802,
+      "step": 5177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6666687729100516e-06,
+      "loss": 0.6108,
+      "step": 5178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.663837406319808e-06,
+      "loss": 0.7162,
+      "step": 5179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661008228515454e-06,
+      "loss": 0.7089,
+      "step": 5180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6581812402398323e-06,
+      "loss": 0.6595,
+      "step": 5181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655356442235214e-06,
+      "loss": 0.6818,
+      "step": 5182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.652533835243293e-06,
+      "loss": 0.733,
+      "step": 5183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.649713420005188e-06,
+      "loss": 0.6901,
+      "step": 5184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6468951972614377e-06,
+      "loss": 0.7007,
+      "step": 5185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6440791677520117e-06,
+      "loss": 0.7449,
+      "step": 5186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6412653322162986e-06,
+      "loss": 0.7346,
+      "step": 5187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6384536913931183e-06,
+      "loss": 0.7642,
+      "step": 5188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6356442460207079e-06,
+      "loss": 0.7527,
+      "step": 5189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6328369968367286e-06,
+      "loss": 0.7532,
+      "step": 5190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6300319445782676e-06,
+      "loss": 0.7515,
+      "step": 5191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6272290899818332e-06,
+      "loss": 0.7545,
+      "step": 5192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6244284337833583e-06,
+      "loss": 0.7354,
+      "step": 5193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6216299767181964e-06,
+      "loss": 0.6064,
+      "step": 5194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6188337195211223e-06,
+      "loss": 0.6953,
+      "step": 5195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6160396629263452e-06,
+      "loss": 0.7134,
+      "step": 5196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6132478076674774e-06,
+      "loss": 0.7563,
+      "step": 5197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6104581544775655e-06,
+      "loss": 0.7101,
+      "step": 5198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607670704089076e-06,
+      "loss": 0.7046,
+      "step": 5199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6048854572338969e-06,
+      "loss": 0.7183,
+      "step": 5200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.602102414643333e-06,
+      "loss": 0.7209,
+      "step": 5201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5993215770481196e-06,
+      "loss": 0.7417,
+      "step": 5202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5965429451784064e-06,
+      "loss": 0.7343,
+      "step": 5203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5937665197637642e-06,
+      "loss": 0.6996,
+      "step": 5204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5909923015331863e-06,
+      "loss": 0.7749,
+      "step": 5205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5882202912150857e-06,
+      "loss": 0.7717,
+      "step": 5206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5854504895372946e-06,
+      "loss": 0.6169,
+      "step": 5207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.582682897227068e-06,
+      "loss": 0.7312,
+      "step": 5208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579917515011078e-06,
+      "loss": 0.6218,
+      "step": 5209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577154343615419e-06,
+      "loss": 0.7102,
+      "step": 5210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5743933837656023e-06,
+      "loss": 0.7576,
+      "step": 5211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5716346361865598e-06,
+      "loss": 0.7285,
+      "step": 5212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5688781016026434e-06,
+      "loss": 0.6479,
+      "step": 5213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5661237807376217e-06,
+      "loss": 0.2355,
+      "step": 5214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5633716743146832e-06,
+      "loss": 0.6899,
+      "step": 5215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5606217830564375e-06,
+      "loss": 0.7186,
+      "step": 5216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.557874107684909e-06,
+      "loss": 0.7368,
+      "step": 5217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5551286489215423e-06,
+      "loss": 0.7888,
+      "step": 5218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5523854074872003e-06,
+      "loss": 0.6978,
+      "step": 5219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5496443841021592e-06,
+      "loss": 0.7991,
+      "step": 5220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5469055794861143e-06,
+      "loss": 0.6731,
+      "step": 5221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5441689943581873e-06,
+      "loss": 0.2546,
+      "step": 5222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5414346294369054e-06,
+      "loss": 0.7163,
+      "step": 5223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.53870248544022e-06,
+      "loss": 0.6396,
+      "step": 5224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535972563085495e-06,
+      "loss": 0.7361,
+      "step": 5225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5332448630895136e-06,
+      "loss": 0.3008,
+      "step": 5226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5305193861684753e-06,
+      "loss": 0.7239,
+      "step": 5227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.527796133037992e-06,
+      "loss": 0.2579,
+      "step": 5228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5250751044131006e-06,
+      "loss": 0.3085,
+      "step": 5229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5223563010082465e-06,
+      "loss": 0.736,
+      "step": 5230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.519639723537294e-06,
+      "loss": 0.6095,
+      "step": 5231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.516925372713517e-06,
+      "loss": 0.7036,
+      "step": 5232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5142132492496131e-06,
+      "loss": 0.7008,
+      "step": 5233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5115033538576874e-06,
+      "loss": 0.6847,
+      "step": 5234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5087956872492703e-06,
+      "loss": 0.7074,
+      "step": 5235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.506090250135297e-06,
+      "loss": 0.7365,
+      "step": 5236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5033870432261222e-06,
+      "loss": 0.6421,
+      "step": 5237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5006860672315139e-06,
+      "loss": 0.6744,
+      "step": 5238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4979873228606533e-06,
+      "loss": 0.6776,
+      "step": 5239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4952908108221365e-06,
+      "loss": 0.6792,
+      "step": 5240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4925965318239754e-06,
+      "loss": 0.6927,
+      "step": 5241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489904486573589e-06,
+      "loss": 0.6896,
+      "step": 5242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.487214675777824e-06,
+      "loss": 0.7456,
+      "step": 5243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4845271001429228e-06,
+      "loss": 0.6997,
+      "step": 5244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4818417603745516e-06,
+      "loss": 0.8013,
+      "step": 5245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.479158657177786e-06,
+      "loss": 0.6436,
+      "step": 5246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4764777912571183e-06,
+      "loss": 0.6637,
+      "step": 5247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.473799163316445e-06,
+      "loss": 0.738,
+      "step": 5248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4711227740590871e-06,
+      "loss": 0.7434,
+      "step": 5249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4684486241877683e-06,
+      "loss": 0.7275,
+      "step": 5250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4657767144046264e-06,
+      "loss": 0.2792,
+      "step": 5251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4631070454112127e-06,
+      "loss": 0.7333,
+      "step": 5252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4604396179084901e-06,
+      "loss": 0.7393,
+      "step": 5253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4577744325968313e-06,
+      "loss": 0.7441,
+      "step": 5254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.45511149017602e-06,
+      "loss": 0.6571,
+      "step": 5255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4524507913452536e-06,
+      "loss": 0.7195,
+      "step": 5256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4497923368031387e-06,
+      "loss": 0.7129,
+      "step": 5257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4471361272476925e-06,
+      "loss": 0.7122,
+      "step": 5258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4444821633763428e-06,
+      "loss": 0.6537,
+      "step": 5259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4418304458859288e-06,
+      "loss": 0.7029,
+      "step": 5260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4391809754726982e-06,
+      "loss": 0.728,
+      "step": 5261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4365337528323087e-06,
+      "loss": 0.6808,
+      "step": 5262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4338887786598322e-06,
+      "loss": 0.7086,
+      "step": 5263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4312460536497452e-06,
+      "loss": 0.7156,
+      "step": 5264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.428605578495934e-06,
+      "loss": 0.7117,
+      "step": 5265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4259673538916996e-06,
+      "loss": 0.704,
+      "step": 5266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4233313805297421e-06,
+      "loss": 0.7168,
+      "step": 5267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.420697659102176e-06,
+      "loss": 0.7203,
+      "step": 5268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.41806619030053e-06,
+      "loss": 0.6913,
+      "step": 5269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4154369748157327e-06,
+      "loss": 0.6752,
+      "step": 5270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.412810013338126e-06,
+      "loss": 0.6732,
+      "step": 5271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410185306557459e-06,
+      "loss": 0.7404,
+      "step": 5272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4075628551628863e-06,
+      "loss": 0.762,
+      "step": 5273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4049426598429726e-06,
+      "loss": 0.6829,
+      "step": 5274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4023247212856895e-06,
+      "loss": 0.7136,
+      "step": 5275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3997090401784185e-06,
+      "loss": 0.6757,
+      "step": 5276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3970956172079453e-06,
+      "loss": 0.7148,
+      "step": 5277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3944844530604663e-06,
+      "loss": 0.7019,
+      "step": 5278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3918755484215774e-06,
+      "loss": 0.7384,
+      "step": 5279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3892689039762874e-06,
+      "loss": 0.6733,
+      "step": 5280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.38666452040901e-06,
+      "loss": 0.6934,
+      "step": 5281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3840623984035695e-06,
+      "loss": 0.7133,
+      "step": 5282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.381462538643189e-06,
+      "loss": 0.2364,
+      "step": 5283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3788649418105015e-06,
+      "loss": 0.7029,
+      "step": 5284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376269608587546e-06,
+      "loss": 0.6833,
+      "step": 5285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3736765396557672e-06,
+      "loss": 0.3005,
+      "step": 5286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3710857356960128e-06,
+      "loss": 0.6663,
+      "step": 5287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3684971973885397e-06,
+      "loss": 0.7064,
+      "step": 5288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3659109254130042e-06,
+      "loss": 0.7137,
+      "step": 5289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363326920448479e-06,
+      "loss": 0.6787,
+      "step": 5290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3607451831734265e-06,
+      "loss": 0.6875,
+      "step": 5291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3581657142657224e-06,
+      "loss": 0.7375,
+      "step": 5292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3555885144026458e-06,
+      "loss": 0.7449,
+      "step": 5293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3530135842608816e-06,
+      "loss": 0.7284,
+      "step": 5294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3504409245165118e-06,
+      "loss": 0.7598,
+      "step": 5295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3478705358450329e-06,
+      "loss": 0.7412,
+      "step": 5296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3453024189213372e-06,
+      "loss": 0.6798,
+      "step": 5297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342736574419723e-06,
+      "loss": 0.6881,
+      "step": 5298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3401730030138914e-06,
+      "loss": 0.7242,
+      "step": 5299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3376117053769455e-06,
+      "loss": 0.7599,
+      "step": 5300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3350526821813957e-06,
+      "loss": 0.7461,
+      "step": 5301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3324959340991505e-06,
+      "loss": 0.6678,
+      "step": 5302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3299414618015226e-06,
+      "loss": 0.6448,
+      "step": 5303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.327389265959228e-06,
+      "loss": 0.6713,
+      "step": 5304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3248393472423847e-06,
+      "loss": 0.7297,
+      "step": 5305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3222917063205133e-06,
+      "loss": 0.719,
+      "step": 5306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3197463438625336e-06,
+      "loss": 0.7058,
+      "step": 5307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3172032605367701e-06,
+      "loss": 0.7172,
+      "step": 5308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3146624570109455e-06,
+      "loss": 0.6846,
+      "step": 5309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3121239339521908e-06,
+      "loss": 0.7363,
+      "step": 5310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309587692027031e-06,
+      "loss": 0.7019,
+      "step": 5311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3070537319013965e-06,
+      "loss": 0.7231,
+      "step": 5312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3045220542406178e-06,
+      "loss": 0.6519,
+      "step": 5313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3019926597094212e-06,
+      "loss": 0.7271,
+      "step": 5314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2994655489719366e-06,
+      "loss": 0.6932,
+      "step": 5315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2969407226917019e-06,
+      "loss": 0.75,
+      "step": 5316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2944181815316437e-06,
+      "loss": 0.7271,
+      "step": 5317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2918979261540954e-06,
+      "loss": 0.7469,
+      "step": 5318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2893799572207877e-06,
+      "loss": 0.6979,
+      "step": 5319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2868642753928507e-06,
+      "loss": 0.7588,
+      "step": 5320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2843508813308159e-06,
+      "loss": 0.6984,
+      "step": 5321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281839775694611e-06,
+      "loss": 0.6884,
+      "step": 5322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2793309591435676e-06,
+      "loss": 0.7107,
+      "step": 5323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2768244323364132e-06,
+      "loss": 0.741,
+      "step": 5324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2743201959312756e-06,
+      "loss": 0.7058,
+      "step": 5325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2718182505856758e-06,
+      "loss": 0.7588,
+      "step": 5326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2693185969565403e-06,
+      "loss": 0.6923,
+      "step": 5327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2668212357001898e-06,
+      "loss": 0.6739,
+      "step": 5328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.264326167472343e-06,
+      "loss": 0.6594,
+      "step": 5329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2618333929281223e-06,
+      "loss": 0.7532,
+      "step": 5330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2593429127220414e-06,
+      "loss": 0.7549,
+      "step": 5331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2568547275080135e-06,
+      "loss": 0.7108,
+      "step": 5332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2543688379393483e-06,
+      "loss": 0.7151,
+      "step": 5333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2518852446687557e-06,
+      "loss": 0.8,
+      "step": 5334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.24940394834834e-06,
+      "loss": 0.6906,
+      "step": 5335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2469249496295999e-06,
+      "loss": 0.6985,
+      "step": 5336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2444482491634413e-06,
+      "loss": 0.7013,
+      "step": 5337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2419738476001542e-06,
+      "loss": 0.63,
+      "step": 5338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.239501745589431e-06,
+      "loss": 0.6809,
+      "step": 5339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2370319437803601e-06,
+      "loss": 0.7009,
+      "step": 5340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2345644428214254e-06,
+      "loss": 0.6844,
+      "step": 5341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232099243360504e-06,
+      "loss": 0.6917,
+      "step": 5342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2296363460448758e-06,
+      "loss": 0.7327,
+      "step": 5343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2271757515212102e-06,
+      "loss": 0.6811,
+      "step": 5344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.224717460435574e-06,
+      "loss": 0.7231,
+      "step": 5345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.222261473433428e-06,
+      "loss": 0.7329,
+      "step": 5346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2198077911596284e-06,
+      "loss": 0.6227,
+      "step": 5347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2173564142584281e-06,
+      "loss": 0.7039,
+      "step": 5348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2149073433734726e-06,
+      "loss": 0.6473,
+      "step": 5349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2124605791478016e-06,
+      "loss": 0.6793,
+      "step": 5350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2100161222238516e-06,
+      "loss": 0.6597,
+      "step": 5351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.207573973243451e-06,
+      "loss": 0.7052,
+      "step": 5352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.205134132847824e-06,
+      "loss": 0.6884,
+      "step": 5353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2026966016775865e-06,
+      "loss": 0.6794,
+      "step": 5354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2002613803727492e-06,
+      "loss": 0.6841,
+      "step": 5355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.197828469572716e-06,
+      "loss": 0.2538,
+      "step": 5356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1953978699162872e-06,
+      "loss": 0.7603,
+      "step": 5357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.192969582041651e-06,
+      "loss": 0.6625,
+      "step": 5358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190543606586393e-06,
+      "loss": 0.7096,
+      "step": 5359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188119944187489e-06,
+      "loss": 0.6502,
+      "step": 5360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1856985954813093e-06,
+      "loss": 0.7021,
+      "step": 5361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1832795611036119e-06,
+      "loss": 0.7174,
+      "step": 5362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1808628416895551e-06,
+      "loss": 0.6624,
+      "step": 5363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1784484378736838e-06,
+      "loss": 0.7549,
+      "step": 5364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1760363502899353e-06,
+      "loss": 0.7153,
+      "step": 5365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1736265795716429e-06,
+      "loss": 0.6671,
+      "step": 5366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1712191263515248e-06,
+      "loss": 0.6676,
+      "step": 5367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1688139912616959e-06,
+      "loss": 0.6619,
+      "step": 5368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1664111749336616e-06,
+      "loss": 0.688,
+      "step": 5369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1640106779983152e-06,
+      "loss": 0.256,
+      "step": 5370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.161612501085947e-06,
+      "loss": 0.2896,
+      "step": 5371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1592166448262354e-06,
+      "loss": 0.6718,
+      "step": 5372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1568231098482442e-06,
+      "loss": 0.6914,
+      "step": 5373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1544318967804348e-06,
+      "loss": 0.7554,
+      "step": 5374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.152043006250656e-06,
+      "loss": 0.6548,
+      "step": 5375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1496564388861463e-06,
+      "loss": 0.7686,
+      "step": 5376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.147272195313538e-06,
+      "loss": 0.7451,
+      "step": 5377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1448902761588487e-06,
+      "loss": 0.6791,
+      "step": 5378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1425106820474875e-06,
+      "loss": 0.704,
+      "step": 5379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.140133413604252e-06,
+      "loss": 0.6904,
+      "step": 5380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1377584714533318e-06,
+      "loss": 0.7932,
+      "step": 5381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.135385856218303e-06,
+      "loss": 0.6589,
+      "step": 5382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.13301556852213e-06,
+      "loss": 0.7181,
+      "step": 5383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1306476089871732e-06,
+      "loss": 0.7319,
+      "step": 5384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1282819782351706e-06,
+      "loss": 0.7327,
+      "step": 5385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1259186768872576e-06,
+      "loss": 0.6743,
+      "step": 5386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1235577055639523e-06,
+      "loss": 0.6556,
+      "step": 5387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1211990648851657e-06,
+      "loss": 0.6805,
+      "step": 5388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.118842755470192e-06,
+      "loss": 0.7,
+      "step": 5389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164887779377198e-06,
+      "loss": 0.6853,
+      "step": 5390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1141371329058192e-06,
+      "loss": 0.7217,
+      "step": 5391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1117878209919507e-06,
+      "loss": 0.7542,
+      "step": 5392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1094408428129632e-06,
+      "loss": 0.6836,
+      "step": 5393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1070961989850882e-06,
+      "loss": 0.7014,
+      "step": 5394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1047538901239508e-06,
+      "loss": 0.7463,
+      "step": 5395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.102413916844557e-06,
+      "loss": 0.7568,
+      "step": 5396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.100076279761304e-06,
+      "loss": 0.6668,
+      "step": 5397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0977409794879733e-06,
+      "loss": 0.683,
+      "step": 5398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0954080166377334e-06,
+      "loss": 0.7194,
+      "step": 5399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0930773918231386e-06,
+      "loss": 0.6597,
+      "step": 5400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0907491056561316e-06,
+      "loss": 0.7466,
+      "step": 5401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0884231587480366e-06,
+      "loss": 0.7131,
+      "step": 5402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0860995517095663e-06,
+      "loss": 0.7157,
+      "step": 5403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0837782851508216e-06,
+      "loss": 0.7283,
+      "step": 5404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0814593596812851e-06,
+      "loss": 0.7108,
+      "step": 5405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.079142775909825e-06,
+      "loss": 0.6978,
+      "step": 5406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0768285344446971e-06,
+      "loss": 0.7698,
+      "step": 5407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0745166358935411e-06,
+      "loss": 0.6412,
+      "step": 5408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0722070808633755e-06,
+      "loss": 0.7524,
+      "step": 5409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0698998699606145e-06,
+      "loss": 0.6987,
+      "step": 5410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.067595003791051e-06,
+      "loss": 0.6423,
+      "step": 5411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0652924829598599e-06,
+      "loss": 0.7207,
+      "step": 5412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0629923080716053e-06,
+      "loss": 0.2819,
+      "step": 5413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.060694479730232e-06,
+      "loss": 0.7003,
+      "step": 5414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0583989985390696e-06,
+      "loss": 0.73,
+      "step": 5415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0561058651008316e-06,
+      "loss": 0.2512,
+      "step": 5416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0538150800176139e-06,
+      "loss": 0.6912,
+      "step": 5417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0515266438908989e-06,
+      "loss": 0.7139,
+      "step": 5418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0492405573215503e-06,
+      "loss": 0.6897,
+      "step": 5419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0469568209098146e-06,
+      "loss": 0.6764,
+      "step": 5420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.044675435255319e-06,
+      "loss": 0.6824,
+      "step": 5421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0423964009570776e-06,
+      "loss": 0.6938,
+      "step": 5422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0401197186134837e-06,
+      "loss": 0.7563,
+      "step": 5423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0378453888223173e-06,
+      "loss": 0.7223,
+      "step": 5424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0355734121807382e-06,
+      "loss": 0.7615,
+      "step": 5425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033303789285287e-06,
+      "loss": 0.7023,
+      "step": 5426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0310365207318884e-06,
+      "loss": 0.7269,
+      "step": 5427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.028771607115847e-06,
+      "loss": 0.6714,
+      "step": 5428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026509049031852e-06,
+      "loss": 0.6807,
+      "step": 5429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0242488470739686e-06,
+      "loss": 0.7347,
+      "step": 5430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0219910018356526e-06,
+      "loss": 0.2756,
+      "step": 5431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019735513909732e-06,
+      "loss": 0.7268,
+      "step": 5432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0174823838884196e-06,
+      "loss": 0.6471,
+      "step": 5433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01523161236331e-06,
+      "loss": 0.7081,
+      "step": 5434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0129831999253759e-06,
+      "loss": 0.7173,
+      "step": 5435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0107371471649708e-06,
+      "loss": 0.6643,
+      "step": 5436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0084934546718339e-06,
+      "loss": 0.6882,
+      "step": 5437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0062521230350797e-06,
+      "loss": 0.6693,
+      "step": 5438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0040131528432017e-06,
+      "loss": 0.6774,
+      "step": 5439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0017765446840777e-06,
+      "loss": 0.7173,
+      "step": 5440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.99542299144961e-07,
+      "loss": 0.7046,
+      "step": 5441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97310416812488e-07,
+      "loss": 0.734,
+      "step": 5442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.950808982726724e-07,
+      "loss": 0.7017,
+      "step": 5443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.928537441109088e-07,
+      "loss": 0.7324,
+      "step": 5444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906289549119708e-07,
+      "loss": 0.7451,
+      "step": 5445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.884065312600088e-07,
+      "loss": 0.7037,
+      "step": 5446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.86186473738555e-07,
+      "loss": 0.2812,
+      "step": 5447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.8396878293052e-07,
+      "loss": 0.7002,
+      "step": 5448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.817534594181898e-07,
+      "loss": 0.7266,
+      "step": 5449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795405037832316e-07,
+      "loss": 0.7258,
+      "step": 5450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77329916606693e-07,
+      "loss": 0.7666,
+      "step": 5451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.75121698468996e-07,
+      "loss": 0.7119,
+      "step": 5452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.729158499499403e-07,
+      "loss": 0.702,
+      "step": 5453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.70712371628707e-07,
+      "loss": 0.7334,
+      "step": 5454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.685112640838524e-07,
+      "loss": 0.6378,
+      "step": 5455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.663125278933072e-07,
+      "loss": 0.7081,
+      "step": 5456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.641161636343865e-07,
+      "loss": 0.7162,
+      "step": 5457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61922171883779e-07,
+      "loss": 0.6274,
+      "step": 5458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.597305532175495e-07,
+      "loss": 0.7264,
+      "step": 5459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.575413082111406e-07,
+      "loss": 0.7109,
+      "step": 5460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.553544374393731e-07,
+      "loss": 0.7098,
+      "step": 5461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.531699414764406e-07,
+      "loss": 0.7338,
+      "step": 5462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.509878208959178e-07,
+      "loss": 0.6854,
+      "step": 5463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.488080762707508e-07,
+      "loss": 0.7371,
+      "step": 5464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.466307081732696e-07,
+      "loss": 0.7158,
+      "step": 5465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.444557171751712e-07,
+      "loss": 0.8086,
+      "step": 5466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.422831038475355e-07,
+      "loss": 0.7327,
+      "step": 5467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.401128687608108e-07,
+      "loss": 0.7007,
+      "step": 5468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.379450124848278e-07,
+      "loss": 0.7219,
+      "step": 5469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.357795355887877e-07,
+      "loss": 0.7087,
+      "step": 5470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336164386412738e-07,
+      "loss": 0.7139,
+      "step": 5471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.314557222102372e-07,
+      "loss": 0.739,
+      "step": 5472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.292973868630062e-07,
+      "loss": 0.7119,
+      "step": 5473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.271414331662865e-07,
+      "loss": 0.6899,
+      "step": 5474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.249878616861552e-07,
+      "loss": 0.627,
+      "step": 5475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.228366729880655e-07,
+      "loss": 0.6733,
+      "step": 5476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.206878676368414e-07,
+      "loss": 0.641,
+      "step": 5477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.185414461966923e-07,
+      "loss": 0.6665,
+      "step": 5478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.163974092311867e-07,
+      "loss": 0.272,
+      "step": 5479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.142557573032751e-07,
+      "loss": 0.6918,
+      "step": 5480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.121164909752822e-07,
+      "loss": 0.7395,
+      "step": 5481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.099796108089043e-07,
+      "loss": 0.7526,
+      "step": 5482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.078451173652114e-07,
+      "loss": 0.7217,
+      "step": 5483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.057130112046453e-07,
+      "loss": 0.7148,
+      "step": 5484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.03583292887027e-07,
+      "loss": 0.8112,
+      "step": 5485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.014559629715436e-07,
+      "loss": 0.7089,
+      "step": 5486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.993310220167583e-07,
+      "loss": 0.7454,
+      "step": 5487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.972084705806072e-07,
+      "loss": 0.7239,
+      "step": 5488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.950883092203977e-07,
+      "loss": 0.7317,
+      "step": 5489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.929705384928111e-07,
+      "loss": 0.7012,
+      "step": 5490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.908551589538983e-07,
+      "loss": 0.7,
+      "step": 5491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.88742171159086e-07,
+      "loss": 0.7108,
+      "step": 5492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866315756631705e-07,
+      "loss": 0.2626,
+      "step": 5493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.845233730203217e-07,
+      "loss": 0.7042,
+      "step": 5494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824175637840782e-07,
+      "loss": 0.6956,
+      "step": 5495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.80314148507354e-07,
+      "loss": 0.7227,
+      "step": 5496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7821312774243e-07,
+      "loss": 0.662,
+      "step": 5497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.761145020409645e-07,
+      "loss": 0.7281,
+      "step": 5498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.740182719539825e-07,
+      "loss": 0.6962,
+      "step": 5499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.719244380318803e-07,
+      "loss": 0.692,
+      "step": 5500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69833000824426e-07,
+      "loss": 0.6569,
+      "step": 5501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.677439608807592e-07,
+      "loss": 0.7209,
+      "step": 5502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.656573187493855e-07,
+      "loss": 0.6877,
+      "step": 5503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.635730749781879e-07,
+      "loss": 0.7654,
+      "step": 5504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.614912301144152e-07,
+      "loss": 0.693,
+      "step": 5505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.594117847046868e-07,
+      "loss": 0.7111,
+      "step": 5506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.573347392949938e-07,
+      "loss": 0.6919,
+      "step": 5507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.552600944306955e-07,
+      "loss": 0.6971,
+      "step": 5508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.531878506565205e-07,
+      "loss": 0.7285,
+      "step": 5509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.511180085165682e-07,
+      "loss": 0.6823,
+      "step": 5510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.49050568554306e-07,
+      "loss": 0.7419,
+      "step": 5511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.469855313125752e-07,
+      "loss": 0.7222,
+      "step": 5512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.449228973335799e-07,
+      "loss": 0.7375,
+      "step": 5513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.42862667158898e-07,
+      "loss": 0.678,
+      "step": 5514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.408048413294712e-07,
+      "loss": 0.7205,
+      "step": 5515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.387494203856139e-07,
+      "loss": 0.7314,
+      "step": 5516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.366964048670056e-07,
+      "loss": 0.7734,
+      "step": 5517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34645795312703e-07,
+      "loss": 0.7476,
+      "step": 5518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.325975922611195e-07,
+      "loss": 0.7587,
+      "step": 5519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.305517962500442e-07,
+      "loss": 0.2673,
+      "step": 5520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.285084078166317e-07,
+      "loss": 0.6371,
+      "step": 5521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.264674274974038e-07,
+      "loss": 0.7052,
+      "step": 5522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.244288558282509e-07,
+      "loss": 0.7703,
+      "step": 5523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.223926933444315e-07,
+      "loss": 0.6877,
+      "step": 5524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.203589405805679e-07,
+      "loss": 0.7729,
+      "step": 5525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.183275980706595e-07,
+      "loss": 0.7308,
+      "step": 5526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.162986663480599e-07,
+      "loss": 0.7596,
+      "step": 5527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.142721459454984e-07,
+      "loss": 0.6753,
+      "step": 5528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.12248037395068e-07,
+      "loss": 0.7269,
+      "step": 5529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.102263412282297e-07,
+      "loss": 0.7434,
+      "step": 5530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.082070579758072e-07,
+      "loss": 0.7531,
+      "step": 5531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.061901881680001e-07,
+      "loss": 0.7125,
+      "step": 5532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.041757323343635e-07,
+      "loss": 0.6631,
+      "step": 5533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.021636910038255e-07,
+      "loss": 0.7085,
+      "step": 5534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.001540647046768e-07,
+      "loss": 0.7397,
+      "step": 5535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.981468539645765e-07,
+      "loss": 0.6847,
+      "step": 5536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.961420593105462e-07,
+      "loss": 0.739,
+      "step": 5537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.941396812689772e-07,
+      "loss": 0.7108,
+      "step": 5538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.921397203656233e-07,
+      "loss": 0.6326,
+      "step": 5539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.901421771256046e-07,
+      "loss": 0.7275,
+      "step": 5540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.881470520734069e-07,
+      "loss": 0.6851,
+      "step": 5541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.86154345732879e-07,
+      "loss": 0.7073,
+      "step": 5542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.841640586272381e-07,
+      "loss": 0.7299,
+      "step": 5543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.821761912790604e-07,
+      "loss": 0.3058,
+      "step": 5544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.801907442102963e-07,
+      "loss": 0.687,
+      "step": 5545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.78207717942252e-07,
+      "loss": 0.2583,
+      "step": 5546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.762271129956023e-07,
+      "loss": 0.7396,
+      "step": 5547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.742489298903832e-07,
+      "loss": 0.7213,
+      "step": 5548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.722731691459994e-07,
+      "loss": 0.6743,
+      "step": 5549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.702998312812115e-07,
+      "loss": 0.7506,
+      "step": 5550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.68328916814155e-07,
+      "loss": 0.699,
+      "step": 5551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.663604262623214e-07,
+      "loss": 0.2363,
+      "step": 5552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.643943601425674e-07,
+      "loss": 0.7317,
+      "step": 5553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.624307189711144e-07,
+      "loss": 0.7146,
+      "step": 5554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.604695032635445e-07,
+      "loss": 0.644,
+      "step": 5555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.585107135348057e-07,
+      "loss": 0.672,
+      "step": 5556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.565543502992079e-07,
+      "loss": 0.7408,
+      "step": 5557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546004140704222e-07,
+      "loss": 0.2654,
+      "step": 5558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.526489053614871e-07,
+      "loss": 0.746,
+      "step": 5559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.506998246847996e-07,
+      "loss": 0.7668,
+      "step": 5560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.487531725521224e-07,
+      "loss": 0.7217,
+      "step": 5561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.468089494745745e-07,
+      "loss": 0.7029,
+      "step": 5562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.44867155962643e-07,
+      "loss": 0.6772,
+      "step": 5563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429277925261747e-07,
+      "loss": 0.6808,
+      "step": 5564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.409908596743797e-07,
+      "loss": 0.6523,
+      "step": 5565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.390563579158294e-07,
+      "loss": 0.6847,
+      "step": 5566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.371242877584561e-07,
+      "loss": 0.6807,
+      "step": 5567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.351946497095541e-07,
+      "loss": 0.7732,
+      "step": 5568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.332674442757792e-07,
+      "loss": 0.6639,
+      "step": 5569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.313426719631478e-07,
+      "loss": 0.7375,
+      "step": 5570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.294203332770389e-07,
+      "loss": 0.6702,
+      "step": 5571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.275004287221898e-07,
+      "loss": 0.7031,
+      "step": 5572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.255829588027053e-07,
+      "loss": 0.7197,
+      "step": 5573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.236679240220412e-07,
+      "loss": 0.6826,
+      "step": 5574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.217553248830211e-07,
+      "loss": 0.7349,
+      "step": 5575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.198451618878266e-07,
+      "loss": 0.7002,
+      "step": 5576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179374355379997e-07,
+      "loss": 0.7163,
+      "step": 5577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.16032146334441e-07,
+      "loss": 0.6927,
+      "step": 5578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.14129294777417e-07,
+      "loss": 0.7634,
+      "step": 5579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.122288813665479e-07,
+      "loss": 0.7185,
+      "step": 5580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103309066008169e-07,
+      "loss": 0.6593,
+      "step": 5581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.084353709785663e-07,
+      "loss": 0.7014,
+      "step": 5582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.065422749974959e-07,
+      "loss": 0.696,
+      "step": 5583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.046516191546693e-07,
+      "loss": 0.6888,
+      "step": 5584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.027634039465037e-07,
+      "loss": 0.6904,
+      "step": 5585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.008776298687814e-07,
+      "loss": 0.7185,
+      "step": 5586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.989942974166409e-07,
+      "loss": 0.6925,
+      "step": 5587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.971134070845775e-07,
+      "loss": 0.2525,
+      "step": 5588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.952349593664487e-07,
+      "loss": 0.647,
+      "step": 5589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.933589547554709e-07,
+      "loss": 0.7078,
+      "step": 5590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.914853937442134e-07,
+      "loss": 0.741,
+      "step": 5591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.896142768246128e-07,
+      "loss": 0.7367,
+      "step": 5592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.877456044879571e-07,
+      "loss": 0.6752,
+      "step": 5593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.858793772248951e-07,
+      "loss": 0.7026,
+      "step": 5594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.840155955254313e-07,
+      "loss": 0.7019,
+      "step": 5595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.821542598789343e-07,
+      "loss": 0.6863,
+      "step": 5596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.802953707741189e-07,
+      "loss": 0.5922,
+      "step": 5597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.784389286990711e-07,
+      "loss": 0.7441,
+      "step": 5598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.765849341412245e-07,
+      "loss": 0.7046,
+      "step": 5599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.74733387587374e-07,
+      "loss": 0.6736,
+      "step": 5600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.728842895236732e-07,
+      "loss": 0.2509,
+      "step": 5601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.710376404356278e-07,
+      "loss": 0.681,
+      "step": 5602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.691934408081058e-07,
+      "loss": 0.7361,
+      "step": 5603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.673516911253286e-07,
+      "loss": 0.2911,
+      "step": 5604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.65512391870875e-07,
+      "loss": 0.7051,
+      "step": 5605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.636755435276831e-07,
+      "loss": 0.7213,
+      "step": 5606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.618411465780439e-07,
+      "loss": 0.7292,
+      "step": 5607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.600092015036075e-07,
+      "loss": 0.7231,
+      "step": 5608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.581797087853758e-07,
+      "loss": 0.6841,
+      "step": 5609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.563526689037114e-07,
+      "loss": 0.7581,
+      "step": 5610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.545280823383316e-07,
+      "loss": 0.6749,
+      "step": 5611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.527059495683086e-07,
+      "loss": 0.7,
+      "step": 5612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.508862710720731e-07,
+      "loss": 0.6879,
+      "step": 5613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.490690473274075e-07,
+      "loss": 0.6765,
+      "step": 5614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.472542788114511e-07,
+      "loss": 0.738,
+      "step": 5615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.454419660007006e-07,
+      "loss": 0.7402,
+      "step": 5616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.436321093710041e-07,
+      "loss": 0.7385,
+      "step": 5617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.418247093975682e-07,
+      "loss": 0.7073,
+      "step": 5618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.400197665549512e-07,
+      "loss": 0.7205,
+      "step": 5619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.38217281317074e-07,
+      "loss": 0.6869,
+      "step": 5620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.364172541572e-07,
+      "loss": 0.6799,
+      "step": 5621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.346196855479559e-07,
+      "loss": 0.6857,
+      "step": 5622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.328245759613206e-07,
+      "loss": 0.7463,
+      "step": 5623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.310319258686281e-07,
+      "loss": 0.7114,
+      "step": 5624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.292417357405622e-07,
+      "loss": 0.7141,
+      "step": 5625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.274540060471701e-07,
+      "loss": 0.2911,
+      "step": 5626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.256687372578452e-07,
+      "loss": 0.6836,
+      "step": 5627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.238859298413347e-07,
+      "loss": 0.6696,
+      "step": 5628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.221055842657442e-07,
+      "loss": 0.7036,
+      "step": 5629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.203277009985298e-07,
+      "loss": 0.7137,
+      "step": 5630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185522805065014e-07,
+      "loss": 0.7638,
+      "step": 5631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.167793232558228e-07,
+      "loss": 0.6827,
+      "step": 5632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.150088297120105e-07,
+      "loss": 0.6929,
+      "step": 5633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.132408003399349e-07,
+      "loss": 0.7264,
+      "step": 5634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.114752356038179e-07,
+      "loss": 0.6674,
+      "step": 5635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097121359672365e-07,
+      "loss": 0.7014,
+      "step": 5636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.079515018931192e-07,
+      "loss": 0.7599,
+      "step": 5637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.061933338437464e-07,
+      "loss": 0.7406,
+      "step": 5638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.044376322807499e-07,
+      "loss": 0.7036,
+      "step": 5639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.026843976651209e-07,
+      "loss": 0.8101,
+      "step": 5640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.009336304571933e-07,
+      "loss": 0.7108,
+      "step": 5641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.991853311166606e-07,
+      "loss": 0.7229,
+      "step": 5642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.974395001025634e-07,
+      "loss": 0.6953,
+      "step": 5643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.956961378732973e-07,
+      "loss": 0.6998,
+      "step": 5644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.939552448866049e-07,
+      "loss": 0.7045,
+      "step": 5645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922168215995883e-07,
+      "loss": 0.7415,
+      "step": 5646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.904808684686958e-07,
+      "loss": 0.7114,
+      "step": 5647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.887473859497284e-07,
+      "loss": 0.6586,
+      "step": 5648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.870163744978364e-07,
+      "loss": 0.7368,
+      "step": 5649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85287834567525e-07,
+      "loss": 0.6864,
+      "step": 5650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83561766612648e-07,
+      "loss": 0.7534,
+      "step": 5651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.818381710864096e-07,
+      "loss": 0.6764,
+      "step": 5652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.801170484413677e-07,
+      "loss": 0.7012,
+      "step": 5653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.783983991294284e-07,
+      "loss": 0.6794,
+      "step": 5654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.76682223601851e-07,
+      "loss": 0.6813,
+      "step": 5655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.749685223092405e-07,
+      "loss": 0.6975,
+      "step": 5656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.732572957015559e-07,
+      "loss": 0.6525,
+      "step": 5657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.715485442281043e-07,
+      "loss": 0.7144,
+      "step": 5658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.698422683375482e-07,
+      "loss": 0.7296,
+      "step": 5659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.681384684778935e-07,
+      "loss": 0.7133,
+      "step": 5660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.664371450964989e-07,
+      "loss": 0.748,
+      "step": 5661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.64738298640074e-07,
+      "loss": 0.724,
+      "step": 5662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.630419295546752e-07,
+      "loss": 0.6973,
+      "step": 5663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.613480382857106e-07,
+      "loss": 0.6929,
+      "step": 5664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.596566252779367e-07,
+      "loss": 0.7461,
+      "step": 5665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.579676909754583e-07,
+      "loss": 0.6558,
+      "step": 5666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.562812358217351e-07,
+      "loss": 0.7368,
+      "step": 5667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.545972602595673e-07,
+      "loss": 0.7039,
+      "step": 5668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.52915764731109e-07,
+      "loss": 0.7068,
+      "step": 5669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.512367496778637e-07,
+      "loss": 0.7134,
+      "step": 5670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.495602155406799e-07,
+      "loss": 0.6988,
+      "step": 5671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.478861627597587e-07,
+      "loss": 0.7533,
+      "step": 5672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.462145917746486e-07,
+      "loss": 0.7231,
+      "step": 5673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.445455030242453e-07,
+      "loss": 0.7357,
+      "step": 5674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.428788969467935e-07,
+      "loss": 0.6832,
+      "step": 5675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.412147739798856e-07,
+      "loss": 0.7385,
+      "step": 5676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.395531345604621e-07,
+      "loss": 0.7195,
+      "step": 5677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.378939791248128e-07,
+      "loss": 0.7053,
+      "step": 5678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.362373081085725e-07,
+      "loss": 0.6609,
+      "step": 5679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.345831219467257e-07,
+      "loss": 0.6444,
+      "step": 5680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.329314210736047e-07,
+      "loss": 0.6953,
+      "step": 5681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.312822059228884e-07,
+      "loss": 0.6975,
+      "step": 5682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.296354769276035e-07,
+      "loss": 0.6831,
+      "step": 5683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279912345201221e-07,
+      "loss": 0.7153,
+      "step": 5684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.263494791321677e-07,
+      "loss": 0.7609,
+      "step": 5685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.247102111948044e-07,
+      "loss": 0.696,
+      "step": 5686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2307343113845e-07,
+      "loss": 0.7229,
+      "step": 5687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.214391393928653e-07,
+      "loss": 0.7421,
+      "step": 5688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.198073363871581e-07,
+      "loss": 0.6854,
+      "step": 5689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181780225497834e-07,
+      "loss": 0.7208,
+      "step": 5690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.165511983085425e-07,
+      "loss": 0.7391,
+      "step": 5691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.149268640905791e-07,
+      "loss": 0.6674,
+      "step": 5692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.133050203223922e-07,
+      "loss": 0.7032,
+      "step": 5693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116856674298177e-07,
+      "loss": 0.7175,
+      "step": 5694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100688058380432e-07,
+      "loss": 0.7087,
+      "step": 5695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084544359715993e-07,
+      "loss": 0.694,
+      "step": 5696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.068425582543646e-07,
+      "loss": 0.7488,
+      "step": 5697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.052331731095595e-07,
+      "loss": 0.7053,
+      "step": 5698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.036262809597536e-07,
+      "loss": 0.6759,
+      "step": 5699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.020218822268619e-07,
+      "loss": 0.6698,
+      "step": 5700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004199773321439e-07,
+      "loss": 0.6985,
+      "step": 5701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.988205666962021e-07,
+      "loss": 0.6997,
+      "step": 5702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972236507389893e-07,
+      "loss": 0.7432,
+      "step": 5703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.956292298797949e-07,
+      "loss": 0.2819,
+      "step": 5704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.940373045372604e-07,
+      "loss": 0.7388,
+      "step": 5705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.924478751293715e-07,
+      "loss": 0.6432,
+      "step": 5706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.908609420734567e-07,
+      "loss": 0.7463,
+      "step": 5707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.892765057861882e-07,
+      "loss": 0.6898,
+      "step": 5708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.876945666835842e-07,
+      "loss": 0.6567,
+      "step": 5709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.86115125181007e-07,
+      "loss": 0.6897,
+      "step": 5710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.845381816931627e-07,
+      "loss": 0.6581,
+      "step": 5711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829637366341022e-07,
+      "loss": 0.7512,
+      "step": 5712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.813917904172183e-07,
+      "loss": 0.7228,
+      "step": 5713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.798223434552518e-07,
+      "loss": 0.7581,
+      "step": 5714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.782553961602831e-07,
+      "loss": 0.7666,
+      "step": 5715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.766909489437388e-07,
+      "loss": 0.7483,
+      "step": 5716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7512900221638793e-07,
+      "loss": 0.7041,
+      "step": 5717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.735695563883425e-07,
+      "loss": 0.6901,
+      "step": 5718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7201261186905824e-07,
+      "loss": 0.6876,
+      "step": 5719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.70458169067336e-07,
+      "loss": 0.6615,
+      "step": 5720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.689062283913193e-07,
+      "loss": 0.7556,
+      "step": 5721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.673567902484921e-07,
+      "loss": 0.7573,
+      "step": 5722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658098550456824e-07,
+      "loss": 0.7271,
+      "step": 5723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.642654231890631e-07,
+      "loss": 0.2371,
+      "step": 5724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.627234950841464e-07,
+      "loss": 0.6902,
+      "step": 5725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6118407113579066e-07,
+      "loss": 0.7081,
+      "step": 5726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5964715174819374e-07,
+      "loss": 0.7255,
+      "step": 5727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.581127373248984e-07,
+      "loss": 0.7319,
+      "step": 5728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.565808282687867e-07,
+      "loss": 0.7646,
+      "step": 5729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5505142498208475e-07,
+      "loss": 0.674,
+      "step": 5730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5352452786636246e-07,
+      "loss": 0.7515,
+      "step": 5731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.52000137322528e-07,
+      "loss": 0.7354,
+      "step": 5732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.504782537508323e-07,
+      "loss": 0.7185,
+      "step": 5733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.489588775508724e-07,
+      "loss": 0.681,
+      "step": 5734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4744200912158144e-07,
+      "loss": 0.7471,
+      "step": 5735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.459276488612374e-07,
+      "loss": 0.6951,
+      "step": 5736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.444157971674568e-07,
+      "loss": 0.6875,
+      "step": 5737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429064544372008e-07,
+      "loss": 0.734,
+      "step": 5738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.41399621066767e-07,
+      "loss": 0.7264,
+      "step": 5739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3989529745180113e-07,
+      "loss": 0.6758,
+      "step": 5740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3839348398728607e-07,
+      "loss": 0.7529,
+      "step": 5741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.368941810675431e-07,
+      "loss": 0.7017,
+      "step": 5742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3539738908623844e-07,
+      "loss": 0.7803,
+      "step": 5743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.339031084363776e-07,
+      "loss": 0.7393,
+      "step": 5744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.324113395103069e-07,
+      "loss": 0.761,
+      "step": 5745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.309220826997096e-07,
+      "loss": 0.7488,
+      "step": 5746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.294353383956185e-07,
+      "loss": 0.7126,
+      "step": 5747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.279511069883968e-07,
+      "loss": 0.6917,
+      "step": 5748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.264693888677529e-07,
+      "loss": 0.7009,
+      "step": 5749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2499018442273665e-07,
+      "loss": 0.7283,
+      "step": 5750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.235134940417307e-07,
+      "loss": 0.6757,
+      "step": 5751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.220393181124649e-07,
+      "loss": 0.6992,
+      "step": 5752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2056765702200855e-07,
+      "loss": 0.7306,
+      "step": 5753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.190985111567658e-07,
+      "loss": 0.714,
+      "step": 5754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1763188090248486e-07,
+      "loss": 0.7418,
+      "step": 5755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1616776664424987e-07,
+      "loss": 0.6895,
+      "step": 5756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1470616876648886e-07,
+      "loss": 0.6998,
+      "step": 5757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1324708765296483e-07,
+      "loss": 0.7429,
+      "step": 5758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.117905236867814e-07,
+      "loss": 0.7144,
+      "step": 5759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103364772503804e-07,
+      "loss": 0.7231,
+      "step": 5760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.088849487255486e-07,
+      "loss": 0.6709,
+      "step": 5761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0743593849340234e-07,
+      "loss": 0.7393,
+      "step": 5762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.059894469344028e-07,
+      "loss": 0.7455,
+      "step": 5763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0454547442834726e-07,
+      "loss": 0.6914,
+      "step": 5764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.031040213543735e-07,
+      "loss": 0.6788,
+      "step": 5765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.016650880909556e-07,
+      "loss": 0.7302,
+      "step": 5766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0022867501591013e-07,
+      "loss": 0.718,
+      "step": 5767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.987947825063865e-07,
+      "loss": 0.67,
+      "step": 5768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.973634109388769e-07,
+      "loss": 0.7085,
+      "step": 5769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.959345606892073e-07,
+      "loss": 0.2776,
+      "step": 5770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.945082321325466e-07,
+      "loss": 0.7654,
+      "step": 5771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.930844256433963e-07,
+      "loss": 0.7305,
+      "step": 5772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.916631415956007e-07,
+      "loss": 0.6636,
+      "step": 5773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.902443803623379e-07,
+      "loss": 0.7261,
+      "step": 5774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.888281423161266e-07,
+      "loss": 0.7013,
+      "step": 5775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8741442782882054e-07,
+      "loss": 0.7341,
+      "step": 5776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.860032372716116e-07,
+      "loss": 0.6721,
+      "step": 5777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8459457101503006e-07,
+      "loss": 0.7015,
+      "step": 5778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.831884294289423e-07,
+      "loss": 0.7329,
+      "step": 5779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8178481288255074e-07,
+      "loss": 0.7771,
+      "step": 5780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.803837217443995e-07,
+      "loss": 0.7769,
+      "step": 5781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.789851563823643e-07,
+      "loss": 0.7511,
+      "step": 5782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.775891171636592e-07,
+      "loss": 0.7123,
+      "step": 5783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.761956044548365e-07,
+      "loss": 0.7231,
+      "step": 5784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.748046186217846e-07,
+      "loss": 0.6316,
+      "step": 5785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.734161600297237e-07,
+      "loss": 0.6976,
+      "step": 5786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.720302290432198e-07,
+      "loss": 0.2693,
+      "step": 5787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7064682602616863e-07,
+      "loss": 0.7546,
+      "step": 5788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.692659513418029e-07,
+      "loss": 0.7203,
+      "step": 5789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6788760535269254e-07,
+      "loss": 0.7371,
+      "step": 5790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6651178842074365e-07,
+      "loss": 0.7126,
+      "step": 5791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6513850090719617e-07,
+      "loss": 0.6862,
+      "step": 5792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6376774317262944e-07,
+      "loss": 0.7397,
+      "step": 5793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6239951557695554e-07,
+      "loss": 0.7126,
+      "step": 5794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6103381847942375e-07,
+      "loss": 0.6882,
+      "step": 5795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.596706522386184e-07,
+      "loss": 0.752,
+      "step": 5796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5831001721246206e-07,
+      "loss": 0.6481,
+      "step": 5797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.569519137582056e-07,
+      "loss": 0.7664,
+      "step": 5798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.555963422324427e-07,
+      "loss": 0.748,
+      "step": 5799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5424330299109635e-07,
+      "loss": 0.6592,
+      "step": 5800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.528927963894302e-07,
+      "loss": 0.7178,
+      "step": 5801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.515448227820406e-07,
+      "loss": 0.7166,
+      "step": 5802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.501993825228567e-07,
+      "loss": 0.7048,
+      "step": 5803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.488564759651447e-07,
+      "loss": 0.7954,
+      "step": 5804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4751610346150487e-07,
+      "loss": 0.6227,
+      "step": 5805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4617826536387344e-07,
+      "loss": 0.6936,
+      "step": 5806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.448429620235183e-07,
+      "loss": 0.7609,
+      "step": 5807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4351019379104565e-07,
+      "loss": 0.6959,
+      "step": 5808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.421799610163945e-07,
+      "loss": 0.7256,
+      "step": 5809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.408522640488343e-07,
+      "loss": 0.6765,
+      "step": 5810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3952710323697403e-07,
+      "loss": 0.7666,
+      "step": 5811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.382044789287553e-07,
+      "loss": 0.6783,
+      "step": 5812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3688439147145036e-07,
+      "loss": 0.731,
+      "step": 5813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3556684121167195e-07,
+      "loss": 0.6941,
+      "step": 5814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.342518284953611e-07,
+      "loss": 0.7273,
+      "step": 5815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3293935366779495e-07,
+      "loss": 0.7358,
+      "step": 5816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.316294170735823e-07,
+      "loss": 0.7235,
+      "step": 5817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3032201905666807e-07,
+      "loss": 0.7639,
+      "step": 5818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2901715996032867e-07,
+      "loss": 0.7329,
+      "step": 5819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.277148401271757e-07,
+      "loss": 0.7195,
+      "step": 5820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.264150598991522e-07,
+      "loss": 0.6626,
+      "step": 5821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.251178196175353e-07,
+      "loss": 0.7634,
+      "step": 5822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2382311962293575e-07,
+      "loss": 0.6963,
+      "step": 5823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2253096025529505e-07,
+      "loss": 0.6938,
+      "step": 5824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2124134185389066e-07,
+      "loss": 0.6895,
+      "step": 5825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.199542647573317e-07,
+      "loss": 0.7617,
+      "step": 5826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1866972930355676e-07,
+      "loss": 0.6696,
+      "step": 5827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1738773582984384e-07,
+      "loss": 0.723,
+      "step": 5828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.161082846727992e-07,
+      "loss": 0.6954,
+      "step": 5829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.148313761683608e-07,
+      "loss": 0.7294,
+      "step": 5830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1355701065180045e-07,
+      "loss": 0.701,
+      "step": 5831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.122851884577238e-07,
+      "loss": 0.6978,
+      "step": 5832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.110159099200638e-07,
+      "loss": 0.658,
+      "step": 5833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0974917537209157e-07,
+      "loss": 0.7563,
+      "step": 5834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.084849851464067e-07,
+      "loss": 0.7327,
+      "step": 5835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.072233395749424e-07,
+      "loss": 0.6917,
+      "step": 5836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0596423898896056e-07,
+      "loss": 0.7598,
+      "step": 5837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.047076837190588e-07,
+      "loss": 0.6879,
+      "step": 5838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0345367409516544e-07,
+      "loss": 0.7411,
+      "step": 5839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.022022104465372e-07,
+      "loss": 0.6462,
+      "step": 5840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.009532931017667e-07,
+      "loss": 0.6978,
+      "step": 5841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.997069223887761e-07,
+      "loss": 0.7532,
+      "step": 5842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9846309863481806e-07,
+      "loss": 0.7053,
+      "step": 5843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.972218221664802e-07,
+      "loss": 0.6726,
+      "step": 5844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.959830933096741e-07,
+      "loss": 0.255,
+      "step": 5845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.947469123896496e-07,
+      "loss": 0.6669,
+      "step": 5846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.935132797309825e-07,
+      "loss": 0.6324,
+      "step": 5847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9228219565758496e-07,
+      "loss": 0.6246,
+      "step": 5848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.910536604926939e-07,
+      "loss": 0.7036,
+      "step": 5849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8982767455888237e-07,
+      "loss": 0.6346,
+      "step": 5850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.886042381780496e-07,
+      "loss": 0.7921,
+      "step": 5851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.873833516714286e-07,
+      "loss": 0.6797,
+      "step": 5852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8616501535958073e-07,
+      "loss": 0.2538,
+      "step": 5853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.849492295623979e-07,
+      "loss": 0.6737,
+      "step": 5854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837359945991047e-07,
+      "loss": 0.257,
+      "step": 5855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.825253107882553e-07,
+      "loss": 0.7664,
+      "step": 5856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.813171784477309e-07,
+      "loss": 0.6803,
+      "step": 5857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8011159789474553e-07,
+      "loss": 0.2797,
+      "step": 5858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7890856944584267e-07,
+      "loss": 0.7,
+      "step": 5859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.777080934168952e-07,
+      "loss": 0.7173,
+      "step": 5860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.765101701231077e-07,
+      "loss": 0.7324,
+      "step": 5861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.753147998790129e-07,
+      "loss": 0.7188,
+      "step": 5862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.74121982998472e-07,
+      "loss": 0.7261,
+      "step": 5863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7293171979467793e-07,
+      "loss": 0.7141,
+      "step": 5864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.717440105801539e-07,
+      "loss": 0.6941,
+      "step": 5865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705588556667482e-07,
+      "loss": 0.6877,
+      "step": 5866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6937625536564206e-07,
+      "loss": 0.7661,
+      "step": 5867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6819620998734696e-07,
+      "loss": 0.7316,
+      "step": 5868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.670187198416996e-07,
+      "loss": 0.7349,
+      "step": 5869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6584378523786816e-07,
+      "loss": 0.7322,
+      "step": 5870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.646714064843503e-07,
+      "loss": 0.7045,
+      "step": 5871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6350158388897095e-07,
+      "loss": 0.6914,
+      "step": 5872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.623343177588866e-07,
+      "loss": 0.7615,
+      "step": 5873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.611696084005777e-07,
+      "loss": 0.6979,
+      "step": 5874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6000745611985844e-07,
+      "loss": 0.6912,
+      "step": 5875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.588478612218692e-07,
+      "loss": 0.7026,
+      "step": 5876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.576908240110798e-07,
+      "loss": 0.6547,
+      "step": 5877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.565363447912872e-07,
+      "loss": 0.6921,
+      "step": 5878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.553844238656189e-07,
+      "loss": 0.6802,
+      "step": 5879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.542350615365241e-07,
+      "loss": 0.7299,
+      "step": 5880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5308825810579143e-07,
+      "loss": 0.7349,
+      "step": 5881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.519440138745288e-07,
+      "loss": 0.7104,
+      "step": 5882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508023291431749e-07,
+      "loss": 0.6372,
+      "step": 5883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.496632042114966e-07,
+      "loss": 0.7493,
+      "step": 5884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.485266393785879e-07,
+      "loss": 0.7053,
+      "step": 5885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4739263494287236e-07,
+      "loss": 0.7374,
+      "step": 5886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.462611912020996e-07,
+      "loss": 0.7427,
+      "step": 5887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4513230845334525e-07,
+      "loss": 0.691,
+      "step": 5888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4400598699301783e-07,
+      "loss": 0.7114,
+      "step": 5889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.428822271168496e-07,
+      "loss": 0.6674,
+      "step": 5890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.417610291198991e-07,
+      "loss": 0.5968,
+      "step": 5891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4064239329655514e-07,
+      "loss": 0.7014,
+      "step": 5892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.395263199405307e-07,
+      "loss": 0.6926,
+      "step": 5893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.38412809344869e-07,
+      "loss": 0.7114,
+      "step": 5894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.373018618019396e-07,
+      "loss": 0.749,
+      "step": 5895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3619347760343915e-07,
+      "loss": 0.7463,
+      "step": 5896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3508765704038928e-07,
+      "loss": 0.7776,
+      "step": 5897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3398440040313997e-07,
+      "loss": 0.6478,
+      "step": 5898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3288370798136838e-07,
+      "loss": 0.6422,
+      "step": 5899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.317855800640789e-07,
+      "loss": 0.6735,
+      "step": 5900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.306900169395998e-07,
+      "loss": 0.6927,
+      "step": 5901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2959701889558871e-07,
+      "loss": 0.7295,
+      "step": 5902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2850658621903055e-07,
+      "loss": 0.6631,
+      "step": 5903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2741871919623294e-07,
+      "loss": 0.7473,
+      "step": 5904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2633341811283183e-07,
+      "loss": 0.7,
+      "step": 5905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.252506832537904e-07,
+      "loss": 0.6215,
+      "step": 5906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2417051490339459e-07,
+      "loss": 0.7157,
+      "step": 5907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2309291334526307e-07,
+      "loss": 0.7273,
+      "step": 5908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2201787886233394e-07,
+      "loss": 0.7112,
+      "step": 5909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2094541173687478e-07,
+      "loss": 0.7144,
+      "step": 5910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.19875512250477e-07,
+      "loss": 0.7351,
+      "step": 5911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.188081806840603e-07,
+      "loss": 0.7384,
+      "step": 5912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.177434173178672e-07,
+      "loss": 0.6919,
+      "step": 5913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.166812224314685e-07,
+      "loss": 0.6523,
+      "step": 5914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.156215963037589e-07,
+      "loss": 0.7483,
+      "step": 5915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1456453921296027e-07,
+      "loss": 0.6863,
+      "step": 5916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1351005143661842e-07,
+      "loss": 0.6566,
+      "step": 5917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1245813325160514e-07,
+      "loss": 0.77,
+      "step": 5918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1140878493411844e-07,
+      "loss": 0.7134,
+      "step": 5919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.10362006759679e-07,
+      "loss": 0.7336,
+      "step": 5920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0931779900313366e-07,
+      "loss": 0.7119,
+      "step": 5921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0827616193865751e-07,
+      "loss": 0.7026,
+      "step": 5922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0723709583974626e-07,
+      "loss": 0.7449,
+      "step": 5923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0620060097922389e-07,
+      "loss": 0.762,
+      "step": 5924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0516667762923713e-07,
+      "loss": 0.7238,
+      "step": 5925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0413532606125886e-07,
+      "loss": 0.7207,
+      "step": 5926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0310654654608464e-07,
+      "loss": 0.7058,
+      "step": 5927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.020803393538373e-07,
+      "loss": 0.7124,
+      "step": 5928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0105670475396243e-07,
+      "loss": 0.7255,
+      "step": 5929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.000356430152317e-07,
+      "loss": 0.7563,
+      "step": 5930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9901715440573955e-07,
+      "loss": 0.7739,
+      "step": 5931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9800123919290537e-07,
+      "loss": 0.7515,
+      "step": 5932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.969878976434747e-07,
+      "loss": 0.7628,
+      "step": 5933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9597713002351359e-07,
+      "loss": 0.759,
+      "step": 5934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9496893659841531e-07,
+      "loss": 0.739,
+      "step": 5935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.93963317632897e-07,
+      "loss": 0.7446,
+      "step": 5936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9296027339099855e-07,
+      "loss": 0.6805,
+      "step": 5937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9195980413608483e-07,
+      "loss": 0.6425,
+      "step": 5938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9096191013084354e-07,
+      "loss": 0.6833,
+      "step": 5939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8996659163728726e-07,
+      "loss": 0.7084,
+      "step": 5940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8897384891675142e-07,
+      "loss": 0.751,
+      "step": 5941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8798368222989637e-07,
+      "loss": 0.7054,
+      "step": 5942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8699609183670531e-07,
+      "loss": 0.6472,
+      "step": 5943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.860110779964852e-07,
+      "loss": 0.6993,
+      "step": 5944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8502864096786588e-07,
+      "loss": 0.6381,
+      "step": 5945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8404878100880097e-07,
+      "loss": 0.7881,
+      "step": 5946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8307149837656914e-07,
+      "loss": 0.6721,
+      "step": 5947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820967933277684e-07,
+      "loss": 0.7581,
+      "step": 5948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8112466611832302e-07,
+      "loss": 0.2791,
+      "step": 5949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.801551170034821e-07,
+      "loss": 0.7168,
+      "step": 5950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.79188146237812e-07,
+      "loss": 0.6945,
+      "step": 5951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7822375407520742e-07,
+      "loss": 0.7412,
+      "step": 5952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7726194076888358e-07,
+      "loss": 0.7092,
+      "step": 5953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7630270657138072e-07,
+      "loss": 0.6946,
+      "step": 5954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.753460517345573e-07,
+      "loss": 0.7546,
+      "step": 5955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7439197650959917e-07,
+      "loss": 0.7041,
+      "step": 5956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7344048114701474e-07,
+      "loss": 0.7324,
+      "step": 5957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7249156589663196e-07,
+      "loss": 0.6981,
+      "step": 5958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7154523100760269e-07,
+      "loss": 0.7288,
+      "step": 5959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7060147672840144e-07,
+      "loss": 0.7393,
+      "step": 5960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.696603033068267e-07,
+      "loss": 0.7159,
+      "step": 5961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6872171098999633e-07,
+      "loss": 0.7114,
+      "step": 5962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6778570002435213e-07,
+      "loss": 0.6917,
+      "step": 5963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6685227065565858e-07,
+      "loss": 0.7324,
+      "step": 5964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6592142312900184e-07,
+      "loss": 0.6794,
+      "step": 5965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6499315768878865e-07,
+      "loss": 0.7068,
+      "step": 5966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6406747457875073e-07,
+      "loss": 0.7275,
+      "step": 5967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6314437404193917e-07,
+      "loss": 0.6477,
+      "step": 5968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6222385632072902e-07,
+      "loss": 0.2599,
+      "step": 5969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.613059216568158e-07,
+      "loss": 0.7393,
+      "step": 5970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039057029121673e-07,
+      "loss": 0.7323,
+      "step": 5971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.594778024642718e-07,
+      "loss": 0.29,
+      "step": 5972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.585676184156415e-07,
+      "loss": 0.7061,
+      "step": 5973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.57660018384308e-07,
+      "loss": 0.7241,
+      "step": 5974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5675500260857735e-07,
+      "loss": 0.6373,
+      "step": 5975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.558525713260728e-07,
+      "loss": 0.717,
+      "step": 5976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.549527247737437e-07,
+      "loss": 0.6384,
+      "step": 5977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5405546318785547e-07,
+      "loss": 0.7083,
+      "step": 5978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.531607868040008e-07,
+      "loss": 0.7231,
+      "step": 5979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.522686958570896e-07,
+      "loss": 0.6995,
+      "step": 5980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.513791905813522e-07,
+      "loss": 0.7482,
+      "step": 5981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5049227121034292e-07,
+      "loss": 0.2858,
+      "step": 5982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4960793797693663e-07,
+      "loss": 0.7129,
+      "step": 5983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4872619111332754e-07,
+      "loss": 0.7513,
+      "step": 5984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4784703085103268e-07,
+      "loss": 0.7421,
+      "step": 5985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4697045742088746e-07,
+      "loss": 0.7029,
+      "step": 5986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.460964710530499e-07,
+      "loss": 0.6842,
+      "step": 5987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4522507197699874e-07,
+      "loss": 0.6653,
+      "step": 5988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4435626042153316e-07,
+      "loss": 0.7368,
+      "step": 5989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.434900366147729e-07,
+      "loss": 0.7273,
+      "step": 5990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4262640078415713e-07,
+      "loss": 0.668,
+      "step": 5991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.417653531564478e-07,
+      "loss": 0.7111,
+      "step": 5992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4090689395772516e-07,
+      "loss": 0.6838,
+      "step": 5993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4005102341339117e-07,
+      "loss": 0.7683,
+      "step": 5994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3919774174816714e-07,
+      "loss": 0.7183,
+      "step": 5995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3834704918609497e-07,
+      "loss": 0.7286,
+      "step": 5996
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3749894595053825e-07,
+      "loss": 0.6783,
+      "step": 5997
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3665343226417772e-07,
+      "loss": 0.7112,
+      "step": 5998
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3581050834901687e-07,
+      "loss": 0.7386,
+      "step": 5999
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.349701744263776e-07,
+      "loss": 0.71,
+      "step": 6000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.341324307169034e-07,
+      "loss": 0.679,
+      "step": 6001
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3329727744055388e-07,
+      "loss": 0.6697,
+      "step": 6002
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.324647148166136e-07,
+      "loss": 0.7131,
+      "step": 6003
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.316347430636844e-07,
+      "loss": 0.7844,
+      "step": 6004
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.308073623996875e-07,
+      "loss": 0.7297,
+      "step": 6005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2998257304186468e-07,
+      "loss": 0.6628,
+      "step": 6006
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2916037520677605e-07,
+      "loss": 0.7231,
+      "step": 6007
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2834076911030225e-07,
+      "loss": 0.7207,
+      "step": 6008
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2752375496764335e-07,
+      "loss": 0.7183,
+      "step": 6009
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2670933299332e-07,
+      "loss": 0.7241,
+      "step": 6010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2589750340116892e-07,
+      "loss": 0.6801,
+      "step": 6011
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2508826640434957e-07,
+      "loss": 0.7266,
+      "step": 6012
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.242816222153398e-07,
+      "loss": 0.7332,
+      "step": 6013
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2347757104593462e-07,
+      "loss": 0.7302,
+      "step": 6014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.226761131072507e-07,
+      "loss": 0.7454,
+      "step": 6015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2187724860972417e-07,
+      "loss": 0.7313,
+      "step": 6016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2108097776310833e-07,
+      "loss": 0.7095,
+      "step": 6017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.202873007764749e-07,
+      "loss": 0.7332,
+      "step": 6018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1949621785821708e-07,
+      "loss": 0.6818,
+      "step": 6019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1870772921604657e-07,
+      "loss": 0.7041,
+      "step": 6020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1792183505698995e-07,
+      "loss": 0.6978,
+      "step": 6021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1713853558739885e-07,
+      "loss": 0.74,
+      "step": 6022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1635783101293874e-07,
+      "loss": 0.7041,
+      "step": 6023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1557972153859564e-07,
+      "loss": 0.6984,
+      "step": 6024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1480420736867392e-07,
+      "loss": 0.7262,
+      "step": 6025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1403128870679736e-07,
+      "loss": 0.677,
+      "step": 6026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1326096575590695e-07,
+      "loss": 0.7415,
+      "step": 6027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1249323871826201e-07,
+      "loss": 0.7327,
+      "step": 6028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1172810779544129e-07,
+      "loss": 0.7642,
+      "step": 6029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.10965573188343e-07,
+      "loss": 0.7009,
+      "step": 6030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.102056350971803e-07,
+      "loss": 0.6851,
+      "step": 6031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0944829372148691e-07,
+      "loss": 0.7136,
+      "step": 6032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0869354926011488e-07,
+      "loss": 0.7031,
+      "step": 6033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0794140191123348e-07,
+      "loss": 0.7207,
+      "step": 6034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0719185187232916e-07,
+      "loss": 0.7041,
+      "step": 6035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0644489934021007e-07,
+      "loss": 0.7294,
+      "step": 6036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0570054451099821e-07,
+      "loss": 0.7268,
+      "step": 6037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0495878758013611e-07,
+      "loss": 0.7546,
+      "step": 6038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0421962874238245e-07,
+      "loss": 0.6829,
+      "step": 6039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.034830681918153e-07,
+      "loss": 0.7178,
+      "step": 6040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0274910612182887e-07,
+      "loss": 0.6625,
+      "step": 6041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0201774272513676e-07,
+      "loss": 0.7732,
+      "step": 6042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0128897819376981e-07,
+      "loss": 0.7373,
+      "step": 6043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0056281271907609e-07,
+      "loss": 0.6954,
+      "step": 6044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.983924649172083e-08,
+      "loss": 0.2682,
+      "step": 6045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.911827970168763e-08,
+      "loss": 0.7125,
+      "step": 6046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.83999125382773e-08,
+      "loss": 0.7211,
+      "step": 6047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.768414519010782e-08,
+      "loss": 0.7316,
+      "step": 6048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.697097784511444e-08,
+      "loss": 0.6833,
+      "step": 6049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.62604106905507e-08,
+      "loss": 0.2874,
+      "step": 6050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.555244391298735e-08,
+      "loss": 0.7019,
+      "step": 6051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.484707769831014e-08,
+      "loss": 0.7526,
+      "step": 6052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.414431223172538e-08,
+      "loss": 0.703,
+      "step": 6053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.344414769775323e-08,
+      "loss": 0.7561,
+      "step": 6054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.274658428023331e-08,
+      "loss": 0.6952,
+      "step": 6055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.205162216232133e-08,
+      "loss": 0.728,
+      "step": 6056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.135926152649133e-08,
+      "loss": 0.7377,
+      "step": 6057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.066950255453011e-08,
+      "loss": 0.2706,
+      "step": 6058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.998234542754725e-08,
+      "loss": 0.7402,
+      "step": 6059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.929779032596619e-08,
+      "loss": 0.7576,
+      "step": 6060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.861583742952429e-08,
+      "loss": 0.7224,
+      "step": 6061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.793648691728163e-08,
+      "loss": 0.6252,
+      "step": 6062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.725973896761108e-08,
+      "loss": 0.7328,
+      "step": 6063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.658559375820275e-08,
+      "loss": 0.724,
+      "step": 6064
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.59140514660639e-08,
+      "loss": 0.2652,
+      "step": 6065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.524511226751797e-08,
+      "loss": 0.7354,
+      "step": 6066
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.457877633820555e-08,
+      "loss": 0.7205,
+      "step": 6067
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.391504385308336e-08,
+      "loss": 0.7292,
+      "step": 6068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.325391498642421e-08,
+      "loss": 0.7856,
+      "step": 6069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.259538991181703e-08,
+      "loss": 0.2751,
+      "step": 6070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.193946880216685e-08,
+      "loss": 0.7434,
+      "step": 6071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.128615182969812e-08,
+      "loss": 0.7637,
+      "step": 6072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.063543916594696e-08,
+      "loss": 0.8303,
+      "step": 6073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.998733098176892e-08,
+      "loss": 0.686,
+      "step": 6074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.934182744733454e-08,
+      "loss": 0.7559,
+      "step": 6075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.869892873212937e-08,
+      "loss": 0.7305,
+      "step": 6076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.805863500495724e-08,
+      "loss": 0.7656,
+      "step": 6077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.742094643393816e-08,
+      "loss": 0.7141,
+      "step": 6078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.678586318650483e-08,
+      "loss": 0.2737,
+      "step": 6079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.615338542940941e-08,
+      "loss": 0.7256,
+      "step": 6080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.552351332871799e-08,
+      "loss": 0.74,
+      "step": 6081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.489624704981158e-08,
+      "loss": 0.6714,
+      "step": 6082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.427158675739177e-08,
+      "loss": 0.7163,
+      "step": 6083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.364953261546958e-08,
+      "loss": 0.7214,
+      "step": 6084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.303008478737662e-08,
+      "loss": 0.7095,
+      "step": 6085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.241324343575718e-08,
+      "loss": 0.7021,
+      "step": 6086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.179900872257283e-08,
+      "loss": 0.6946,
+      "step": 6087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.118738080909904e-08,
+      "loss": 0.7261,
+      "step": 6088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.057835985593065e-08,
+      "loss": 0.7529,
+      "step": 6089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.997194602297197e-08,
+      "loss": 0.7094,
+      "step": 6090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.936813946944898e-08,
+      "loss": 0.6992,
+      "step": 6091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.876694035390041e-08,
+      "loss": 0.6787,
+      "step": 6092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.816834883417778e-08,
+      "loss": 0.6901,
+      "step": 6093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.75723650674509e-08,
+      "loss": 0.6895,
+      "step": 6094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.697898921020574e-08,
+      "loss": 0.6982,
+      "step": 6095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.638822141824097e-08,
+      "loss": 0.6855,
+      "step": 6096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.580006184667254e-08,
+      "loss": 0.6526,
+      "step": 6097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.521451064993023e-08,
+      "loss": 0.7222,
+      "step": 6098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.463156798175884e-08,
+      "loss": 0.7495,
+      "step": 6099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.40512339952204e-08,
+      "loss": 0.658,
+      "step": 6100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.347350884268855e-08,
+      "loss": 0.7749,
+      "step": 6101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.289839267585418e-08,
+      "loss": 0.658,
+      "step": 6102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.232588564572317e-08,
+      "loss": 0.6549,
+      "step": 6103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.17559879026164e-08,
+      "loss": 0.7338,
+      "step": 6104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.118869959616858e-08,
+      "loss": 0.6859,
+      "step": 6105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.062402087532948e-08,
+      "loss": 0.6786,
+      "step": 6106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.006195188836495e-08,
+      "loss": 0.7112,
+      "step": 6107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.950249278285358e-08,
+      "loss": 0.7212,
+      "step": 6108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8945643705691225e-08,
+      "loss": 0.6389,
+      "step": 6109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8391404803085364e-08,
+      "loss": 0.7476,
+      "step": 6110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.78397762205618e-08,
+      "loss": 0.7129,
+      "step": 6111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7290758102958035e-08,
+      "loss": 0.6959,
+      "step": 6112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6744350594426514e-08,
+      "loss": 0.6901,
+      "step": 6113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.620055383843581e-08,
+      "loss": 0.7212,
+      "step": 6114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5659367977768383e-08,
+      "loss": 0.7935,
+      "step": 6115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.512079315451835e-08,
+      "loss": 0.6584,
+      "step": 6116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4584829510099246e-08,
+      "loss": 0.5712,
+      "step": 6117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.405147718523518e-08,
+      "loss": 0.6523,
+      "step": 6118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.352073631996635e-08,
+      "loss": 0.7043,
+      "step": 6119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.299260705364684e-08,
+      "loss": 0.7479,
+      "step": 6120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2467089524944614e-08,
+      "loss": 0.6968,
+      "step": 6121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1944183871842635e-08,
+      "loss": 0.6966,
+      "step": 6122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1423890231636633e-08,
+      "loss": 0.6887,
+      "step": 6123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.090620874093954e-08,
+      "loss": 0.7051,
+      "step": 6124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.039113953567598e-08,
+      "loss": 0.6725,
+      "step": 6125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.987868275108443e-08,
+      "loss": 0.6953,
+      "step": 6126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.936883852171837e-08,
+      "loss": 0.7275,
+      "step": 6127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8861606981445156e-08,
+      "loss": 0.686,
+      "step": 6128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8356988263444925e-08,
+      "loss": 0.7589,
+      "step": 6129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.785498250021503e-08,
+      "loss": 0.7759,
+      "step": 6130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.735558982356447e-08,
+      "loss": 0.7279,
+      "step": 6131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.685881036461615e-08,
+      "loss": 0.6796,
+      "step": 6132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.636464425380571e-08,
+      "loss": 0.7192,
+      "step": 6133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5873091620884933e-08,
+      "loss": 0.6772,
+      "step": 6134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.538415259491946e-08,
+      "loss": 0.7125,
+      "step": 6135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.489782730428549e-08,
+      "loss": 0.7032,
+      "step": 6136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.441411587667643e-08,
+      "loss": 0.6765,
+      "step": 6137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3933018439097366e-08,
+      "loss": 0.678,
+      "step": 6138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3454535117869454e-08,
+      "loss": 0.7312,
+      "step": 6139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.297866603862333e-08,
+      "loss": 0.7856,
+      "step": 6140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.250541132630792e-08,
+      "loss": 0.7351,
+      "step": 6141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2034771105181615e-08,
+      "loss": 0.761,
+      "step": 6142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.156674549881889e-08,
+      "loss": 0.6855,
+      "step": 6143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1101334630107015e-08,
+      "loss": 0.8192,
+      "step": 6144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0638538621246e-08,
+      "loss": 0.6639,
+      "step": 6145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.017835759375088e-08,
+      "loss": 0.7134,
+      "step": 6146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.972079166844833e-08,
+      "loss": 0.7185,
+      "step": 6147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.926584096548114e-08,
+      "loss": 0.7246,
+      "step": 6148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8813505604300413e-08,
+      "loss": 0.6882,
+      "step": 6149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.836378570367672e-08,
+      "loss": 0.6794,
+      "step": 6150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7916681381687805e-08,
+      "loss": 0.6379,
+      "step": 6151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7472192755730883e-08,
+      "loss": 0.6953,
+      "step": 6152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.703031994251038e-08,
+      "loss": 0.2822,
+      "step": 6153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.659106305804905e-08,
+      "loss": 0.6931,
+      "step": 6154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6154422217677954e-08,
+      "loss": 0.7074,
+      "step": 6155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5720397536046504e-08,
+      "loss": 0.6594,
+      "step": 6156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5288989127111315e-08,
+      "loss": 0.7786,
+      "step": 6157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4860197104148454e-08,
+      "loss": 0.7212,
+      "step": 6158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4434021579742297e-08,
+      "loss": 0.7188,
+      "step": 6159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.401046266579111e-08,
+      "loss": 0.7156,
+      "step": 6160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.358952047350705e-08,
+      "loss": 0.7375,
+      "step": 6161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3171195113416156e-08,
+      "loss": 0.7197,
+      "step": 6162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2755486695354996e-08,
+      "loss": 0.712,
+      "step": 6163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.234239532847294e-08,
+      "loss": 0.738,
+      "step": 6164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1931921121235444e-08,
+      "loss": 0.7368,
+      "step": 6165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.15240641814174e-08,
+      "loss": 0.7495,
+      "step": 6166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.111882461610982e-08,
+      "loss": 0.7166,
+      "step": 6167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0716202531710924e-08,
+      "loss": 0.6797,
+      "step": 6168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.031619803393837e-08,
+      "loss": 0.7151,
+      "step": 6169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.991881122781815e-08,
+      "loss": 0.6899,
+      "step": 6170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.952404221769123e-08,
+      "loss": 0.7129,
+      "step": 6171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9131891107209153e-08,
+      "loss": 0.713,
+      "step": 6172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8742357999336222e-08,
+      "loss": 0.6989,
+      "step": 6173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8355442996352845e-08,
+      "loss": 0.6769,
+      "step": 6174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7971146199847756e-08,
+      "loss": 0.7104,
+      "step": 6175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.758946771072357e-08,
+      "loss": 0.7098,
+      "step": 6176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7210407629195688e-08,
+      "loss": 0.7754,
+      "step": 6177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.683396605479338e-08,
+      "loss": 0.7424,
+      "step": 6178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6460143086357582e-08,
+      "loss": 0.7122,
+      "step": 6179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.608893882203867e-08,
+      "loss": 0.6655,
+      "step": 6180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.572035335930312e-08,
+      "loss": 0.6934,
+      "step": 6181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5354386794930184e-08,
+      "loss": 0.7158,
+      "step": 6182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.499103922500745e-08,
+      "loss": 0.6626,
+      "step": 6183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4630310744938603e-08,
+      "loss": 0.6865,
+      "step": 6184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4272201449438982e-08,
+      "loss": 0.6772,
+      "step": 6185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3916711432534488e-08,
+      "loss": 0.7358,
+      "step": 6186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3563840787564905e-08,
+      "loss": 0.6906,
+      "step": 6187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3213589607180563e-08,
+      "loss": 0.6586,
+      "step": 6188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2865957983347896e-08,
+      "loss": 0.7686,
+      "step": 6189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.252094600734056e-08,
+      "loss": 0.6693,
+      "step": 6190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2178553769747202e-08,
+      "loss": 0.6742,
+      "step": 6191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.183878136046813e-08,
+      "loss": 0.7297,
+      "step": 6192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1501628868716428e-08,
+      "loss": 0.6631,
+      "step": 6193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.116709638301573e-08,
+      "loss": 0.7349,
+      "step": 6194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0835183991203546e-08,
+      "loss": 0.6802,
+      "step": 6195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0505891780427944e-08,
+      "loss": 0.6619,
+      "step": 6196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0179219837148657e-08,
+      "loss": 0.6748,
+      "step": 6197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9855168247139288e-08,
+      "loss": 0.7081,
+      "step": 6198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9533737095485118e-08,
+      "loss": 0.6974,
+      "step": 6199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.921492646658196e-08,
+      "loss": 0.751,
+      "step": 6200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8898736444138422e-08,
+      "loss": 0.7164,
+      "step": 6201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858516711117586e-08,
+      "loss": 0.7183,
+      "step": 6202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.827421855002509e-08,
+      "loss": 0.6854,
+      "step": 6203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.79658908423308e-08,
+      "loss": 0.7382,
+      "step": 6204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7660184069050456e-08,
+      "loss": 0.7604,
+      "step": 6205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.735709831044985e-08,
+      "loss": 0.6855,
+      "step": 6206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7056633646110876e-08,
+      "loss": 0.6691,
+      "step": 6207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6758790154923766e-08,
+      "loss": 0.692,
+      "step": 6208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6463567915093737e-08,
+      "loss": 0.6814,
+      "step": 6209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.617096700413212e-08,
+      "loss": 0.6814,
+      "step": 6210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5880987498868572e-08,
+      "loss": 0.7059,
+      "step": 6211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5593629475442185e-08,
+      "loss": 0.6631,
+      "step": 6212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5308893009300386e-08,
+      "loss": 0.6802,
+      "step": 6213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5026778175207814e-08,
+      "loss": 0.6768,
+      "step": 6214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4747285047236325e-08,
+      "loss": 0.7371,
+      "step": 6215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4470413698771668e-08,
+      "loss": 0.7036,
+      "step": 6216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4196164202510132e-08,
+      "loss": 0.7412,
+      "step": 6217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3924536630460784e-08,
+      "loss": 0.7245,
+      "step": 6218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3655531053943238e-08,
+      "loss": 0.276,
+      "step": 6219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3389147543588776e-08,
+      "loss": 0.7273,
+      "step": 6220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3125386169341448e-08,
+      "loss": 0.2623,
+      "step": 6221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2864247000453634e-08,
+      "loss": 0.7148,
+      "step": 6222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2605730105493818e-08,
+      "loss": 0.7358,
+      "step": 6223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2349835552337708e-08,
+      "loss": 0.6893,
+      "step": 6224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2096563408176e-08,
+      "loss": 0.6719,
+      "step": 6225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1845913739507719e-08,
+      "loss": 0.6741,
+      "step": 6226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1597886612144672e-08,
+      "loss": 0.7135,
+      "step": 6227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1352482091211426e-08,
+      "loss": 0.7156,
+      "step": 6228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1109700241141997e-08,
+      "loss": 0.7112,
+      "step": 6229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0869541125682059e-08,
+      "loss": 0.6987,
+      "step": 6230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0632004807888951e-08,
+      "loss": 0.679,
+      "step": 6231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0397091350131671e-08,
+      "loss": 0.7292,
+      "step": 6232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0164800814090881e-08,
+      "loss": 0.7041,
+      "step": 6233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.93513326075779e-09,
+      "loss": 0.7048,
+      "step": 6234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.708088750434163e-09,
+      "loss": 0.6595,
+      "step": 6235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.483667342735647e-09,
+      "loss": 0.7018,
+      "step": 6236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.261869096586217e-09,
+      "loss": 0.7024,
+      "step": 6237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.042694070221513e-09,
+      "loss": 0.7595,
+      "step": 6238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.826142321192165e-09,
+      "loss": 0.7529,
+      "step": 6239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.612213906353806e-09,
+      "loss": 0.6523,
+      "step": 6240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.400908881879278e-09,
+      "loss": 0.7101,
+      "step": 6241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.192227303247535e-09,
+      "loss": 0.7164,
+      "step": 6242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.986169225253638e-09,
+      "loss": 0.6792,
+      "step": 6243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.78273470199986e-09,
+      "loss": 0.7134,
+      "step": 6244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.581923786900148e-09,
+      "loss": 0.6571,
+      "step": 6245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.383736532682318e-09,
+      "loss": 0.7134,
+      "step": 6246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.188172991381415e-09,
+      "loss": 0.6787,
+      "step": 6247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.995233214347474e-09,
+      "loss": 0.6552,
+      "step": 6248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.804917252239973e-09,
+      "loss": 0.7153,
+      "step": 6249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.617225155026718e-09,
+      "loss": 0.6947,
+      "step": 6250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.432156971991621e-09,
+      "loss": 0.6707,
+      "step": 6251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.249712751725811e-09,
+      "loss": 0.6749,
+      "step": 6252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0698925421331935e-09,
+      "loss": 0.6992,
+      "step": 6253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.892696390428221e-09,
+      "loss": 0.6359,
+      "step": 6254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.71812434313701e-09,
+      "loss": 0.6993,
+      "step": 6255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.54617644609512e-09,
+      "loss": 0.7317,
+      "step": 6256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.376852744450878e-09,
+      "loss": 0.7361,
+      "step": 6257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.210153282662056e-09,
+      "loss": 0.6804,
+      "step": 6258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.046078104499197e-09,
+      "loss": 0.7141,
+      "step": 6259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.884627253041174e-09,
+      "loss": 0.7307,
+      "step": 6260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.725800770680744e-09,
+      "loss": 0.8044,
+      "step": 6261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.569598699120103e-09,
+      "loss": 0.6935,
+      "step": 6262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.416021079370891e-09,
+      "loss": 0.6987,
+      "step": 6263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.265067951759738e-09,
+      "loss": 0.645,
+      "step": 6264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.116739355919386e-09,
+      "loss": 0.6987,
+      "step": 6265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.971035330797568e-09,
+      "loss": 0.7529,
+      "step": 6266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.827955914649239e-09,
+      "loss": 0.6812,
+      "step": 6267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6875011450443476e-09,
+      "loss": 0.7939,
+      "step": 6268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5496710588589502e-09,
+      "loss": 0.7125,
+      "step": 6269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.414465692284097e-09,
+      "loss": 0.7225,
+      "step": 6270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2818850808191695e-09,
+      "loss": 0.6843,
+      "step": 6271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1519292592763207e-09,
+      "loss": 0.7551,
+      "step": 6272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.024598261776035e-09,
+      "loss": 0.2465,
+      "step": 6273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8998921217515685e-09,
+      "loss": 0.6602,
+      "step": 6274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.777810871947839e-09,
+      "loss": 0.7454,
+      "step": 6275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6583545444158756e-09,
+      "loss": 0.7455,
+      "step": 6276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5415231705239184e-09,
+      "loss": 0.6804,
+      "step": 6277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4273167809463207e-09,
+      "loss": 0.674,
+      "step": 6278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.315735405670205e-09,
+      "loss": 0.7002,
+      "step": 6279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.206779073992138e-09,
+      "loss": 0.7288,
+      "step": 6280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1004478145203457e-09,
+      "loss": 0.7429,
+      "step": 6281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9967416551758267e-09,
+      "loss": 0.6963,
+      "step": 6282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8956606231845808e-09,
+      "loss": 0.7296,
+      "step": 6283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.79720474508982e-09,
+      "loss": 0.6934,
+      "step": 6284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.701374046741977e-09,
+      "loss": 0.7047,
+      "step": 6285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6081685533031466e-09,
+      "loss": 0.7532,
+      "step": 6286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5175882892437544e-09,
+      "loss": 0.694,
+      "step": 6287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4296332783503286e-09,
+      "loss": 0.6936,
+      "step": 6288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3443035437132879e-09,
+      "loss": 0.7495,
+      "step": 6289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2615991077402634e-09,
+      "loss": 0.6727,
+      "step": 6290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1815199921449972e-09,
+      "loss": 0.7333,
+      "step": 6291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1040662179540029e-09,
+      "loss": 0.7432,
+      "step": 6292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.029237805503236e-09,
+      "loss": 0.6414,
+      "step": 6293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.57034774440313e-10,
+      "loss": 0.7391,
+      "step": 6294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.874571437234025e-10,
+      "loss": 0.7236,
+      "step": 6295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.205049316223346e-10,
+      "loss": 0.7067,
+      "step": 6296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.5617815571416e-10,
+      "loss": 0.6622,
+      "step": 6297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.944768328909224e-10,
+      "loss": 0.6752,
+      "step": 6298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.354009793507754e-10,
+      "loss": 0.6951,
+      "step": 6299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.789506106079757e-10,
+      "loss": 0.6921,
+      "step": 6300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.251257414828903e-10,
+      "loss": 0.6791,
+      "step": 6301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.739263861075483e-10,
+      "loss": 0.7623,
+      "step": 6302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.253525579256401e-10,
+      "loss": 0.6903,
+      "step": 6303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7940426969140797e-10,
+      "loss": 0.66,
+      "step": 6304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3608153346853523e-10,
+      "loss": 0.7336,
+      "step": 6305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9538436063236697e-10,
+      "loss": 0.7314,
+      "step": 6306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5731276186879984e-10,
+      "loss": 0.7224,
+      "step": 6307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.218667471731717e-10,
+      "loss": 0.7366,
+      "step": 6308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8904632585470262e-10,
+      "loss": 0.6956,
+      "step": 6309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5885150652761305e-10,
+      "loss": 0.7126,
+      "step": 6310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3128229712222605e-10,
+      "loss": 0.7449,
+      "step": 6311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0633870487719577e-10,
+      "loss": 0.7024,
+      "step": 6312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.402073634172781e-11,
+      "loss": 0.6858,
+      "step": 6313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.432839737446906e-11,
+      "loss": 0.7354,
+      "step": 6314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.726169314683837e-11,
+      "loss": 0.6555,
+      "step": 6315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.282062814080611e-11,
+      "loss": 0.739,
+      "step": 6316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1005206146673675e-11,
+      "loss": 0.7156,
+      "step": 6317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1815430266404193e-11,
+      "loss": 0.713,
+      "step": 6318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2513029158429706e-12,
+      "loss": 0.7407,
+      "step": 6319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3128258147254713e-12,
+      "loss": 0.4417,
+      "step": 6320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5141,
+      "step": 6321
+    },
+    {
+      "epoch": 1.0,
+      "step": 6321,
+      "total_flos": 4.058200602525355e+19,
+      "train_loss": 0.7335893932427622,
+      "train_runtime": 69735.416,
+      "train_samples_per_second": 11.602,
+      "train_steps_per_second": 0.091
+    }
+  ],
+  "max_steps": 6321,
+  "num_train_epochs": 1,
+  "total_flos": 4.058200602525355e+19,
+  "trial_name": null,
+  "trial_params": null
+}