diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,22705 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998677423621214,
+  "global_step": 3780,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 1.3385,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 1.29,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.2743,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 1.3181,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.771929824561403e-06,
+      "loss": 1.2876,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 1.2442,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2280701754385964e-05,
+      "loss": 1.2495,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.2654,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5789473684210526e-05,
+      "loss": 1.2151,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7543859649122806e-05,
+      "loss": 1.2537,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 1.249,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 1.1509,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.280701754385965e-05,
+      "loss": 1.191,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.456140350877193e-05,
+      "loss": 1.1221,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 1.1223,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 1.1213,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9824561403508772e-05,
+      "loss": 1.1291,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 1.1522,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.0844,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 1.0426,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 1.1298,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 1.084,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0350877192982455e-05,
+      "loss": 1.0996,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 1.0359,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3859649122807014e-05,
+      "loss": 1.0964,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.56140350877193e-05,
+      "loss": 1.0368,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 1.1031,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.912280701754386e-05,
+      "loss": 1.0223,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.087719298245615e-05,
+      "loss": 1.0264,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2631578947368424e-05,
+      "loss": 1.0703,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.438596491228071e-05,
+      "loss": 1.0152,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6140350877192984e-05,
+      "loss": 1.0979,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.789473684210527e-05,
+      "loss": 1.0037,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9649122807017544e-05,
+      "loss": 1.0791,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.140350877192983e-05,
+      "loss": 0.974,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.31578947368421e-05,
+      "loss": 1.0512,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.49122807017544e-05,
+      "loss": 0.3297,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.0101,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.842105263157895e-05,
+      "loss": 0.9863,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.017543859649122e-05,
+      "loss": 1.0068,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.192982456140351e-05,
+      "loss": 1.0117,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.368421052631579e-05,
+      "loss": 0.9474,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.543859649122808e-05,
+      "loss": 1.0245,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.719298245614036e-05,
+      "loss": 1.0086,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.894736842105263e-05,
+      "loss": 0.2937,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.070175438596491e-05,
+      "loss": 0.9911,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.24561403508772e-05,
+      "loss": 1.0161,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.421052631578948e-05,
+      "loss": 0.9596,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.596491228070177e-05,
+      "loss": 0.989,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.771929824561403e-05,
+      "loss": 0.9973,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.947368421052632e-05,
+      "loss": 0.2977,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.12280701754386e-05,
+      "loss": 1.0062,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.298245614035089e-05,
+      "loss": 1.0154,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.473684210526316e-05,
+      "loss": 1.0024,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.649122807017544e-05,
+      "loss": 0.9711,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.824561403508771e-05,
+      "loss": 0.9982,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.9833,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001017543859649123,
+      "loss": 0.9571,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010350877192982457,
+      "loss": 0.303,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010526315789473685,
+      "loss": 0.94,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010701754385964912,
+      "loss": 0.9525,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010877192982456141,
+      "loss": 0.95,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001105263157894737,
+      "loss": 0.9821,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011228070175438597,
+      "loss": 0.939,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011403508771929824,
+      "loss": 1.008,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011578947368421053,
+      "loss": 0.9647,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011754385964912282,
+      "loss": 0.938,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011929824561403509,
+      "loss": 0.9264,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012105263157894738,
+      "loss": 1.0119,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012280701754385965,
+      "loss": 0.9364,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012456140350877194,
+      "loss": 0.9787,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001263157894736842,
+      "loss": 0.9383,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001280701754385965,
+      "loss": 0.9527,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001298245614035088,
+      "loss": 0.9709,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013157894736842108,
+      "loss": 0.9536,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.9213,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013508771929824563,
+      "loss": 0.9183,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001368421052631579,
+      "loss": 0.9758,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013859649122807018,
+      "loss": 0.9445,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014035087719298245,
+      "loss": 0.9956,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014210526315789474,
+      "loss": 0.2735,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014385964912280703,
+      "loss": 0.9596,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014561403508771932,
+      "loss": 0.9325,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014736842105263158,
+      "loss": 0.9479,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014912280701754387,
+      "loss": 0.9281,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015087719298245616,
+      "loss": 0.9077,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015263157894736845,
+      "loss": 0.9294,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001543859649122807,
+      "loss": 0.3024,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015614035087719297,
+      "loss": 0.9528,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 0.974,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015964912280701756,
+      "loss": 0.988,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016140350877192982,
+      "loss": 0.9337,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001631578947368421,
+      "loss": 0.9538,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001649122807017544,
+      "loss": 0.942,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.9634,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016842105263157895,
+      "loss": 0.9299,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017017543859649124,
+      "loss": 0.9796,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017192982456140353,
+      "loss": 0.9038,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001736842105263158,
+      "loss": 0.9829,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017543859649122806,
+      "loss": 1.0089,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017719298245614035,
+      "loss": 0.9281,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017894736842105264,
+      "loss": 0.9549,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018070175438596493,
+      "loss": 0.942,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001824561403508772,
+      "loss": 0.984,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018421052631578948,
+      "loss": 0.9338,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018596491228070177,
+      "loss": 0.9625,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018771929824561406,
+      "loss": 0.2818,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018947368421052632,
+      "loss": 0.9732,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001912280701754386,
+      "loss": 0.9239,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019298245614035088,
+      "loss": 0.8975,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019473684210526317,
+      "loss": 0.9337,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019649122807017543,
+      "loss": 0.9485,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019824561403508772,
+      "loss": 0.9234,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9503,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999996328151095,
+      "loss": 0.9645,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999985312607076,
+      "loss": 0.9145,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999966953376032,
+      "loss": 0.9383,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999994125047145,
+      "loss": 0.2892,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999908203912194,
+      "loss": 1.0131,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999867813722544,
+      "loss": 0.956,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999820079932157,
+      "loss": 0.9454,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999765002576085,
+      "loss": 0.9676,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999702581694783,
+      "loss": 0.9365,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999963281733408,
+      "loss": 0.9374,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999555709545217,
+      "loss": 0.9163,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999471258384815,
+      "loss": 0.945,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999379463914898,
+      "loss": 0.9737,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999280326202872,
+      "loss": 0.9496,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999173845321538,
+      "loss": 0.9717,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999060021349104,
+      "loss": 0.9189,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998938854369147,
+      "loss": 0.9308,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999881034447065,
+      "loss": 0.9242,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998674491747995,
+      "loss": 0.8892,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998531296300942,
+      "loss": 0.9552,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998380758234647,
+      "loss": 0.9417,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998222877659668,
+      "loss": 0.9045,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998057654691938,
+      "loss": 0.9283,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199978850894528,
+      "loss": 0.9601,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997705182068977,
+      "loss": 0.9225,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999751793267259,
+      "loss": 0.881,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999732334140115,
+      "loss": 0.3218,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997121408397552,
+      "loss": 0.9295,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996912133810098,
+      "loss": 1.0057,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996695517792467,
+      "loss": 0.9453,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996471560503737,
+      "loss": 0.2986,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996240262108378,
+      "loss": 0.317,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996001622776244,
+      "loss": 0.9222,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999575564268259,
+      "loss": 0.918,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995502322008051,
+      "loss": 0.9436,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999524166093866,
+      "loss": 0.9546,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994973659665842,
+      "loss": 0.9013,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994698318386402,
+      "loss": 0.9453,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994415637302547,
+      "loss": 0.9076,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999412561662187,
+      "loss": 0.9122,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993828256557351,
+      "loss": 0.9304,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993523557327366,
+      "loss": 0.9237,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999321151915567,
+      "loss": 0.9474,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999289214227142,
+      "loss": 0.8891,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999256542690916,
+      "loss": 0.9526,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992231373308812,
+      "loss": 0.9113,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991889981715698,
+      "loss": 0.9297,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991541252380526,
+      "loss": 0.9153,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991185185559392,
+      "loss": 0.9741,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990821781513783,
+      "loss": 0.927,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990451040510565,
+      "loss": 0.946,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990072962822007,
+      "loss": 0.9619,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989687548725756,
+      "loss": 0.9314,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989294798504844,
+      "loss": 0.9239,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199888947124477,
+      "loss": 0.9048,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988487290848134,
+      "loss": 0.9214,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988072534005342,
+      "loss": 0.9643,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998765044222391,
+      "loss": 0.8936,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998722101581381,
+      "loss": 0.9075,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986784255090397,
+      "loss": 0.9136,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998634016037442,
+      "loss": 0.8762,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985888731992004,
+      "loss": 0.9234,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985429970274668,
+      "loss": 0.9263,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984963875559311,
+      "loss": 0.8847,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984490448188218,
+      "loss": 0.3004,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998400968850906,
+      "loss": 0.8988,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983521596874895,
+      "loss": 0.941,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998302617364416,
+      "loss": 0.9059,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998252341918068,
+      "loss": 0.9431,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982013333853663,
+      "loss": 0.8959,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981495918037698,
+      "loss": 0.9335,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998097117211276,
+      "loss": 0.8936,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998043909646421,
+      "loss": 0.9012,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979899691482787,
+      "loss": 0.9279,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997935295756461,
+      "loss": 0.941,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978798895111188,
+      "loss": 0.9524,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978237504529408,
+      "loss": 0.9389,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977668786231534,
+      "loss": 0.9146,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977092740635217,
+      "loss": 0.9389,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976509368163493,
+      "loss": 0.9162,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975918669244762,
+      "loss": 0.9146,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975320644312827,
+      "loss": 0.9451,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974715293806855,
+      "loss": 0.9297,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974102618171394,
+      "loss": 0.9409,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997348261785638,
+      "loss": 0.9172,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972855293317116,
+      "loss": 0.8949,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972220645014297,
+      "loss": 0.3613,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971578673413985,
+      "loss": 0.9448,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970929378987625,
+      "loss": 0.9668,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970272762212044,
+      "loss": 0.9523,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969608823569433,
+      "loss": 0.9276,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968937563547375,
+      "loss": 0.873,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968258982638821,
+      "loss": 0.9195,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967573081342103,
+      "loss": 0.9376,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996687986016092,
+      "loss": 0.9337,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966179319604355,
+      "loss": 0.9314,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996547146018687,
+      "loss": 0.2763,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964756282428287,
+      "loss": 0.9391,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964033786853817,
+      "loss": 0.8973,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963303973994037,
+      "loss": 0.289,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962566844384898,
+      "loss": 0.9129,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961822398567727,
+      "loss": 0.9332,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961070637089222,
+      "loss": 0.8977,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960311560501454,
+      "loss": 0.9814,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959545169361865,
+      "loss": 0.9124,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958771464233274,
+      "loss": 0.9225,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957990445683863,
+      "loss": 0.3177,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957202114287187,
+      "loss": 0.9075,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956406470622172,
+      "loss": 0.293,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955603515273122,
+      "loss": 0.9335,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954793248829695,
+      "loss": 0.9249,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995397567188693,
+      "loss": 0.9291,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995315078504523,
+      "loss": 0.9548,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995231858891037,
+      "loss": 0.9098,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951479084093484,
+      "loss": 0.8822,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950632271211084,
+      "loss": 0.9142,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949778150885042,
+      "loss": 0.955,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948916723742596,
+      "loss": 0.9238,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948047990416354,
+      "loss": 0.8844,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994717195154429,
+      "loss": 0.9051,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946288607769735,
+      "loss": 0.9155,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945397959741394,
+      "loss": 0.887,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994450000811333,
+      "loss": 0.9162,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943594753544974,
+      "loss": 0.9285,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942682196701114,
+      "loss": 0.8971,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941762338251907,
+      "loss": 0.8746,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994083517887287,
+      "loss": 0.9545,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939900719244876,
+      "loss": 0.9546,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938958960054171,
+      "loss": 0.9079,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993800990199235,
+      "loss": 0.9532,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993705354575637,
+      "loss": 0.9179,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019936089892048556,
+      "loss": 0.9045,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935118941576582,
+      "loss": 0.9162,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934140695053486,
+      "loss": 0.9106,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933155153197666,
+      "loss": 0.8784,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932162316732866,
+      "loss": 0.9266,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931162186388206,
+      "loss": 0.9371,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930154762898142,
+      "loss": 0.9515,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199291400470025,
+      "loss": 0.8999,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928118039446458,
+      "loss": 0.9513,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992708874098054,
+      "loss": 0.9219,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992605215236064,
+      "loss": 0.9191,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925008274347995,
+      "loss": 0.9299,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923957107709195,
+      "loss": 0.9295,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922898653216187,
+      "loss": 0.8984,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921832911646268,
+      "loss": 0.914,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920759883782085,
+      "loss": 0.9574,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991967957041164,
+      "loss": 0.8891,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918591972328277,
+      "loss": 0.9041,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917497090330707,
+      "loss": 0.8908,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991639492522296,
+      "loss": 0.8735,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991528547781445,
+      "loss": 0.8752,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914168748919915,
+      "loss": 0.8958,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913044739359442,
+      "loss": 0.8805,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911913449958477,
+      "loss": 0.9184,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199107748815478,
+      "loss": 0.9147,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909629034963548,
+      "loss": 0.9083,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908475911047186,
+      "loss": 0.9135,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907315510645543,
+      "loss": 0.8907,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906147834610773,
+      "loss": 0.8906,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990497288380039,
+      "loss": 0.9165,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990379065907724,
+      "loss": 0.9116,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019902601161309508,
+      "loss": 0.9411,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901404391370732,
+      "loss": 0.8547,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990020035013978,
+      "loss": 0.9192,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898989038500867,
+      "loss": 0.9424,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897770457343538,
+      "loss": 0.9233,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989654460756269,
+      "loss": 0.9148,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019895311490058542,
+      "loss": 0.8723,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019894071105736662,
+      "loss": 0.9337,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989282345550795,
+      "loss": 0.9573,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891568540288644,
+      "loss": 0.9512,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019890306361000315,
+      "loss": 0.9589,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889036918569868,
+      "loss": 0.9151,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887760213929543,
+      "loss": 0.9032,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886476248016916,
+      "loss": 0.9154,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885185021774896,
+      "loss": 0.9318,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988388653615171,
+      "loss": 0.9369,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882580792100933,
+      "loss": 0.9095,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881267790581466,
+      "loss": 0.9196,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879947532557536,
+      "loss": 0.8869,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878620018998696,
+      "loss": 0.9214,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877285250879836,
+      "loss": 0.9631,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019875943229181173,
+      "loss": 0.9416,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987459395488824,
+      "loss": 0.9489,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873237428991907,
+      "loss": 0.8926,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871873652488363,
+      "loss": 0.9024,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870502626379127,
+      "loss": 0.9419,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019869124351671043,
+      "loss": 0.8789,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867738829376264,
+      "loss": 0.9118,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866346060512286,
+      "loss": 0.35,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986494604610191,
+      "loss": 0.9113,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863538787173266,
+      "loss": 0.9217,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862124284759802,
+      "loss": 0.9059,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860702539900287,
+      "loss": 0.8594,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019859273553638808,
+      "loss": 0.8961,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857837327024765,
+      "loss": 0.8977,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856393861112883,
+      "loss": 0.8887,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854943156963199,
+      "loss": 0.9597,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985348521564107,
+      "loss": 0.8813,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985202003821716,
+      "loss": 0.8828,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985054762576745,
+      "loss": 0.9456,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849067979373235,
+      "loss": 0.9075,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847581100121127,
+      "loss": 0.908,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846086989103043,
+      "loss": 0.9054,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019844585647416213,
+      "loss": 0.8997,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843077076163175,
+      "loss": 0.9546,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001984156127645178,
+      "loss": 0.8987,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840038249395185,
+      "loss": 0.2943,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838507996111857,
+      "loss": 0.9115,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019836970517725564,
+      "loss": 0.8643,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019835425815365387,
+      "loss": 0.2979,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833873890165706,
+      "loss": 0.8503,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983231474326621,
+      "loss": 0.909,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830748375811893,
+      "loss": 0.9558,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019829174788953038,
+      "loss": 0.8833,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982759398384525,
+      "loss": 0.8998,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826005961649416,
+      "loss": 0.8924,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019824410723531734,
+      "loss": 0.9385,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822808270663703,
+      "loss": 0.8992,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821198604222113,
+      "loss": 0.8614,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019819581725389056,
+      "loss": 0.3527,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019817957635351915,
+      "loss": 0.8806,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816326335303375,
+      "loss": 0.8939,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019814687826441414,
+      "loss": 0.9443,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198130421099693,
+      "loss": 0.8674,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019811389187095605,
+      "loss": 0.897,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980972905903418,
+      "loss": 0.9719,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019808061727004172,
+      "loss": 0.3063,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980638719223002,
+      "loss": 0.9072,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019804705455941457,
+      "loss": 0.901,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980301651937349,
+      "loss": 0.3198,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801320383766433,
+      "loss": 0.9259,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979961705036587,
+      "loss": 0.9186,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019797906520422677,
+      "loss": 0.9321,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796188795193022,
+      "loss": 0.8927,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794463875938341,
+      "loss": 0.9162,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019792731763925372,
+      "loss": 0.8837,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979099246042612,
+      "loss": 0.8761,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789245966717883,
+      "loss": 0.8884,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019787492284083222,
+      "loss": 0.9106,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785731413809999,
+      "loss": 0.884,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019783963357191339,
+      "loss": 0.8789,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782188115525652,
+      "loss": 0.9221,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001978040569011662,
+      "loss": 0.8918,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019778616082273204,
+      "loss": 0.8967,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776819293309633,
+      "loss": 0.927,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977501532454542,
+      "loss": 0.8841,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019773204177305344,
+      "loss": 0.2967,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019771385852919456,
+      "loss": 0.9667,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976956035272308,
+      "loss": 0.8453,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019767727678056805,
+      "loss": 0.8656,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019765887830266494,
+      "loss": 0.8859,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764040810703277,
+      "loss": 0.8995,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019762186620723545,
+      "loss": 0.8811,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019760325261688964,
+      "loss": 0.8653,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019758456734966456,
+      "loss": 0.9123,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019756581041928213,
+      "loss": 0.8997,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754698183951688,
+      "loss": 0.8625,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975280816241959,
+      "loss": 0.9001,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197509109787199,
+      "loss": 0.8967,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974900663424585,
+      "loss": 0.8883,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001974709513039593,
+      "loss": 0.9184,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745176468573893,
+      "loss": 0.8416,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743250650188749,
+      "loss": 0.8987,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741317676654757,
+      "loss": 0.9243,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019739377549391436,
+      "loss": 0.8897,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973743026982356,
+      "loss": 0.9376,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735475839381144,
+      "loss": 0.8809,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019733514259499466,
+      "loss": 0.9067,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019731545531619056,
+      "loss": 0.9606,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729569657185684,
+      "loss": 0.872,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019727586637650373,
+      "loss": 0.9512,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019725596474469393,
+      "loss": 0.9175,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019723599169104258,
+      "loss": 0.9078,
+      "step": 389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019721594723021732,
+      "loss": 0.9116,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719583137693816,
+      "loss": 0.9056,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717564414597761,
+      "loss": 0.8453,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715538555216053,
+      "loss": 0.8656,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713505561036424,
+      "loss": 0.8858,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019711465433551846,
+      "loss": 0.8789,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970941817426052,
+      "loss": 0.9314,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019707363784665898,
+      "loss": 0.8969,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970530226627666,
+      "loss": 0.9111,
+      "step": 398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019703233620606717,
+      "loss": 0.9072,
+      "step": 399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701157849175228,
+      "loss": 0.9042,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969907495350657,
+      "loss": 0.9573,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019696984935130364,
+      "loss": 0.8805,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019694887795581456,
+      "loss": 0.8692,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969278353639992,
+      "loss": 0.8579,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019690672159131058,
+      "loss": 0.9243,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019688553665325406,
+      "loss": 0.8809,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968642805653872,
+      "loss": 0.8655,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684295334331984,
+      "loss": 0.9069,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019682155500271403,
+      "loss": 0.9052,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019680008555928406,
+      "loss": 0.898,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019677854502879645,
+      "loss": 0.92,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019675693342706993,
+      "loss": 0.8823,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967352507699754,
+      "loss": 0.8579,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671349707343593,
+      "loss": 0.9209,
+      "step": 414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966916723534268,
+      "loss": 0.857,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966697766259754,
+      "loss": 0.8984,
+      "step": 416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019664780990716132,
+      "loss": 0.8795,
+      "step": 417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019662577221311625,
+      "loss": 0.8658,
+      "step": 418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196603663560024,
+      "loss": 0.8677,
+      "step": 419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019658148396412047,
+      "loss": 0.9536,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965592334416937,
+      "loss": 0.3245,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019653691200908382,
+      "loss": 0.8986,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196514519682683,
+      "loss": 0.8707,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964920564789355,
+      "loss": 0.8755,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019646952241433762,
+      "loss": 0.9091,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644691750543767,
+      "loss": 0.8594,
+      "step": 426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019642424176883605,
+      "loss": 0.9358,
+      "step": 427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964014952211851,
+      "loss": 0.3432,
+      "step": 428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963786778791892,
+      "loss": 0.298,
+      "step": 429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019635578975960475,
+      "loss": 0.899,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019633283087924005,
+      "loss": 0.9376,
+      "step": 431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019630980125495543,
+      "loss": 0.9333,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019628670090366318,
+      "loss": 0.8794,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019626352984232744,
+      "loss": 0.9123,
+      "step": 434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019624028808796438,
+      "loss": 0.9138,
+      "step": 435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000196216975657642,
+      "loss": 0.8842,
+      "step": 436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961935925684803,
+      "loss": 0.9062,
+      "step": 437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961701388376511,
+      "loss": 0.8792,
+      "step": 438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961466144823781,
+      "loss": 0.8873,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019612301951993686,
+      "loss": 0.9542,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960993539676548,
+      "loss": 0.8646,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019607561784291125,
+      "loss": 0.9084,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019605181116313724,
+      "loss": 0.9281,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019602793394581572,
+      "loss": 0.9189,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600398620848132,
+      "loss": 0.8792,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019597996796872063,
+      "loss": 0.9257,
+      "step": 446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019595587924417184,
+      "loss": 0.9235,
+      "step": 447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019593172005252506,
+      "loss": 0.9308,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019590749041152204,
+      "loss": 0.9001,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019588319033895623,
+      "loss": 0.8866,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019585881985267295,
+      "loss": 0.9112,
+      "step": 451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019583437897056915,
+      "loss": 0.9044,
+      "step": 452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019580986771059342,
+      "loss": 0.8823,
+      "step": 453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019578528609074614,
+      "loss": 0.905,
+      "step": 454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019576063412907928,
+      "loss": 0.9402,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019573591184369646,
+      "loss": 0.8937,
+      "step": 456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571111925275307,
+      "loss": 0.8985,
+      "step": 457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019568625637445596,
+      "loss": 0.9123,
+      "step": 458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019566132322706374,
+      "loss": 0.8968,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956363198288865,
+      "loss": 0.874,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195611246198286,
+      "loss": 0.8967,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955861023536756,
+      "loss": 0.9117,
+      "step": 462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019556088831352013,
+      "loss": 0.8764,
+      "step": 463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019553560409633601,
+      "loss": 0.9234,
+      "step": 464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019551024972069126,
+      "loss": 0.8585,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019548482520520533,
+      "loss": 0.8421,
+      "step": 466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954593305685492,
+      "loss": 0.903,
+      "step": 467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954337658294454,
+      "loss": 0.8914,
+      "step": 468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019540813100666786,
+      "loss": 0.9011,
+      "step": 469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019538242611904204,
+      "loss": 0.9013,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019535665118544486,
+      "loss": 0.9259,
+      "step": 471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019533080622480462,
+      "loss": 0.883,
+      "step": 472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019530489125610108,
+      "loss": 0.9188,
+      "step": 473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952789062983654,
+      "loss": 0.8992,
+      "step": 474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952528513706802,
+      "loss": 0.878,
+      "step": 475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019522672649217938,
+      "loss": 0.9463,
+      "step": 476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520053168204827,
+      "loss": 0.8771,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019517426695952358,
+      "loss": 0.911,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514793234389326,
+      "loss": 0.9443,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019512152785449672,
+      "loss": 0.3074,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019509505351072458,
+      "loss": 0.3192,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019506850933201882,
+      "loss": 0.93,
+      "step": 482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019504189533787268,
+      "loss": 0.9079,
+      "step": 483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950152115478307,
+      "loss": 0.9431,
+      "step": 484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019498845798148859,
+      "loss": 0.8909,
+      "step": 485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949616346584934,
+      "loss": 0.3111,
+      "step": 486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019493474159854333,
+      "loss": 0.9177,
+      "step": 487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019490777882138788,
+      "loss": 0.906,
+      "step": 488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019488074634682766,
+      "loss": 0.8893,
+      "step": 489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019485364419471454,
+      "loss": 0.8947,
+      "step": 490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019482647238495152,
+      "loss": 0.9058,
+      "step": 491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019479923093749274,
+      "loss": 0.9192,
+      "step": 492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019477191987234344,
+      "loss": 0.8761,
+      "step": 493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019474453920956012,
+      "loss": 0.3181,
+      "step": 494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947170889692503,
+      "loss": 0.9189,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019468956917157256,
+      "loss": 0.9256,
+      "step": 496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019466197983673668,
+      "loss": 0.9227,
+      "step": 497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019463432098500337,
+      "loss": 0.8884,
+      "step": 498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946065926366845,
+      "loss": 0.8589,
+      "step": 499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019457879481214288,
+      "loss": 0.8627,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019455092753179243,
+      "loss": 0.8811,
+      "step": 501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000194522990816098,
+      "loss": 0.9267,
+      "step": 502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019449498468557556,
+      "loss": 0.9043,
+      "step": 503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944669091607919,
+      "loss": 0.8971,
+      "step": 504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944387642623648,
+      "loss": 0.8868,
+      "step": 505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944105500109631,
+      "loss": 0.8163,
+      "step": 506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019438226642730646,
+      "loss": 0.8622,
+      "step": 507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019435391353216547,
+      "loss": 0.9169,
+      "step": 508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019432549134636168,
+      "loss": 0.3126,
+      "step": 509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019429699989076746,
+      "loss": 0.866,
+      "step": 510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019426843918630607,
+      "loss": 0.8623,
+      "step": 511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019423980925395163,
+      "loss": 0.8892,
+      "step": 512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942111101147291,
+      "loss": 0.8695,
+      "step": 513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019418234178971425,
+      "loss": 0.8496,
+      "step": 514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019415350430003372,
+      "loss": 0.8904,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941245976668648,
+      "loss": 0.9093,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019409562191143577,
+      "loss": 0.8914,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019406657705502544,
+      "loss": 0.9165,
+      "step": 518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019403746311896352,
+      "loss": 0.9265,
+      "step": 519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940082801246304,
+      "loss": 0.8569,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019397902809345716,
+      "loss": 0.8495,
+      "step": 521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019394970704692566,
+      "loss": 0.8777,
+      "step": 522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392031700656836,
+      "loss": 0.9102,
+      "step": 523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019389085799396844,
+      "loss": 0.9111,
+      "step": 524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019386133003075967,
+      "loss": 0.8768,
+      "step": 525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019383173313862652,
+      "loss": 0.8783,
+      "step": 526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019380206733930408,
+      "loss": 0.3157,
+      "step": 527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019377233265457794,
+      "loss": 0.8906,
+      "step": 528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019374252910628442,
+      "loss": 0.925,
+      "step": 529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019371265671631037,
+      "loss": 0.9077,
+      "step": 530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019368271550659308,
+      "loss": 0.8904,
+      "step": 531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019365270549912052,
+      "loss": 0.8537,
+      "step": 532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019362262671593112,
+      "loss": 0.8922,
+      "step": 533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019359247917911384,
+      "loss": 0.8534,
+      "step": 534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019356226291080813,
+      "loss": 0.8917,
+      "step": 535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019353197793320389,
+      "loss": 0.9117,
+      "step": 536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935016242685415,
+      "loss": 0.8964,
+      "step": 537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019347120193911175,
+      "loss": 0.8923,
+      "step": 538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344071096725592,
+      "loss": 0.8686,
+      "step": 539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019341015137536562,
+      "loss": 0.8855,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019337952318588292,
+      "loss": 0.8753,
+      "step": 541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001933488264213002,
+      "loss": 0.8809,
+      "step": 542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019331806110416027,
+      "loss": 0.8388,
+      "step": 543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019328722725705624,
+      "loss": 0.8817,
+      "step": 544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019325632490263157,
+      "loss": 0.8927,
+      "step": 545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019322535406358,
+      "loss": 0.8851,
+      "step": 546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019319431476264555,
+      "loss": 0.8667,
+      "step": 547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001931632070226226,
+      "loss": 0.8437,
+      "step": 548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019313203086635566,
+      "loss": 0.9172,
+      "step": 549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019310078631673964,
+      "loss": 0.8561,
+      "step": 550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019306947339671956,
+      "loss": 0.8798,
+      "step": 551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930380921292907,
+      "loss": 0.8369,
+      "step": 552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019300664253749848,
+      "loss": 0.9097,
+      "step": 553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929751246444385,
+      "loss": 0.8584,
+      "step": 554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019294353847325662,
+      "loss": 0.9322,
+      "step": 555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019291188404714878,
+      "loss": 0.875,
+      "step": 556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019288016138936092,
+      "loss": 0.9145,
+      "step": 557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019284837052318933,
+      "loss": 0.8857,
+      "step": 558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928165114719802,
+      "loss": 0.885,
+      "step": 559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927845842591299,
+      "loss": 0.9216,
+      "step": 560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927525889080847,
+      "loss": 0.8934,
+      "step": 561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019272052544234112,
+      "loss": 0.9001,
+      "step": 562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926883938854456,
+      "loss": 0.8743,
+      "step": 563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019265619426099453,
+      "loss": 0.8628,
+      "step": 564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019262392659263438,
+      "loss": 0.8525,
+      "step": 565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925915909040615,
+      "loss": 0.861,
+      "step": 566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019255918721902235,
+      "loss": 0.8711,
+      "step": 567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925267155613131,
+      "loss": 0.9066,
+      "step": 568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019249417595478002,
+      "loss": 0.8866,
+      "step": 569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019246156842331918,
+      "loss": 0.9033,
+      "step": 570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019242889299087658,
+      "loss": 0.8608,
+      "step": 571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923961496814481,
+      "loss": 0.9004,
+      "step": 572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019236333851907936,
+      "loss": 0.8939,
+      "step": 573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019233045952786598,
+      "loss": 0.8841,
+      "step": 574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019229751273195324,
+      "loss": 0.8747,
+      "step": 575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019226449815553629,
+      "loss": 0.9215,
+      "step": 576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019223141582286001,
+      "loss": 0.8959,
+      "step": 577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019219826575821912,
+      "loss": 0.8671,
+      "step": 578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019216504798595798,
+      "loss": 0.8276,
+      "step": 579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921317625304707,
+      "loss": 0.8784,
+      "step": 580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920984094162012,
+      "loss": 0.8675,
+      "step": 581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019206498866764288,
+      "loss": 0.8931,
+      "step": 582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019203150030933902,
+      "loss": 0.8421,
+      "step": 583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019199794436588243,
+      "loss": 0.8903,
+      "step": 584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019196432086191555,
+      "loss": 0.8469,
+      "step": 585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019193062982213055,
+      "loss": 0.8613,
+      "step": 586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019189687127126902,
+      "loss": 0.8841,
+      "step": 587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019186304523412226,
+      "loss": 0.9365,
+      "step": 588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918291517355311,
+      "loss": 0.8879,
+      "step": 589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019179519080038586,
+      "loss": 0.8499,
+      "step": 590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019176116245362646,
+      "loss": 0.8988,
+      "step": 591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917270667202423,
+      "loss": 0.8666,
+      "step": 592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019169290362527222,
+      "loss": 0.9134,
+      "step": 593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019165867319380456,
+      "loss": 0.8893,
+      "step": 594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019162437545097719,
+      "loss": 0.8609,
+      "step": 595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019159001042197722,
+      "loss": 0.8351,
+      "step": 596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019155557813204136,
+      "loss": 0.932,
+      "step": 597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019152107860645568,
+      "loss": 0.9209,
+      "step": 598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914865118705555,
+      "loss": 0.8576,
+      "step": 599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019145187794972565,
+      "loss": 0.8865,
+      "step": 600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914171768694002,
+      "loss": 0.881,
+      "step": 601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913824086550626,
+      "loss": 0.8871,
+      "step": 602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019134757333224558,
+      "loss": 0.8927,
+      "step": 603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913126709265311,
+      "loss": 0.9233,
+      "step": 604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019127770146355047,
+      "loss": 0.9251,
+      "step": 605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912426649689842,
+      "loss": 0.8564,
+      "step": 606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019120756146856203,
+      "loss": 0.9031,
+      "step": 607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019117239098806295,
+      "loss": 0.865,
+      "step": 608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019113715355331502,
+      "loss": 0.9022,
+      "step": 609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019110184919019565,
+      "loss": 0.8715,
+      "step": 610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019106647792463118,
+      "loss": 0.9094,
+      "step": 611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019103103978259725,
+      "loss": 0.9012,
+      "step": 612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909955347901186,
+      "loss": 0.9565,
+      "step": 613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019095996297326898,
+      "loss": 0.877,
+      "step": 614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019092432435817127,
+      "loss": 0.9219,
+      "step": 615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908886189709974,
+      "loss": 0.8457,
+      "step": 616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908528468379683,
+      "loss": 0.8725,
+      "step": 617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019081700798535397,
+      "loss": 0.9054,
+      "step": 618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019078110243947336,
+      "loss": 0.9146,
+      "step": 619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019074513022669445,
+      "loss": 0.9067,
+      "step": 620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070909137343408,
+      "loss": 0.8907,
+      "step": 621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906729859061582,
+      "loss": 0.8957,
+      "step": 622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019063681385138144,
+      "loss": 0.8469,
+      "step": 623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019060057523566756,
+      "loss": 0.9064,
+      "step": 624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019056427008562913,
+      "loss": 0.8801,
+      "step": 625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019052789842792747,
+      "loss": 0.8767,
+      "step": 626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019049146028927286,
+      "loss": 0.887,
+      "step": 627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019045495569642437,
+      "loss": 0.8652,
+      "step": 628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019041838467618986,
+      "loss": 0.8795,
+      "step": 629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019038174725542604,
+      "loss": 0.8513,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019034504346103823,
+      "loss": 0.9053,
+      "step": 631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019030827331998067,
+      "loss": 0.9151,
+      "step": 632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001902714368592562,
+      "loss": 0.8848,
+      "step": 633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019023453410591635,
+      "loss": 0.884,
+      "step": 634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901975650870615,
+      "loss": 0.9023,
+      "step": 635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019016052982984052,
+      "loss": 0.9173,
+      "step": 636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019012342836145102,
+      "loss": 0.8893,
+      "step": 637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900862607091391,
+      "loss": 0.8612,
+      "step": 638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019004902690019967,
+      "loss": 0.879,
+      "step": 639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019001172696197607,
+      "loss": 0.9025,
+      "step": 640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018997436092186022,
+      "loss": 0.9098,
+      "step": 641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899369288072927,
+      "loss": 0.8862,
+      "step": 642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018989943064576242,
+      "loss": 0.9418,
+      "step": 643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018986186646480696,
+      "loss": 0.8633,
+      "step": 644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018982423629201227,
+      "loss": 0.8516,
+      "step": 645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018978654015501287,
+      "loss": 0.8223,
+      "step": 646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897487780814916,
+      "loss": 0.8879,
+      "step": 647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018971095009917986,
+      "loss": 0.8663,
+      "step": 648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018967305623585732,
+      "loss": 0.8879,
+      "step": 649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018963509651935212,
+      "loss": 0.87,
+      "step": 650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018959707097754067,
+      "loss": 0.9071,
+      "step": 651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895589796383479,
+      "loss": 0.905,
+      "step": 652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895208225297468,
+      "loss": 0.9196,
+      "step": 653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018948259967975888,
+      "loss": 0.9095,
+      "step": 654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018944431111645382,
+      "loss": 0.8708,
+      "step": 655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894059568679496,
+      "loss": 0.3453,
+      "step": 656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018936753696241238,
+      "loss": 0.897,
+      "step": 657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018932905142805665,
+      "loss": 0.8753,
+      "step": 658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018929050029314496,
+      "loss": 0.8613,
+      "step": 659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018925188358598813,
+      "loss": 0.8924,
+      "step": 660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001892132013349451,
+      "loss": 0.8544,
+      "step": 661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891744535684229,
+      "loss": 0.8835,
+      "step": 662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891356403148768,
+      "loss": 0.8756,
+      "step": 663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018909676160281,
+      "loss": 0.816,
+      "step": 664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018905781746077395,
+      "loss": 0.3531,
+      "step": 665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018901880791736793,
+      "loss": 0.9619,
+      "step": 666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018897973300123947,
+      "loss": 0.9299,
+      "step": 667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018894059274108396,
+      "loss": 0.8891,
+      "step": 668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018890138716564484,
+      "loss": 0.8457,
+      "step": 669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018886211630371347,
+      "loss": 0.8616,
+      "step": 670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888227801841292,
+      "loss": 0.9484,
+      "step": 671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887833788357793,
+      "loss": 0.8997,
+      "step": 672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018874391228759893,
+      "loss": 0.8743,
+      "step": 673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018870438056857114,
+      "loss": 0.9083,
+      "step": 674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886647837077268,
+      "loss": 0.8955,
+      "step": 675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018862512173414465,
+      "loss": 0.8519,
+      "step": 676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018858539467695126,
+      "loss": 0.8625,
+      "step": 677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188545602565321,
+      "loss": 0.8844,
+      "step": 678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018850574542847594,
+      "loss": 0.89,
+      "step": 679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188465823295686,
+      "loss": 0.3416,
+      "step": 680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884258361962688,
+      "loss": 0.8659,
+      "step": 681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018838578415958962,
+      "loss": 0.8683,
+      "step": 682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018834566721506147,
+      "loss": 0.8698,
+      "step": 683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018830548539214506,
+      "loss": 0.8894,
+      "step": 684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018826523872034865,
+      "loss": 0.8796,
+      "step": 685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882249272292282,
+      "loss": 0.9261,
+      "step": 686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018818455094838727,
+      "loss": 0.9099,
+      "step": 687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018814410990747695,
+      "loss": 0.8845,
+      "step": 688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018810360413619597,
+      "loss": 0.7995,
+      "step": 689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880630336642905,
+      "loss": 0.8796,
+      "step": 690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018802239852155425,
+      "loss": 0.8471,
+      "step": 691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879816987378285,
+      "loss": 0.8666,
+      "step": 692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018794093434300191,
+      "loss": 0.8887,
+      "step": 693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879001053670106,
+      "loss": 0.8764,
+      "step": 694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018785921183983817,
+      "loss": 0.8744,
+      "step": 695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018781825379151558,
+      "loss": 0.8724,
+      "step": 696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777723125212119,
+      "loss": 0.8691,
+      "step": 697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001877361442517807,
+      "loss": 0.9153,
+      "step": 698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018769499282066717,
+      "loss": 0.8832,
+      "step": 699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018765377698900094,
+      "loss": 0.8967,
+      "step": 700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876124967870497,
+      "loss": 0.7931,
+      "step": 701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875711522451284,
+      "loss": 0.8064,
+      "step": 702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875297433935992,
+      "loss": 0.8809,
+      "step": 703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018748827026287147,
+      "loss": 0.8939,
+      "step": 704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874467328834019,
+      "loss": 0.8926,
+      "step": 705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018740513128569426,
+      "loss": 0.907,
+      "step": 706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018736346550029945,
+      "loss": 0.8611,
+      "step": 707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018732173555781567,
+      "loss": 0.8746,
+      "step": 708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000187279941488888,
+      "loss": 0.8888,
+      "step": 709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018723808332420886,
+      "loss": 0.8379,
+      "step": 710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018719616109451754,
+      "loss": 0.914,
+      "step": 711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871541748306005,
+      "loss": 0.8698,
+      "step": 712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018711212456329118,
+      "loss": 0.826,
+      "step": 713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018707001032347,
+      "loss": 0.8892,
+      "step": 714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018702783214206444,
+      "loss": 0.3092,
+      "step": 715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018698559005004878,
+      "loss": 0.9004,
+      "step": 716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018694328407844446,
+      "loss": 0.8489,
+      "step": 717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018690091425831962,
+      "loss": 0.8972,
+      "step": 718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868584806207894,
+      "loss": 0.8352,
+      "step": 719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868159831970158,
+      "loss": 0.2979,
+      "step": 720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018677342201820762,
+      "loss": 0.882,
+      "step": 721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018673079711562052,
+      "loss": 0.8914,
+      "step": 722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018668810852055692,
+      "loss": 0.8966,
+      "step": 723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018664535626436606,
+      "loss": 0.9092,
+      "step": 724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.8731,
+      "step": 725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865596608942331,
+      "loss": 0.8427,
+      "step": 726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018651671784322308,
+      "loss": 0.852,
+      "step": 727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018647371125694992,
+      "loss": 0.8847,
+      "step": 728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018643064116699636,
+      "loss": 0.8993,
+      "step": 729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018638750760499177,
+      "loss": 0.8977,
+      "step": 730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018634431060261214,
+      "loss": 0.8612,
+      "step": 731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018630105019158005,
+      "loss": 0.8814,
+      "step": 732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018625772640366463,
+      "loss": 0.3225,
+      "step": 733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018621433927068154,
+      "loss": 0.9097,
+      "step": 734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186170888824493,
+      "loss": 0.8751,
+      "step": 735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001861273750970077,
+      "loss": 0.8866,
+      "step": 736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860837981201808,
+      "loss": 0.8463,
+      "step": 737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018604015792601396,
+      "loss": 0.8762,
+      "step": 738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018599645454655514,
+      "loss": 0.8203,
+      "step": 739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018595268801389883,
+      "loss": 0.876,
+      "step": 740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859088583601859,
+      "loss": 0.9084,
+      "step": 741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018586496561760337,
+      "loss": 0.8707,
+      "step": 742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858210098183849,
+      "loss": 0.8521,
+      "step": 743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018577699099481024,
+      "loss": 0.9038,
+      "step": 744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018573290917920545,
+      "loss": 0.8446,
+      "step": 745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018568876440394296,
+      "loss": 0.886,
+      "step": 746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018564455670144127,
+      "loss": 0.8336,
+      "step": 747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856002861041653,
+      "loss": 0.872,
+      "step": 748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855559526446259,
+      "loss": 0.8822,
+      "step": 749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855115563553803,
+      "loss": 0.9388,
+      "step": 750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018546709726903178,
+      "loss": 0.774,
+      "step": 751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018542257541822978,
+      "loss": 0.8651,
+      "step": 752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018537799083566975,
+      "loss": 0.8453,
+      "step": 753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018533334355409328,
+      "loss": 0.8915,
+      "step": 754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018528863360628802,
+      "loss": 0.8814,
+      "step": 755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018524386102508755,
+      "loss": 0.8894,
+      "step": 756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851990258433715,
+      "loss": 0.8847,
+      "step": 757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018515412809406553,
+      "loss": 0.959,
+      "step": 758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018510916781014117,
+      "loss": 0.8947,
+      "step": 759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018506414502461585,
+      "loss": 0.8651,
+      "step": 760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018501905977055295,
+      "loss": 0.8934,
+      "step": 761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849739120810618,
+      "loss": 0.8868,
+      "step": 762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849287019892974,
+      "loss": 0.8639,
+      "step": 763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018488342952846073,
+      "loss": 0.8541,
+      "step": 764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018483809473179852,
+      "loss": 0.907,
+      "step": 765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018479269763260322,
+      "loss": 0.8494,
+      "step": 766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018474723826421315,
+      "loss": 0.8973,
+      "step": 767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018470171666001226,
+      "loss": 0.877,
+      "step": 768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018465613285343024,
+      "loss": 0.8958,
+      "step": 769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018461048687794249,
+      "loss": 0.8287,
+      "step": 770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018456477876707003,
+      "loss": 0.909,
+      "step": 771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001845190085543795,
+      "loss": 0.8869,
+      "step": 772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018447317627348314,
+      "loss": 0.8302,
+      "step": 773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018442728195803881,
+      "loss": 0.868,
+      "step": 774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018438132564174996,
+      "loss": 0.8744,
+      "step": 775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018433530735836545,
+      "loss": 0.3316,
+      "step": 776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842892271416797,
+      "loss": 0.8705,
+      "step": 777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018424308502553273,
+      "loss": 0.8856,
+      "step": 778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018419688104380982,
+      "loss": 0.8369,
+      "step": 779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018415061523044184,
+      "loss": 0.8828,
+      "step": 780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018410428761940495,
+      "loss": 0.886,
+      "step": 781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001840578982447208,
+      "loss": 0.9478,
+      "step": 782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001840114471404563,
+      "loss": 0.8495,
+      "step": 783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018396493434072374,
+      "loss": 0.9125,
+      "step": 784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018391835987968075,
+      "loss": 0.3457,
+      "step": 785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838717237915302,
+      "loss": 0.8869,
+      "step": 786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018382502611052025,
+      "loss": 0.8772,
+      "step": 787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018377826687094418,
+      "loss": 0.8908,
+      "step": 788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018373144610714061,
+      "loss": 0.8464,
+      "step": 789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018368456385349334,
+      "loss": 0.8698,
+      "step": 790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018363762014443117,
+      "loss": 0.8683,
+      "step": 791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018359061501442824,
+      "loss": 0.8595,
+      "step": 792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018354354849800365,
+      "loss": 0.891,
+      "step": 793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018349642062972165,
+      "loss": 0.9243,
+      "step": 794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018344923144419156,
+      "loss": 0.8992,
+      "step": 795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001834019809760676,
+      "loss": 0.877,
+      "step": 796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018335466926004915,
+      "loss": 0.8461,
+      "step": 797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018330729633088045,
+      "loss": 0.8701,
+      "step": 798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832598622233508,
+      "loss": 0.825,
+      "step": 799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018321236697229436,
+      "loss": 0.8055,
+      "step": 800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018316481061259022,
+      "loss": 0.8804,
+      "step": 801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831171931791623,
+      "loss": 0.8568,
+      "step": 802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018306951470697946,
+      "loss": 0.8737,
+      "step": 803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018302177523105525,
+      "loss": 0.8267,
+      "step": 804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018297397478644816,
+      "loss": 0.9215,
+      "step": 805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018292611340826138,
+      "loss": 0.8636,
+      "step": 806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018287819113164283,
+      "loss": 0.8786,
+      "step": 807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018283020799178524,
+      "loss": 0.864,
+      "step": 808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827821640239259,
+      "loss": 0.8877,
+      "step": 809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018273405926334696,
+      "loss": 0.8747,
+      "step": 810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018268589374537497,
+      "loss": 0.8865,
+      "step": 811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018263766750538132,
+      "loss": 0.8733,
+      "step": 812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018258938057878188,
+      "loss": 0.3491,
+      "step": 813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018254103300103708,
+      "loss": 0.8578,
+      "step": 814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018249262480765196,
+      "loss": 0.8629,
+      "step": 815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018244415603417603,
+      "loss": 0.8691,
+      "step": 816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018239562671620328,
+      "loss": 0.8726,
+      "step": 817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001823470368893722,
+      "loss": 0.8616,
+      "step": 818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018229838658936564,
+      "loss": 0.8997,
+      "step": 819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018224967585191094,
+      "loss": 0.9021,
+      "step": 820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018220090471277983,
+      "loss": 0.8958,
+      "step": 821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821520732077883,
+      "loss": 0.8727,
+      "step": 822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018210318137279675,
+      "loss": 0.9057,
+      "step": 823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018205422924370986,
+      "loss": 0.866,
+      "step": 824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018200521685647663,
+      "loss": 0.8895,
+      "step": 825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018195614424709026,
+      "loss": 0.8722,
+      "step": 826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018190701145158816,
+      "loss": 0.8836,
+      "step": 827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181857818506052,
+      "loss": 0.8502,
+      "step": 828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818085654466076,
+      "loss": 0.8409,
+      "step": 829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001817592523094249,
+      "loss": 0.3291,
+      "step": 830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018170987913071796,
+      "loss": 0.8606,
+      "step": 831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018166044594674502,
+      "loss": 0.8721,
+      "step": 832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018161095279380822,
+      "loss": 0.8956,
+      "step": 833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018156139970825391,
+      "loss": 0.8457,
+      "step": 834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018151178672647236,
+      "loss": 0.8547,
+      "step": 835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018146211388489784,
+      "loss": 0.9382,
+      "step": 836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018141238122000857,
+      "loss": 0.8193,
+      "step": 837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018136258876832672,
+      "loss": 0.3394,
+      "step": 838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001813127365664184,
+      "loss": 0.8861,
+      "step": 839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812628246508935,
+      "loss": 0.86,
+      "step": 840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018121285305840588,
+      "loss": 0.9008,
+      "step": 841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018116282182565311,
+      "loss": 0.9124,
+      "step": 842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018111273098937668,
+      "loss": 0.8964,
+      "step": 843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001810625805863617,
+      "loss": 0.9168,
+      "step": 844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018101237065343723,
+      "loss": 0.883,
+      "step": 845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018096210122747584,
+      "loss": 0.8819,
+      "step": 846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809117723453939,
+      "loss": 0.9164,
+      "step": 847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808613840441514,
+      "loss": 0.8513,
+      "step": 848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000180810936360752,
+      "loss": 0.8524,
+      "step": 849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018076042933224299,
+      "loss": 0.8316,
+      "step": 850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018070986299571518,
+      "loss": 0.8866,
+      "step": 851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018065923738830292,
+      "loss": 0.3265,
+      "step": 852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018060855254718416,
+      "loss": 0.8958,
+      "step": 853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018055780850958033,
+      "loss": 0.897,
+      "step": 854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001805070053127563,
+      "loss": 0.8997,
+      "step": 855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018045614299402041,
+      "loss": 0.9118,
+      "step": 856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018040522159072439,
+      "loss": 0.841,
+      "step": 857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803542411402634,
+      "loss": 0.856,
+      "step": 858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018030320168007594,
+      "loss": 0.9011,
+      "step": 859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018025210324764385,
+      "loss": 0.8785,
+      "step": 860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018020094588049227,
+      "loss": 0.8541,
+      "step": 861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018014972961618962,
+      "loss": 0.8306,
+      "step": 862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001800984544923476,
+      "loss": 0.8748,
+      "step": 863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018004712054662107,
+      "loss": 0.9232,
+      "step": 864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017999572781670815,
+      "loss": 0.8939,
+      "step": 865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017994427634035015,
+      "loss": 0.8457,
+      "step": 866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798927661553314,
+      "loss": 0.818,
+      "step": 867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017984119729947944,
+      "loss": 0.897,
+      "step": 868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017978956981066493,
+      "loss": 0.8606,
+      "step": 869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797378837268015,
+      "loss": 0.8928,
+      "step": 870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017968613908584587,
+      "loss": 0.8469,
+      "step": 871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796343359257977,
+      "loss": 0.9014,
+      "step": 872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001795824742846997,
+      "loss": 0.8333,
+      "step": 873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017953055420063749,
+      "loss": 0.9049,
+      "step": 874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794785757117396,
+      "loss": 0.8924,
+      "step": 875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001794265388561775,
+      "loss": 0.8513,
+      "step": 876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017937444367216543,
+      "loss": 0.8461,
+      "step": 877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793222901979605,
+      "loss": 0.8803,
+      "step": 878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017927007847186275,
+      "loss": 0.9003,
+      "step": 879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792178085322148,
+      "loss": 0.8535,
+      "step": 880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017916548041740213,
+      "loss": 0.8133,
+      "step": 881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017911309416585296,
+      "loss": 0.8036,
+      "step": 882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017906064981603815,
+      "loss": 0.8647,
+      "step": 883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017900814740647123,
+      "loss": 0.8379,
+      "step": 884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017895558697570836,
+      "loss": 0.8405,
+      "step": 885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017890296856234845,
+      "loss": 0.8737,
+      "step": 886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017885029220503275,
+      "loss": 0.8864,
+      "step": 887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017879755794244525,
+      "loss": 0.8543,
+      "step": 888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017874476581331234,
+      "loss": 0.839,
+      "step": 889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017869191585640303,
+      "loss": 0.8934,
+      "step": 890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017863900811052868,
+      "loss": 0.8519,
+      "step": 891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785860426145432,
+      "loss": 0.9053,
+      "step": 892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017853301940734276,
+      "loss": 0.8768,
+      "step": 893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001784799385278661,
+      "loss": 0.828,
+      "step": 894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017842680001509416,
+      "loss": 0.8526,
+      "step": 895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017837360390805026,
+      "loss": 0.8818,
+      "step": 896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017832035024580001,
+      "loss": 0.8845,
+      "step": 897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017826703906745128,
+      "loss": 0.8407,
+      "step": 898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017821367041215424,
+      "loss": 0.8553,
+      "step": 899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017816024431910115,
+      "loss": 0.8603,
+      "step": 900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781067608275266,
+      "loss": 0.85,
+      "step": 901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017805321997670716,
+      "loss": 0.8792,
+      "step": 902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017799962180596167,
+      "loss": 0.3262,
+      "step": 903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177945966354651,
+      "loss": 0.8792,
+      "step": 904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778922536621781,
+      "loss": 0.8517,
+      "step": 905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778384837679879,
+      "loss": 0.8346,
+      "step": 906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017778465671156743,
+      "loss": 0.8873,
+      "step": 907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017773077253244564,
+      "loss": 0.8663,
+      "step": 908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017767683127019342,
+      "loss": 0.8646,
+      "step": 909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017762283296442367,
+      "loss": 0.8909,
+      "step": 910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017756877765479106,
+      "loss": 0.8999,
+      "step": 911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017751466538099217,
+      "loss": 0.8688,
+      "step": 912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017746049618276545,
+      "loss": 0.8432,
+      "step": 913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017740627009989112,
+      "loss": 0.828,
+      "step": 914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017735198717219115,
+      "loss": 0.8809,
+      "step": 915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017729764743952932,
+      "loss": 0.8768,
+      "step": 916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017724325094181103,
+      "loss": 0.8488,
+      "step": 917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017718879771898348,
+      "loss": 0.8898,
+      "step": 918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017713428781103544,
+      "loss": 0.8752,
+      "step": 919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017707972125799735,
+      "loss": 0.8803,
+      "step": 920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017702509809994124,
+      "loss": 0.8631,
+      "step": 921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769704183769807,
+      "loss": 0.8636,
+      "step": 922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017691568212927088,
+      "loss": 0.8554,
+      "step": 923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768608893970084,
+      "loss": 0.8656,
+      "step": 924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017680604022043142,
+      "loss": 0.8176,
+      "step": 925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017675113463981948,
+      "loss": 0.8654,
+      "step": 926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017669617269549362,
+      "loss": 0.3328,
+      "step": 927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766411544278162,
+      "loss": 0.8286,
+      "step": 928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017658607987719097,
+      "loss": 0.9123,
+      "step": 929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017653094908406301,
+      "loss": 0.8777,
+      "step": 930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764757620889188,
+      "loss": 0.9166,
+      "step": 931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764205189322859,
+      "loss": 0.8363,
+      "step": 932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017636521965473323,
+      "loss": 0.8417,
+      "step": 933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763098642968709,
+      "loss": 0.8421,
+      "step": 934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017625445289935028,
+      "loss": 0.841,
+      "step": 935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017619898550286375,
+      "loss": 0.9044,
+      "step": 936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017614346214814495,
+      "loss": 0.902,
+      "step": 937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760878828759685,
+      "loss": 0.8987,
+      "step": 938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017603224772715014,
+      "loss": 0.8701,
+      "step": 939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759765567425467,
+      "loss": 0.8812,
+      "step": 940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017592080996305592,
+      "loss": 0.8693,
+      "step": 941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017586500742961653,
+      "loss": 0.3324,
+      "step": 942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017580914918320826,
+      "loss": 0.8589,
+      "step": 943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757532352648517,
+      "loss": 0.8862,
+      "step": 944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017569726571560837,
+      "loss": 0.8839,
+      "step": 945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017564124057658056,
+      "loss": 0.8302,
+      "step": 946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017558515988891146,
+      "loss": 0.9255,
+      "step": 947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017552902369378506,
+      "loss": 0.8356,
+      "step": 948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017547283203242606,
+      "loss": 0.3256,
+      "step": 949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017541658494609993,
+      "loss": 0.8521,
+      "step": 950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017536028247611281,
+      "loss": 0.3272,
+      "step": 951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017530392466381158,
+      "loss": 0.8779,
+      "step": 952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017524751155058364,
+      "loss": 0.8807,
+      "step": 953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017519104317785717,
+      "loss": 0.813,
+      "step": 954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017513451958710077,
+      "loss": 0.8976,
+      "step": 955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017507794081982365,
+      "loss": 0.8597,
+      "step": 956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017502130691757556,
+      "loss": 0.8515,
+      "step": 957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017496461792194677,
+      "loss": 0.3149,
+      "step": 958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749078738745679,
+      "loss": 0.8798,
+      "step": 959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017485107481711012,
+      "loss": 0.8758,
+      "step": 960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017479422079128492,
+      "loss": 0.9118,
+      "step": 961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017473731183884414,
+      "loss": 0.8568,
+      "step": 962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017468034800158003,
+      "loss": 0.8775,
+      "step": 963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017462332932132515,
+      "loss": 0.8694,
+      "step": 964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017456625583995222,
+      "loss": 0.8832,
+      "step": 965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017450912759937434,
+      "loss": 0.9382,
+      "step": 966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017445194464154473,
+      "loss": 0.8759,
+      "step": 967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017439470700845682,
+      "loss": 0.8644,
+      "step": 968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017433741474214421,
+      "loss": 0.9162,
+      "step": 969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017428006788468064,
+      "loss": 0.9142,
+      "step": 970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017422266647817986,
+      "loss": 0.879,
+      "step": 971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017416521056479577,
+      "loss": 0.8343,
+      "step": 972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017410770018672222,
+      "loss": 0.8622,
+      "step": 973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740501353861931,
+      "loss": 0.8802,
+      "step": 974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739925162054823,
+      "loss": 0.8734,
+      "step": 975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017393484268690354,
+      "loss": 0.8419,
+      "step": 976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017387711487281058,
+      "loss": 0.8819,
+      "step": 977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017381933280559693,
+      "loss": 0.8076,
+      "step": 978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173761496527696,
+      "loss": 0.8291,
+      "step": 979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017370360608158101,
+      "loss": 0.8524,
+      "step": 980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017364566150976498,
+      "loss": 0.8559,
+      "step": 981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017358766285480065,
+      "loss": 0.9158,
+      "step": 982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017352961015928043,
+      "loss": 0.8846,
+      "step": 983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001734715034658365,
+      "loss": 0.8989,
+      "step": 984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017341334281714064,
+      "loss": 0.8765,
+      "step": 985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017335512825590433,
+      "loss": 0.8555,
+      "step": 986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017329685982487853,
+      "loss": 0.3125,
+      "step": 987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001732385375668538,
+      "loss": 0.8727,
+      "step": 988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001731801615246603,
+      "loss": 0.888,
+      "step": 989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017312173174116762,
+      "loss": 0.3315,
+      "step": 990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017306324825928479,
+      "loss": 0.3089,
+      "step": 991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017300471112196032,
+      "loss": 0.8657,
+      "step": 992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017294612037218213,
+      "loss": 0.8927,
+      "step": 993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017288747605297754,
+      "loss": 0.8832,
+      "step": 994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017282877820741307,
+      "loss": 0.8878,
+      "step": 995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017277002687859471,
+      "loss": 0.8493,
+      "step": 996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017271122210966766,
+      "loss": 0.8442,
+      "step": 997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017265236394381633,
+      "loss": 0.8031,
+      "step": 998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725934524242644,
+      "loss": 0.8929,
+      "step": 999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725344875942747,
+      "loss": 0.8997,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017247546949714922,
+      "loss": 0.8589,
+      "step": 1001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001724163981762291,
+      "loss": 0.8257,
+      "step": 1002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001723572736748945,
+      "loss": 0.8169,
+      "step": 1003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017229809603656467,
+      "loss": 0.849,
+      "step": 1004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017223886530469792,
+      "loss": 0.8913,
+      "step": 1005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017217958152279143,
+      "loss": 0.9153,
+      "step": 1006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017212024473438147,
+      "loss": 0.9035,
+      "step": 1007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001720608549830432,
+      "loss": 0.9109,
+      "step": 1008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017200141231239058,
+      "loss": 0.8886,
+      "step": 1009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719419167660766,
+      "loss": 0.8888,
+      "step": 1010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017188236838779295,
+      "loss": 0.8608,
+      "step": 1011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017182276722127018,
+      "loss": 0.9083,
+      "step": 1012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017176311331027754,
+      "loss": 0.915,
+      "step": 1013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717034066986231,
+      "loss": 0.8607,
+      "step": 1014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716436474301536,
+      "loss": 0.8839,
+      "step": 1015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001715838355487544,
+      "loss": 0.824,
+      "step": 1016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017152397109834958,
+      "loss": 0.8528,
+      "step": 1017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001714640541229018,
+      "loss": 0.9056,
+      "step": 1018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017140408466641217,
+      "loss": 0.8544,
+      "step": 1019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017134406277292055,
+      "loss": 0.8712,
+      "step": 1020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017128398848650518,
+      "loss": 0.8219,
+      "step": 1021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017122386185128277,
+      "loss": 0.8099,
+      "step": 1022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017116368291140855,
+      "loss": 0.3146,
+      "step": 1023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001711034517110761,
+      "loss": 0.8451,
+      "step": 1024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017104316829451736,
+      "loss": 0.8513,
+      "step": 1025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017098283270600267,
+      "loss": 0.8516,
+      "step": 1026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017092244498984069,
+      "loss": 0.8641,
+      "step": 1027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017086200519037832,
+      "loss": 0.8935,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017080151335200069,
+      "loss": 0.8423,
+      "step": 1029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017074096951913122,
+      "loss": 0.8511,
+      "step": 1030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017068037373623143,
+      "loss": 0.846,
+      "step": 1031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017061972604780106,
+      "loss": 0.8539,
+      "step": 1032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017055902649837793,
+      "loss": 0.904,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000170498275132538,
+      "loss": 0.8283,
+      "step": 1034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017043747199489516,
+      "loss": 0.8511,
+      "step": 1035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001703766171301014,
+      "loss": 0.894,
+      "step": 1036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017031571058284678,
+      "loss": 0.8613,
+      "step": 1037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017025475239785919,
+      "loss": 0.9057,
+      "step": 1038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017019374261990444,
+      "loss": 0.8187,
+      "step": 1039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017013268129378628,
+      "loss": 0.8724,
+      "step": 1040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001700715684643463,
+      "loss": 0.3495,
+      "step": 1041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017001040417646398,
+      "loss": 0.8488,
+      "step": 1042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016994918847505647,
+      "loss": 0.8279,
+      "step": 1043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016988792140507867,
+      "loss": 0.8929,
+      "step": 1044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016982660301152337,
+      "loss": 0.3203,
+      "step": 1045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016976523333942094,
+      "loss": 0.8681,
+      "step": 1046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016970381243383933,
+      "loss": 0.8784,
+      "step": 1047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016964234033988427,
+      "loss": 0.7932,
+      "step": 1048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016958081710269896,
+      "loss": 0.8272,
+      "step": 1049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016951924276746425,
+      "loss": 0.8599,
+      "step": 1050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016945761737939842,
+      "loss": 0.8733,
+      "step": 1051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016939594098375736,
+      "loss": 0.9024,
+      "step": 1052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016933421362583427,
+      "loss": 0.8759,
+      "step": 1053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016927243535095997,
+      "loss": 0.8768,
+      "step": 1054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692106062045024,
+      "loss": 0.8553,
+      "step": 1055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016914872623186716,
+      "loss": 0.9156,
+      "step": 1056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016908679547849697,
+      "loss": 0.9056,
+      "step": 1057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016902481398987194,
+      "loss": 0.842,
+      "step": 1058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016896278181150932,
+      "loss": 0.8691,
+      "step": 1059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016890069898896375,
+      "loss": 0.8308,
+      "step": 1060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016883856556782696,
+      "loss": 0.8875,
+      "step": 1061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016877638159372782,
+      "loss": 0.87,
+      "step": 1062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687141471123324,
+      "loss": 0.8347,
+      "step": 1063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001686518621693438,
+      "loss": 0.8746,
+      "step": 1064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016858952681050223,
+      "loss": 0.8707,
+      "step": 1065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016852714108158486,
+      "loss": 0.8793,
+      "step": 1066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016846470502840591,
+      "loss": 0.8455,
+      "step": 1067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016840221869681652,
+      "loss": 0.8836,
+      "step": 1068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016833968213270475,
+      "loss": 0.9334,
+      "step": 1069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016827709538199556,
+      "loss": 0.8845,
+      "step": 1070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016821445849065083,
+      "loss": 0.8763,
+      "step": 1071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001681517715046691,
+      "loss": 0.316,
+      "step": 1072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001680890344700859,
+      "loss": 0.8594,
+      "step": 1073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016802624743297333,
+      "loss": 0.8612,
+      "step": 1074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016796341043944036,
+      "loss": 0.8487,
+      "step": 1075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016790052353563253,
+      "loss": 0.8824,
+      "step": 1076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016783758676773206,
+      "loss": 0.9024,
+      "step": 1077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001677746001819579,
+      "loss": 0.9115,
+      "step": 1078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016771156382456543,
+      "loss": 0.8537,
+      "step": 1079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016764847774184663,
+      "loss": 0.8018,
+      "step": 1080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001675853419801301,
+      "loss": 0.8994,
+      "step": 1081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016752215658578075,
+      "loss": 0.8457,
+      "step": 1082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016745892160520005,
+      "loss": 0.8274,
+      "step": 1083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016739563708482585,
+      "loss": 0.8373,
+      "step": 1084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016733230307113242,
+      "loss": 0.8661,
+      "step": 1085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016726891961063028,
+      "loss": 0.836,
+      "step": 1086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016720548674986643,
+      "loss": 0.8897,
+      "step": 1087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000167142004535424,
+      "loss": 0.8746,
+      "step": 1088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016707847301392236,
+      "loss": 0.8482,
+      "step": 1089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001670148922320172,
+      "loss": 0.8766,
+      "step": 1090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016695126223640027,
+      "loss": 0.325,
+      "step": 1091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668875830737996,
+      "loss": 0.8938,
+      "step": 1092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016682385479097915,
+      "loss": 0.832,
+      "step": 1093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001667600774347391,
+      "loss": 0.8398,
+      "step": 1094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016669625105191562,
+      "loss": 0.8444,
+      "step": 1095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016663237568938086,
+      "loss": 0.8233,
+      "step": 1096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016656845139404296,
+      "loss": 0.8845,
+      "step": 1097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016650447821284594,
+      "loss": 0.9002,
+      "step": 1098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016644045619276985,
+      "loss": 0.9139,
+      "step": 1099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663763853808305,
+      "loss": 0.8759,
+      "step": 1100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016631226582407952,
+      "loss": 0.8655,
+      "step": 1101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016624809756960444,
+      "loss": 0.8686,
+      "step": 1102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016618388066452845,
+      "loss": 0.8507,
+      "step": 1103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661196151560105,
+      "loss": 0.8696,
+      "step": 1104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016605530109124522,
+      "loss": 0.8972,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016599093851746294,
+      "loss": 0.7934,
+      "step": 1106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016592652748192963,
+      "loss": 0.8522,
+      "step": 1107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016586206803194674,
+      "loss": 0.9147,
+      "step": 1108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016579756021485137,
+      "loss": 0.864,
+      "step": 1109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016573300407801616,
+      "loss": 0.871,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016566839966884907,
+      "loss": 0.8769,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001656037470347937,
+      "loss": 0.8797,
+      "step": 1112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553904622332904,
+      "loss": 0.9033,
+      "step": 1113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016547429728196931,
+      "loss": 0.8771,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016540950025826422,
+      "loss": 0.8064,
+      "step": 1115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016534465519979876,
+      "loss": 0.8633,
+      "step": 1116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001652797621541932,
+      "loss": 0.8524,
+      "step": 1117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016521482116910294,
+      "loss": 0.8413,
+      "step": 1118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016514983229221877,
+      "loss": 0.8941,
+      "step": 1119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650847955712665,
+      "loss": 0.8396,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016501971105400718,
+      "loss": 0.878,
+      "step": 1121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001649545787882369,
+      "loss": 0.7963,
+      "step": 1122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016488939882178678,
+      "loss": 0.8171,
+      "step": 1123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648241712025231,
+      "loss": 0.8927,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016475889597834695,
+      "loss": 0.8942,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001646935731971946,
+      "loss": 0.8967,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016462820290703703,
+      "loss": 0.8579,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016456278515588024,
+      "loss": 0.8716,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016449731999176507,
+      "loss": 0.9311,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016443180746276715,
+      "loss": 0.897,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016436624761699686,
+      "loss": 0.8994,
+      "step": 1131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001643006405025994,
+      "loss": 0.8824,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016423498616775466,
+      "loss": 0.8056,
+      "step": 1133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016416928466067723,
+      "loss": 0.7716,
+      "step": 1134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016410353602961623,
+      "loss": 0.8899,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640377403228555,
+      "loss": 0.8593,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016397189758871346,
+      "loss": 0.854,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163906007875543,
+      "loss": 0.9012,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016384007123173154,
+      "loss": 0.8346,
+      "step": 1139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637740877057009,
+      "loss": 0.8277,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016370805734590747,
+      "loss": 0.8436,
+      "step": 1141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001636419802008419,
+      "loss": 0.8518,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016357585631902928,
+      "loss": 0.847,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016350968574902895,
+      "loss": 0.8619,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634434685394346,
+      "loss": 0.8417,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633772047388742,
+      "loss": 0.8243,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001633108943960098,
+      "loss": 0.8794,
+      "step": 1147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016324453755953773,
+      "loss": 0.8409,
+      "step": 1148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631781342781885,
+      "loss": 0.8647,
+      "step": 1149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016311168460072658,
+      "loss": 0.8201,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630451885759507,
+      "loss": 0.8883,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001629786462526935,
+      "loss": 0.8206,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001629120576798216,
+      "loss": 0.8898,
+      "step": 1153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016284542290623567,
+      "loss": 0.8458,
+      "step": 1154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001627787419808703,
+      "loss": 0.887,
+      "step": 1155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016271201495269395,
+      "loss": 0.887,
+      "step": 1156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016264524187070887,
+      "loss": 0.8736,
+      "step": 1157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016257842278395127,
+      "loss": 0.9136,
+      "step": 1158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016251155774149102,
+      "loss": 0.8366,
+      "step": 1159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001624446467924318,
+      "loss": 0.8981,
+      "step": 1160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016237768998591097,
+      "loss": 0.8845,
+      "step": 1161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001623106873710996,
+      "loss": 0.3316,
+      "step": 1162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001622436389972024,
+      "loss": 0.8625,
+      "step": 1163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016217654491345766,
+      "loss": 0.8971,
+      "step": 1164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016210940516913724,
+      "loss": 0.8777,
+      "step": 1165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016204221981354656,
+      "loss": 0.8946,
+      "step": 1166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016197498889602448,
+      "loss": 0.8594,
+      "step": 1167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619077124659434,
+      "loss": 0.8845,
+      "step": 1168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016184039057270906,
+      "loss": 0.3509,
+      "step": 1169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617730232657606,
+      "loss": 0.3234,
+      "step": 1170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016170561059457064,
+      "loss": 0.8394,
+      "step": 1171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001616381526086449,
+      "loss": 0.9552,
+      "step": 1172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016157064935752252,
+      "loss": 0.8685,
+      "step": 1173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016150310089077584,
+      "loss": 0.9135,
+      "step": 1174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016143550725801045,
+      "loss": 0.8734,
+      "step": 1175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016136786850886505,
+      "loss": 0.8481,
+      "step": 1176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016130018469301147,
+      "loss": 0.8302,
+      "step": 1177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001612324558601547,
+      "loss": 0.8871,
+      "step": 1178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016116468206003268,
+      "loss": 0.8365,
+      "step": 1179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016109686334241655,
+      "loss": 0.3409,
+      "step": 1180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016102899975711023,
+      "loss": 0.8219,
+      "step": 1181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609610913539507,
+      "loss": 0.8925,
+      "step": 1182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016089313818280792,
+      "loss": 0.8787,
+      "step": 1183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016082514029358455,
+      "loss": 0.8316,
+      "step": 1184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016075709773621623,
+      "loss": 0.8802,
+      "step": 1185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606890105606713,
+      "loss": 0.8329,
+      "step": 1186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016062087881695103,
+      "loss": 0.8782,
+      "step": 1187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016055270255508925,
+      "loss": 0.858,
+      "step": 1188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016048448182515257,
+      "loss": 0.8521,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016041621667724017,
+      "loss": 0.8486,
+      "step": 1190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016034790716148395,
+      "loss": 0.8982,
+      "step": 1191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602795533280484,
+      "loss": 0.9177,
+      "step": 1192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016021115522713047,
+      "loss": 0.922,
+      "step": 1193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016014271290895965,
+      "loss": 0.8591,
+      "step": 1194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600742264237979,
+      "loss": 0.8909,
+      "step": 1195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016000569582193968,
+      "loss": 0.8445,
+      "step": 1196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015993712115371173,
+      "loss": 0.8611,
+      "step": 1197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015986850246947325,
+      "loss": 0.8465,
+      "step": 1198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597998398196157,
+      "loss": 0.8347,
+      "step": 1199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597311332545629,
+      "loss": 0.8825,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015966238282477083,
+      "loss": 0.8641,
+      "step": 1201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015959358858072778,
+      "loss": 0.8734,
+      "step": 1202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595247505729541,
+      "loss": 0.8714,
+      "step": 1203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015945586885200239,
+      "loss": 0.8622,
+      "step": 1204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593869434684573,
+      "loss": 0.3242,
+      "step": 1205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015931797447293552,
+      "loss": 0.3186,
+      "step": 1206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015924896191608584,
+      "loss": 0.8937,
+      "step": 1207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015917990584858894,
+      "loss": 0.8677,
+      "step": 1208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015911080632115753,
+      "loss": 0.8567,
+      "step": 1209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015904166338453628,
+      "loss": 0.8297,
+      "step": 1210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589724770895016,
+      "loss": 0.8671,
+      "step": 1211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589032474868618,
+      "loss": 0.8594,
+      "step": 1212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015883397462745704,
+      "loss": 0.8375,
+      "step": 1213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015876465856215925,
+      "loss": 0.8689,
+      "step": 1214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000158695299341872,
+      "loss": 0.8713,
+      "step": 1215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586258970175306,
+      "loss": 0.8437,
+      "step": 1216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015855645164010207,
+      "loss": 0.8259,
+      "step": 1217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015848696326058498,
+      "loss": 0.8484,
+      "step": 1218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015841743193000944,
+      "loss": 0.9095,
+      "step": 1219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015834785769943725,
+      "loss": 0.947,
+      "step": 1220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015827824061996157,
+      "loss": 0.8326,
+      "step": 1221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015820858074270708,
+      "loss": 0.8935,
+      "step": 1222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581388781188299,
+      "loss": 0.8441,
+      "step": 1223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580691327995175,
+      "loss": 0.8842,
+      "step": 1224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001579993448359888,
+      "loss": 0.8631,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015792951427949392,
+      "loss": 0.8449,
+      "step": 1226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015785964118131428,
+      "loss": 0.8338,
+      "step": 1227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015778972559276266,
+      "loss": 0.8478,
+      "step": 1228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001577197675651829,
+      "loss": 0.8742,
+      "step": 1229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015764976714995009,
+      "loss": 0.8549,
+      "step": 1230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015757972439847038,
+      "loss": 0.3276,
+      "step": 1231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015750963936218105,
+      "loss": 0.8323,
+      "step": 1232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574395120925505,
+      "loss": 0.8749,
+      "step": 1233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015736934264107798,
+      "loss": 0.3388,
+      "step": 1234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572991310592939,
+      "loss": 0.9211,
+      "step": 1235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015722887739875947,
+      "loss": 0.8658,
+      "step": 1236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015715858171106686,
+      "loss": 0.8842,
+      "step": 1237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015708824404783917,
+      "loss": 0.8612,
+      "step": 1238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015701786446073018,
+      "loss": 0.8535,
+      "step": 1239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015694744300142453,
+      "loss": 0.8977,
+      "step": 1240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001568769797216376,
+      "loss": 0.8523,
+      "step": 1241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015680647467311557,
+      "loss": 0.8303,
+      "step": 1242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567359279076352,
+      "loss": 0.8439,
+      "step": 1243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015666533947700383,
+      "loss": 0.85,
+      "step": 1244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015659470943305955,
+      "loss": 0.8598,
+      "step": 1245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015652403782767085,
+      "loss": 0.8215,
+      "step": 1246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015645332471273687,
+      "loss": 0.8706,
+      "step": 1247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015638257014018723,
+      "loss": 0.8122,
+      "step": 1248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015631177416198185,
+      "loss": 0.8441,
+      "step": 1249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015624093683011124,
+      "loss": 0.8697,
+      "step": 1250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015617005819659613,
+      "loss": 0.8548,
+      "step": 1251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015609913831348774,
+      "loss": 0.884,
+      "step": 1252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015602817723286738,
+      "loss": 0.8884,
+      "step": 1253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001559571750068468,
+      "loss": 0.8589,
+      "step": 1254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015588613168756782,
+      "loss": 0.3395,
+      "step": 1255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001558150473272026,
+      "loss": 0.7986,
+      "step": 1256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015574392197795328,
+      "loss": 0.3293,
+      "step": 1257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015567275569205218,
+      "loss": 0.8328,
+      "step": 1258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015560154852176165,
+      "loss": 0.8558,
+      "step": 1259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015553030051937414,
+      "loss": 0.8341,
+      "step": 1260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015545901173721197,
+      "loss": 0.8752,
+      "step": 1261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001553876822276275,
+      "loss": 0.8829,
+      "step": 1262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015531631204300297,
+      "loss": 0.8441,
+      "step": 1263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001552449012357505,
+      "loss": 0.8638,
+      "step": 1264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015517344985831194,
+      "loss": 0.8438,
+      "step": 1265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015510195796315915,
+      "loss": 0.9085,
+      "step": 1266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550304256027935,
+      "loss": 0.8608,
+      "step": 1267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015495885282974627,
+      "loss": 0.7987,
+      "step": 1268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015488723969657838,
+      "loss": 0.8341,
+      "step": 1269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015481558625588023,
+      "loss": 0.7801,
+      "step": 1270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000154743892560272,
+      "loss": 0.8851,
+      "step": 1271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015467215866240338,
+      "loss": 0.8172,
+      "step": 1272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015460038461495362,
+      "loss": 0.8398,
+      "step": 1273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545285704706313,
+      "loss": 0.8839,
+      "step": 1274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015445671628217466,
+      "loss": 0.8594,
+      "step": 1275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015438482210235118,
+      "loss": 0.8841,
+      "step": 1276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015431288798395783,
+      "loss": 0.8576,
+      "step": 1277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015424091397982082,
+      "loss": 0.8932,
+      "step": 1278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001541689001427957,
+      "loss": 0.8531,
+      "step": 1279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015409684652576718,
+      "loss": 0.8479,
+      "step": 1280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540247531816494,
+      "loss": 0.8743,
+      "step": 1281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015395262016338541,
+      "loss": 0.9054,
+      "step": 1282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015388044752394757,
+      "loss": 0.8892,
+      "step": 1283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015380823531633729,
+      "loss": 0.8863,
+      "step": 1284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153735983593585,
+      "loss": 0.8712,
+      "step": 1285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015366369240875022,
+      "loss": 0.811,
+      "step": 1286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535913618149214,
+      "loss": 0.8586,
+      "step": 1287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535189918652159,
+      "loss": 0.3374,
+      "step": 1288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001534465826127801,
+      "loss": 0.8624,
+      "step": 1289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015337413411078914,
+      "loss": 0.3367,
+      "step": 1290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153301646412447,
+      "loss": 0.8567,
+      "step": 1291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015322911957098642,
+      "loss": 0.7878,
+      "step": 1292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015315655363966896,
+      "loss": 0.8655,
+      "step": 1293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015308394867178486,
+      "loss": 0.8334,
+      "step": 1294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153011304720653,
+      "loss": 0.8695,
+      "step": 1295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001529386218396209,
+      "loss": 0.8399,
+      "step": 1296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015286590008206465,
+      "loss": 0.8327,
+      "step": 1297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015279313950138893,
+      "loss": 0.8488,
+      "step": 1298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015272034015102696,
+      "loss": 0.8474,
+      "step": 1299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015264750208444027,
+      "loss": 0.9043,
+      "step": 1300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015257462535511903,
+      "loss": 0.8687,
+      "step": 1301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015250171001658172,
+      "loss": 0.8804,
+      "step": 1302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015242875612237508,
+      "loss": 0.8646,
+      "step": 1303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015235576372607429,
+      "loss": 0.8933,
+      "step": 1304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522827328812827,
+      "loss": 0.7983,
+      "step": 1305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015220966364163215,
+      "loss": 0.8682,
+      "step": 1306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015213655606078223,
+      "loss": 0.8567,
+      "step": 1307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015206341019242107,
+      "loss": 0.898,
+      "step": 1308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519902260902648,
+      "loss": 0.8798,
+      "step": 1309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015191700380805752,
+      "loss": 0.8235,
+      "step": 1310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015184374339957159,
+      "loss": 0.8696,
+      "step": 1311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015177044491860713,
+      "loss": 0.8298,
+      "step": 1312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015169710841899238,
+      "loss": 0.8628,
+      "step": 1313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015162373395458345,
+      "loss": 0.8693,
+      "step": 1314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515503215792643,
+      "loss": 0.8384,
+      "step": 1315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015147687134694677,
+      "loss": 0.8778,
+      "step": 1316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015140338331157054,
+      "loss": 0.8601,
+      "step": 1317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015132985752710294,
+      "loss": 0.8716,
+      "step": 1318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512562940475391,
+      "loss": 0.8836,
+      "step": 1319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015118269292690184,
+      "loss": 0.8801,
+      "step": 1320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015110905421924156,
+      "loss": 0.8519,
+      "step": 1321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001510353779786363,
+      "loss": 0.8153,
+      "step": 1322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015096166425919175,
+      "loss": 0.8514,
+      "step": 1323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015088791311504096,
+      "loss": 0.8447,
+      "step": 1324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015081412460034454,
+      "loss": 0.8286,
+      "step": 1325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015074029876929057,
+      "loss": 0.8853,
+      "step": 1326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001506664356760945,
+      "loss": 0.3214,
+      "step": 1327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015059253537499915,
+      "loss": 0.8134,
+      "step": 1328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505185979202747,
+      "loss": 0.8472,
+      "step": 1329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015044462336621852,
+      "loss": 0.8401,
+      "step": 1330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015037061176715537,
+      "loss": 0.8498,
+      "step": 1331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015029656317743704,
+      "loss": 0.875,
+      "step": 1332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015022247765144264,
+      "loss": 0.8257,
+      "step": 1333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501483552435783,
+      "loss": 0.8876,
+      "step": 1334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015007419600827736,
+      "loss": 0.8594,
+      "step": 1335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.8966,
+      "step": 1336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001499257672732336,
+      "loss": 0.8224,
+      "step": 1337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014985149788249248,
+      "loss": 0.8132,
+      "step": 1338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014977719188231768,
+      "loss": 0.8066,
+      "step": 1339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014970284932727744,
+      "loss": 0.3598,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014962847027196662,
+      "loss": 0.8135,
+      "step": 1341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014955405477100691,
+      "loss": 0.8588,
+      "step": 1342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014947960287904687,
+      "loss": 0.8664,
+      "step": 1343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014940511465076173,
+      "loss": 0.835,
+      "step": 1344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014933059014085332,
+      "loss": 0.8478,
+      "step": 1345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014925602940405025,
+      "loss": 0.8319,
+      "step": 1346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014918143249510762,
+      "loss": 0.8661,
+      "step": 1347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014910679946880721,
+      "loss": 0.9276,
+      "step": 1348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014903213037995724,
+      "loss": 0.8578,
+      "step": 1349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001489574252833924,
+      "loss": 0.8291,
+      "step": 1350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014888268423397384,
+      "loss": 0.3399,
+      "step": 1351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488079072865892,
+      "loss": 0.8357,
+      "step": 1352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014873309449615237,
+      "loss": 0.8892,
+      "step": 1353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001486582459176036,
+      "loss": 0.8464,
+      "step": 1354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014858336160590942,
+      "loss": 0.8512,
+      "step": 1355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014850844161606264,
+      "loss": 0.8473,
+      "step": 1356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484334860030822,
+      "loss": 0.8811,
+      "step": 1357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014835849482201322,
+      "loss": 0.8717,
+      "step": 1358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014828346812792702,
+      "loss": 0.8859,
+      "step": 1359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482084059759209,
+      "loss": 0.8696,
+      "step": 1360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014813330842111823,
+      "loss": 0.8596,
+      "step": 1361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014805817551866838,
+      "loss": 0.8942,
+      "step": 1362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001479830073237467,
+      "loss": 0.8717,
+      "step": 1363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014790780389155447,
+      "loss": 0.8646,
+      "step": 1364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014783256527731877,
+      "loss": 0.8687,
+      "step": 1365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014775729153629257,
+      "loss": 0.8047,
+      "step": 1366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014768198272375465,
+      "loss": 0.803,
+      "step": 1367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001476066388950095,
+      "loss": 0.8318,
+      "step": 1368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001475312601053874,
+      "loss": 0.8486,
+      "step": 1369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001474558464102442,
+      "loss": 0.8266,
+      "step": 1370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014738039786496148,
+      "loss": 0.854,
+      "step": 1371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014730491452494632,
+      "loss": 0.8594,
+      "step": 1372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014722939644563145,
+      "loss": 0.851,
+      "step": 1373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014715384368247506,
+      "loss": 0.8515,
+      "step": 1374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014707825629096084,
+      "loss": 0.8807,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001470026343265978,
+      "loss": 0.8931,
+      "step": 1376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001469269778449205,
+      "loss": 0.8714,
+      "step": 1377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014685128690148882,
+      "loss": 0.8528,
+      "step": 1378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014677556155188776,
+      "loss": 0.3043,
+      "step": 1379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466998018517278,
+      "loss": 0.8361,
+      "step": 1380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014662400785664463,
+      "loss": 0.888,
+      "step": 1381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014654817962229898,
+      "loss": 0.8334,
+      "step": 1382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014647231720437686,
+      "loss": 0.3207,
+      "step": 1383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014639642065858934,
+      "loss": 0.8493,
+      "step": 1384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014632049004067253,
+      "loss": 0.8467,
+      "step": 1385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001462445254063876,
+      "loss": 0.8347,
+      "step": 1386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014616852681152066,
+      "loss": 0.8327,
+      "step": 1387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014609249431188278,
+      "loss": 0.8975,
+      "step": 1388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014601642796331,
+      "loss": 0.887,
+      "step": 1389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014594032782166307,
+      "loss": 0.8582,
+      "step": 1390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014586419394282762,
+      "loss": 0.8727,
+      "step": 1391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014578802638271415,
+      "loss": 0.838,
+      "step": 1392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014571182519725775,
+      "loss": 0.3264,
+      "step": 1393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001456355904424183,
+      "loss": 0.8818,
+      "step": 1394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014555932217418026,
+      "loss": 0.8314,
+      "step": 1395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454830204485528,
+      "loss": 0.8576,
+      "step": 1396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014540668532156956,
+      "loss": 0.8389,
+      "step": 1397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014533031684928874,
+      "loss": 0.8152,
+      "step": 1398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014525391508779308,
+      "loss": 0.8631,
+      "step": 1399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014517748009318975,
+      "loss": 0.8702,
+      "step": 1400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014510101192161018,
+      "loss": 0.8682,
+      "step": 1401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014502451062921038,
+      "loss": 0.8023,
+      "step": 1402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014494797627217062,
+      "loss": 0.8354,
+      "step": 1403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014487140890669532,
+      "loss": 0.818,
+      "step": 1404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014479480858901327,
+      "loss": 0.8594,
+      "step": 1405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014471817537537747,
+      "loss": 0.8273,
+      "step": 1406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014464150932206506,
+      "loss": 0.8618,
+      "step": 1407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014456481048537716,
+      "loss": 0.866,
+      "step": 1408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014448807892163912,
+      "loss": 0.8191,
+      "step": 1409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014441131468720037,
+      "loss": 0.8625,
+      "step": 1410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014433451783843414,
+      "loss": 0.8133,
+      "step": 1411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014425768843173776,
+      "loss": 0.8755,
+      "step": 1412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001441808265235324,
+      "loss": 0.882,
+      "step": 1413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014410393217026318,
+      "loss": 0.8637,
+      "step": 1414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001440270054283989,
+      "loss": 0.8423,
+      "step": 1415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439500463544323,
+      "loss": 0.8365,
+      "step": 1416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001438730550048798,
+      "loss": 0.8291,
+      "step": 1417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001437960314362815,
+      "loss": 0.8496,
+      "step": 1418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014371897570520116,
+      "loss": 0.8559,
+      "step": 1419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014364188786822622,
+      "loss": 0.3612,
+      "step": 1420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014356476798196762,
+      "loss": 0.8296,
+      "step": 1421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014348761610305993,
+      "loss": 0.8381,
+      "step": 1422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014341043228816112,
+      "loss": 0.8922,
+      "step": 1423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014333321659395264,
+      "loss": 0.8637,
+      "step": 1424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014325596907713937,
+      "loss": 0.8207,
+      "step": 1425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001431786897944496,
+      "loss": 0.8473,
+      "step": 1426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014310137880263482,
+      "loss": 0.8218,
+      "step": 1427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014302403615846994,
+      "loss": 0.8411,
+      "step": 1428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014294666191875304,
+      "loss": 0.8986,
+      "step": 1429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014286925614030542,
+      "loss": 0.8465,
+      "step": 1430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014279181887997159,
+      "loss": 0.9058,
+      "step": 1431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014271435019461905,
+      "loss": 0.8273,
+      "step": 1432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014263685014113858,
+      "loss": 0.8808,
+      "step": 1433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014255931877644372,
+      "loss": 0.835,
+      "step": 1434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001424817561574713,
+      "loss": 0.8715,
+      "step": 1435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001424041623411809,
+      "loss": 0.8293,
+      "step": 1436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014232653738455511,
+      "loss": 0.3387,
+      "step": 1437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001422488813445993,
+      "loss": 0.8694,
+      "step": 1438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014217119427834178,
+      "loss": 0.8704,
+      "step": 1439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001420934762428335,
+      "loss": 0.8033,
+      "step": 1440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014201572729514834,
+      "loss": 0.8234,
+      "step": 1441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014193794749238267,
+      "loss": 0.8586,
+      "step": 1442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014186013689165573,
+      "loss": 0.8712,
+      "step": 1443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417822955501092,
+      "loss": 0.8359,
+      "step": 1444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014170442352490746,
+      "loss": 0.8686,
+      "step": 1445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014162652087323734,
+      "loss": 0.8349,
+      "step": 1446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014154858765230818,
+      "loss": 0.8563,
+      "step": 1447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014147062391935182,
+      "loss": 0.8623,
+      "step": 1448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014139262973162246,
+      "loss": 0.8392,
+      "step": 1449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014131460514639664,
+      "loss": 0.8144,
+      "step": 1450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014123655022097332,
+      "loss": 0.2907,
+      "step": 1451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001411584650126736,
+      "loss": 0.8957,
+      "step": 1452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014108034957884094,
+      "loss": 0.3172,
+      "step": 1453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014100220397684102,
+      "loss": 0.8176,
+      "step": 1454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001409240282640615,
+      "loss": 0.8786,
+      "step": 1455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001408458224979123,
+      "loss": 0.8899,
+      "step": 1456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014076758673582537,
+      "loss": 0.8291,
+      "step": 1457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014068932103525472,
+      "loss": 0.8298,
+      "step": 1458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001406110254536763,
+      "loss": 0.8215,
+      "step": 1459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014053270004858796,
+      "loss": 0.8553,
+      "step": 1460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404543448775096,
+      "loss": 0.8924,
+      "step": 1461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014037595999798284,
+      "loss": 0.8667,
+      "step": 1462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014029754546757115,
+      "loss": 0.8784,
+      "step": 1463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014021910134385984,
+      "loss": 0.8805,
+      "step": 1464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001401406276844559,
+      "loss": 0.8522,
+      "step": 1465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014006212454698797,
+      "loss": 0.2912,
+      "step": 1466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013998359198910644,
+      "loss": 0.818,
+      "step": 1467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013990503006848317,
+      "loss": 0.8901,
+      "step": 1468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398264388428117,
+      "loss": 0.8473,
+      "step": 1469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013974781836980713,
+      "loss": 0.8736,
+      "step": 1470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013966916870720582,
+      "loss": 0.8631,
+      "step": 1471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001395904899127658,
+      "loss": 0.7925,
+      "step": 1472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013951178204426635,
+      "loss": 0.8437,
+      "step": 1473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394330451595082,
+      "loss": 0.8438,
+      "step": 1474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001393542793163133,
+      "loss": 0.861,
+      "step": 1475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001392754845725249,
+      "loss": 0.3217,
+      "step": 1476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013919666098600753,
+      "loss": 0.8713,
+      "step": 1477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013911780861464675,
+      "loss": 0.84,
+      "step": 1478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013903892751634947,
+      "loss": 0.8569,
+      "step": 1479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013896001774904354,
+      "loss": 0.8811,
+      "step": 1480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001388810793706779,
+      "loss": 0.9044,
+      "step": 1481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001388021124392225,
+      "loss": 0.8865,
+      "step": 1482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013872311701266835,
+      "loss": 0.8586,
+      "step": 1483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013864409314902714,
+      "loss": 0.7952,
+      "step": 1484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013856504090633176,
+      "loss": 0.8812,
+      "step": 1485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013848596034263572,
+      "loss": 0.8541,
+      "step": 1486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013840685151601342,
+      "loss": 0.8615,
+      "step": 1487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013832771448455997,
+      "loss": 0.8692,
+      "step": 1488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013824854930639122,
+      "loss": 0.8646,
+      "step": 1489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013816935603964373,
+      "loss": 0.8191,
+      "step": 1490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013809013474247456,
+      "loss": 0.8143,
+      "step": 1491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013801088547306148,
+      "loss": 0.8981,
+      "step": 1492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013793160828960276,
+      "loss": 0.8263,
+      "step": 1493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013785230325031719,
+      "loss": 0.8793,
+      "step": 1494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013777297041344395,
+      "loss": 0.8689,
+      "step": 1495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001376936098372427,
+      "loss": 0.8634,
+      "step": 1496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013761422157999346,
+      "loss": 0.836,
+      "step": 1497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013753480569999655,
+      "loss": 0.8327,
+      "step": 1498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013745536225557258,
+      "loss": 0.8551,
+      "step": 1499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013737589130506246,
+      "loss": 0.8666,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001372963929068272,
+      "loss": 0.8464,
+      "step": 1501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013721686711924805,
+      "loss": 0.8266,
+      "step": 1502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013713731400072634,
+      "loss": 0.8623,
+      "step": 1503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001370577336096835,
+      "loss": 0.8694,
+      "step": 1504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013697812600456093,
+      "loss": 0.8659,
+      "step": 1505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013689849124382005,
+      "loss": 0.8698,
+      "step": 1506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013681882938594223,
+      "loss": 0.8726,
+      "step": 1507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013673914048942877,
+      "loss": 0.8535,
+      "step": 1508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013665942461280075,
+      "loss": 0.7804,
+      "step": 1509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001365796818145991,
+      "loss": 0.9065,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013649991215338448,
+      "loss": 0.8802,
+      "step": 1511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013642011568773744,
+      "loss": 0.8316,
+      "step": 1512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013634029247625796,
+      "loss": 0.872,
+      "step": 1513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013626044257756585,
+      "loss": 0.8476,
+      "step": 1514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013618056605030048,
+      "loss": 0.8404,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013610066295312073,
+      "loss": 0.8888,
+      "step": 1516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013602073334470505,
+      "loss": 0.8548,
+      "step": 1517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013594077728375128,
+      "loss": 0.3019,
+      "step": 1518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001358607948289768,
+      "loss": 0.8748,
+      "step": 1519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013578078603911827,
+      "loss": 0.3317,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001357007509729317,
+      "loss": 0.8673,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013562068968919248,
+      "loss": 0.898,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001355406022466952,
+      "loss": 0.8842,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013546048870425356,
+      "loss": 0.7847,
+      "step": 1524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013538034912070066,
+      "loss": 0.8649,
+      "step": 1525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001353001835548885,
+      "loss": 0.8665,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013521999206568826,
+      "loss": 0.8474,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013513977471199016,
+      "loss": 0.8725,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001350595315527034,
+      "loss": 0.8295,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013497926264675612,
+      "loss": 0.3183,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013489896805309542,
+      "loss": 0.7976,
+      "step": 1531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013481864783068716,
+      "loss": 0.8295,
+      "step": 1532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013473830203851613,
+      "loss": 0.855,
+      "step": 1533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013465793073558585,
+      "loss": 0.8796,
+      "step": 1534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013457753398091855,
+      "loss": 0.8717,
+      "step": 1535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001344971118335552,
+      "loss": 0.8689,
+      "step": 1536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013441666435255539,
+      "loss": 0.8633,
+      "step": 1537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001343361915969973,
+      "loss": 0.8479,
+      "step": 1538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342556936259777,
+      "loss": 0.8385,
+      "step": 1539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001341751704986119,
+      "loss": 0.8651,
+      "step": 1540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013409462227403356,
+      "loss": 0.839,
+      "step": 1541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000134014049011395,
+      "loss": 0.8789,
+      "step": 1542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013393345076986667,
+      "loss": 0.8869,
+      "step": 1543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013385282760863758,
+      "loss": 0.8154,
+      "step": 1544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013377217958691483,
+      "loss": 0.3394,
+      "step": 1545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000133691506763924,
+      "loss": 0.855,
+      "step": 1546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336108091989087,
+      "loss": 0.8947,
+      "step": 1547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013353008695113082,
+      "loss": 0.8808,
+      "step": 1548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001334493400798703,
+      "loss": 0.8549,
+      "step": 1549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013336856864442525,
+      "loss": 0.8077,
+      "step": 1550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013328777270411174,
+      "loss": 0.8075,
+      "step": 1551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013320695231826388,
+      "loss": 0.8436,
+      "step": 1552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001331261075462337,
+      "loss": 0.8372,
+      "step": 1553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013304523844739124,
+      "loss": 0.8707,
+      "step": 1554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013296434508112422,
+      "loss": 0.8439,
+      "step": 1555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001328834275068383,
+      "loss": 0.8681,
+      "step": 1556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001328024857839569,
+      "loss": 0.8382,
+      "step": 1557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013272151997192127,
+      "loss": 0.8444,
+      "step": 1558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013264053013019013,
+      "loss": 0.8415,
+      "step": 1559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013255951631824004,
+      "loss": 0.854,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013247847859556506,
+      "loss": 0.8367,
+      "step": 1561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001323974170216769,
+      "loss": 0.8346,
+      "step": 1562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013231633165610468,
+      "loss": 0.8244,
+      "step": 1563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013223522255839508,
+      "loss": 0.8343,
+      "step": 1564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321540897881121,
+      "loss": 0.8533,
+      "step": 1565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013207293340483726,
+      "loss": 0.8319,
+      "step": 1566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013199175346816937,
+      "loss": 0.8468,
+      "step": 1567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013191055003772446,
+      "loss": 0.914,
+      "step": 1568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001318293231731359,
+      "loss": 0.8519,
+      "step": 1569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013174807293405428,
+      "loss": 0.8397,
+      "step": 1570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013166679938014726,
+      "loss": 0.8432,
+      "step": 1571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001315855025710997,
+      "loss": 0.8438,
+      "step": 1572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013150418256661358,
+      "loss": 0.8166,
+      "step": 1573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001314228394264078,
+      "loss": 0.8618,
+      "step": 1574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013134147321021829,
+      "loss": 0.8302,
+      "step": 1575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013126008397779796,
+      "loss": 0.8344,
+      "step": 1576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013117867178891662,
+      "loss": 0.3408,
+      "step": 1577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001310972367033609,
+      "loss": 0.8477,
+      "step": 1578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013101577878093425,
+      "loss": 0.7974,
+      "step": 1579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013093429808145693,
+      "loss": 0.8689,
+      "step": 1580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013085279466476594,
+      "loss": 0.8686,
+      "step": 1581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013077126859071484,
+      "loss": 0.8662,
+      "step": 1582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000130689719919174,
+      "loss": 0.872,
+      "step": 1583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013060814871003026,
+      "loss": 0.8162,
+      "step": 1584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013052655502318703,
+      "loss": 0.8565,
+      "step": 1585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013044493891856428,
+      "loss": 0.8585,
+      "step": 1586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001303633004560984,
+      "loss": 0.8402,
+      "step": 1587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013028163969574223,
+      "loss": 0.8866,
+      "step": 1588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013019995669746493,
+      "loss": 0.8977,
+      "step": 1589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013011825152125204,
+      "loss": 0.8345,
+      "step": 1590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013003652422710537,
+      "loss": 0.8191,
+      "step": 1591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000129954774875043,
+      "loss": 0.3495,
+      "step": 1592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012987300352509915,
+      "loss": 0.8485,
+      "step": 1593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012979121023732424,
+      "loss": 0.326,
+      "step": 1594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012970939507178482,
+      "loss": 0.8706,
+      "step": 1595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012962755808856342,
+      "loss": 0.3054,
+      "step": 1596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012954569934775867,
+      "loss": 0.8537,
+      "step": 1597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012946381890948516,
+      "loss": 0.8177,
+      "step": 1598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012938191683387342,
+      "loss": 0.8951,
+      "step": 1599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001292999931810698,
+      "loss": 0.3189,
+      "step": 1600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001292180480112367,
+      "loss": 0.8757,
+      "step": 1601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012913608138455203,
+      "loss": 0.3406,
+      "step": 1602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001290540933612096,
+      "loss": 0.9095,
+      "step": 1603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012897208400141906,
+      "loss": 0.849,
+      "step": 1604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001288900533654055,
+      "loss": 0.8153,
+      "step": 1605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012880800151340982,
+      "loss": 0.8273,
+      "step": 1606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012872592850568832,
+      "loss": 0.3314,
+      "step": 1607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012864383440251302,
+      "loss": 0.8434,
+      "step": 1608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012856171926417133,
+      "loss": 0.8991,
+      "step": 1609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001284795831509661,
+      "loss": 0.8402,
+      "step": 1610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012839742612321564,
+      "loss": 0.8417,
+      "step": 1611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012831524824125356,
+      "loss": 0.8199,
+      "step": 1612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012823304956542889,
+      "loss": 0.8215,
+      "step": 1613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012815083015610572,
+      "loss": 0.8738,
+      "step": 1614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012806859007366355,
+      "loss": 0.8533,
+      "step": 1615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001279863293784971,
+      "loss": 0.8391,
+      "step": 1616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012790404813101607,
+      "loss": 0.8257,
+      "step": 1617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001278217463916453,
+      "loss": 0.818,
+      "step": 1618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277394242208247,
+      "loss": 0.8345,
+      "step": 1619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012765708167900925,
+      "loss": 0.8284,
+      "step": 1620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012757471882666874,
+      "loss": 0.8657,
+      "step": 1621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012749233572428804,
+      "loss": 0.8816,
+      "step": 1622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012740993243236676,
+      "loss": 0.8547,
+      "step": 1623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001273275090114194,
+      "loss": 0.8077,
+      "step": 1624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012724506552197525,
+      "loss": 0.8419,
+      "step": 1625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012716260202457827,
+      "loss": 0.8431,
+      "step": 1626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001270801185797872,
+      "loss": 0.3439,
+      "step": 1627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012699761524817538,
+      "loss": 0.8336,
+      "step": 1628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012691509209033076,
+      "loss": 0.7829,
+      "step": 1629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012683254916685588,
+      "loss": 0.8326,
+      "step": 1630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001267499865383677,
+      "loss": 0.7998,
+      "step": 1631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001266674042654978,
+      "loss": 0.8166,
+      "step": 1632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012658480240889202,
+      "loss": 0.8384,
+      "step": 1633
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012650218102921076,
+      "loss": 0.8414,
+      "step": 1634
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012641954018712863,
+      "loss": 0.3303,
+      "step": 1635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263368799433345,
+      "loss": 0.8333,
+      "step": 1636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012625420035853165,
+      "loss": 0.8126,
+      "step": 1637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012617150149343744,
+      "loss": 0.8427,
+      "step": 1638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001260887834087834,
+      "loss": 0.8377,
+      "step": 1639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012600604616531518,
+      "loss": 0.7611,
+      "step": 1640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012592328982379255,
+      "loss": 0.7999,
+      "step": 1641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012584051444498927,
+      "loss": 0.8638,
+      "step": 1642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012575772008969303,
+      "loss": 0.8564,
+      "step": 1643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012567490681870554,
+      "loss": 0.8849,
+      "step": 1644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012559207469284232,
+      "loss": 0.8645,
+      "step": 1645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012550922377293282,
+      "loss": 0.8526,
+      "step": 1646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012542635411982026,
+      "loss": 0.8638,
+      "step": 1647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001253434657943616,
+      "loss": 0.8331,
+      "step": 1648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001252605588574275,
+      "loss": 0.8918,
+      "step": 1649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001251776333699023,
+      "loss": 0.8234,
+      "step": 1650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012509468939268402,
+      "loss": 0.8156,
+      "step": 1651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012501172698668416,
+      "loss": 0.8385,
+      "step": 1652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001249287462128279,
+      "loss": 0.8129,
+      "step": 1653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012484574713205368,
+      "loss": 0.3421,
+      "step": 1654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012476272980531362,
+      "loss": 0.8971,
+      "step": 1655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012467969429357304,
+      "loss": 0.8564,
+      "step": 1656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001245966406578108,
+      "loss": 0.8452,
+      "step": 1657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012451356895901893,
+      "loss": 0.8531,
+      "step": 1658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012443047925820276,
+      "loss": 0.8871,
+      "step": 1659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012434737161638097,
+      "loss": 0.8467,
+      "step": 1660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012426424609458518,
+      "loss": 0.8345,
+      "step": 1661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012418110275386028,
+      "loss": 0.3471,
+      "step": 1662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012409794165526425,
+      "loss": 0.8109,
+      "step": 1663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001240147628598681,
+      "loss": 0.8919,
+      "step": 1664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001239315664287558,
+      "loss": 0.8513,
+      "step": 1665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012384835242302427,
+      "loss": 0.8602,
+      "step": 1666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012376512090378342,
+      "loss": 0.8579,
+      "step": 1667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012368187193215598,
+      "loss": 0.8513,
+      "step": 1668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012359860556927736,
+      "loss": 0.8368,
+      "step": 1669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012351532187629595,
+      "loss": 0.8735,
+      "step": 1670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012343202091437278,
+      "loss": 0.8465,
+      "step": 1671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012334870274468155,
+      "loss": 0.8486,
+      "step": 1672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001232653674284086,
+      "loss": 0.8868,
+      "step": 1673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012318201502675285,
+      "loss": 0.8688,
+      "step": 1674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012309864560092578,
+      "loss": 0.845,
+      "step": 1675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012301525921215145,
+      "loss": 0.812,
+      "step": 1676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001229318559216662,
+      "loss": 0.8333,
+      "step": 1677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012284843579071897,
+      "loss": 0.8832,
+      "step": 1678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012276499888057093,
+      "loss": 0.831,
+      "step": 1679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012268154525249563,
+      "loss": 0.8165,
+      "step": 1680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012259807496777894,
+      "loss": 0.8001,
+      "step": 1681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012251458808771893,
+      "loss": 0.8457,
+      "step": 1682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012243108467362572,
+      "loss": 0.8732,
+      "step": 1683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001223475647868218,
+      "loss": 0.8482,
+      "step": 1684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012226402848864162,
+      "loss": 0.8438,
+      "step": 1685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221804758404317,
+      "loss": 0.9129,
+      "step": 1686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001220969069035506,
+      "loss": 0.8483,
+      "step": 1687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012201332173936882,
+      "loss": 0.8371,
+      "step": 1688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001219297204092688,
+      "loss": 0.8731,
+      "step": 1689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012184610297464476,
+      "loss": 0.8382,
+      "step": 1690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012176246949690287,
+      "loss": 0.8249,
+      "step": 1691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012167882003746103,
+      "loss": 0.8112,
+      "step": 1692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012159515465774889,
+      "loss": 0.8636,
+      "step": 1693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001215114734192077,
+      "loss": 0.8419,
+      "step": 1694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001214277763832905,
+      "loss": 0.8408,
+      "step": 1695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012134406361146189,
+      "loss": 0.8466,
+      "step": 1696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012126033516519792,
+      "loss": 0.8258,
+      "step": 1697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001211765911059863,
+      "loss": 0.8584,
+      "step": 1698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012109283149532607,
+      "loss": 0.8339,
+      "step": 1699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012100905639472779,
+      "loss": 0.8361,
+      "step": 1700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001209252658657134,
+      "loss": 0.8423,
+      "step": 1701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012084145996981608,
+      "loss": 0.8613,
+      "step": 1702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012075763876858037,
+      "loss": 0.8189,
+      "step": 1703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012067380232356203,
+      "loss": 0.8734,
+      "step": 1704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012058995069632797,
+      "loss": 0.869,
+      "step": 1705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012050608394845637,
+      "loss": 0.9032,
+      "step": 1706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012042220214153637,
+      "loss": 0.771,
+      "step": 1707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012033830533716825,
+      "loss": 0.8558,
+      "step": 1708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001202543935969633,
+      "loss": 0.8063,
+      "step": 1709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012017046698254375,
+      "loss": 0.8407,
+      "step": 1710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012008652555554281,
+      "loss": 0.8117,
+      "step": 1711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012000256937760445,
+      "loss": 0.8958,
+      "step": 1712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001199185985103836,
+      "loss": 0.7975,
+      "step": 1713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011983461301554595,
+      "loss": 0.8497,
+      "step": 1714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011975061295476789,
+      "loss": 0.8701,
+      "step": 1715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011966659838973649,
+      "loss": 0.8071,
+      "step": 1716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011958256938214954,
+      "loss": 0.8509,
+      "step": 1717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011949852599371545,
+      "loss": 0.8194,
+      "step": 1718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011941446828615304,
+      "loss": 0.8947,
+      "step": 1719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001193303963211918,
+      "loss": 0.8719,
+      "step": 1720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011924631016057168,
+      "loss": 0.8311,
+      "step": 1721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011916220986604296,
+      "loss": 0.8159,
+      "step": 1722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011907809549936639,
+      "loss": 0.8345,
+      "step": 1723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011899396712231298,
+      "loss": 0.8255,
+      "step": 1724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011890982479666412,
+      "loss": 0.8359,
+      "step": 1725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011882566858421135,
+      "loss": 0.832,
+      "step": 1726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001187414985467565,
+      "loss": 0.3403,
+      "step": 1727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011865731474611142,
+      "loss": 0.8214,
+      "step": 1728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011857311724409821,
+      "loss": 0.8305,
+      "step": 1729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011848890610254894,
+      "loss": 0.8579,
+      "step": 1730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011840468138330578,
+      "loss": 0.3323,
+      "step": 1731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011832044314822077,
+      "loss": 0.8591,
+      "step": 1732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011823619145915593,
+      "loss": 0.836,
+      "step": 1733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011815192637798314,
+      "loss": 0.8021,
+      "step": 1734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011806764796658418,
+      "loss": 0.8905,
+      "step": 1735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011798335628685054,
+      "loss": 0.8368,
+      "step": 1736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011789905140068346,
+      "loss": 0.8355,
+      "step": 1737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011781473336999392,
+      "loss": 0.8461,
+      "step": 1738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011773040225670256,
+      "loss": 0.8302,
+      "step": 1739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011764605812273956,
+      "loss": 0.8714,
+      "step": 1740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011756170103004472,
+      "loss": 0.9283,
+      "step": 1741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011747733104056733,
+      "loss": 0.8445,
+      "step": 1742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011739294821626621,
+      "loss": 0.9032,
+      "step": 1743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011730855261910949,
+      "loss": 0.8721,
+      "step": 1744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011722414431107478,
+      "loss": 0.8818,
+      "step": 1745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011713972335414895,
+      "loss": 0.8013,
+      "step": 1746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011705528981032827,
+      "loss": 0.8562,
+      "step": 1747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011697084374161816,
+      "loss": 0.8258,
+      "step": 1748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011688638521003322,
+      "loss": 0.8364,
+      "step": 1749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011680191427759728,
+      "loss": 0.8077,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001167174310063432,
+      "loss": 0.8493,
+      "step": 1751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011663293545831302,
+      "loss": 0.8262,
+      "step": 1752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011654842769555765,
+      "loss": 0.826,
+      "step": 1753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011646390778013704,
+      "loss": 0.861,
+      "step": 1754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001163793757741201,
+      "loss": 0.8206,
+      "step": 1755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011629483173958455,
+      "loss": 0.8114,
+      "step": 1756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011621027573861698,
+      "loss": 0.8244,
+      "step": 1757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011612570783331279,
+      "loss": 0.8453,
+      "step": 1758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011604112808577603,
+      "loss": 0.9129,
+      "step": 1759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011595653655811958,
+      "loss": 0.8356,
+      "step": 1760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011587193331246486,
+      "loss": 0.8649,
+      "step": 1761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011578731841094195,
+      "loss": 0.3396,
+      "step": 1762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011570269191568945,
+      "loss": 0.8037,
+      "step": 1763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011561805388885454,
+      "loss": 0.8682,
+      "step": 1764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011553340439259286,
+      "loss": 0.8357,
+      "step": 1765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011544874348906836,
+      "loss": 0.8203,
+      "step": 1766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011536407124045348,
+      "loss": 0.8186,
+      "step": 1767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011527938770892899,
+      "loss": 0.8371,
+      "step": 1768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011519469295668388,
+      "loss": 0.8762,
+      "step": 1769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011510998704591542,
+      "loss": 0.8628,
+      "step": 1770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011502527003882908,
+      "loss": 0.8286,
+      "step": 1771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011494054199763851,
+      "loss": 0.8343,
+      "step": 1772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011485580298456535,
+      "loss": 0.8055,
+      "step": 1773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011477105306183942,
+      "loss": 0.8609,
+      "step": 1774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011468629229169847,
+      "loss": 0.7763,
+      "step": 1775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011460152073638828,
+      "loss": 0.8024,
+      "step": 1776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011451673845816251,
+      "loss": 0.8589,
+      "step": 1777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011443194551928266,
+      "loss": 0.8556,
+      "step": 1778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011434714198201819,
+      "loss": 0.8666,
+      "step": 1779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011426232790864617,
+      "loss": 0.8524,
+      "step": 1780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011417750336145155,
+      "loss": 0.8722,
+      "step": 1781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011409266840272685,
+      "loss": 0.7817,
+      "step": 1782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011400782309477239,
+      "loss": 0.8616,
+      "step": 1783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011392296749989591,
+      "loss": 0.8012,
+      "step": 1784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011383810168041285,
+      "loss": 0.8291,
+      "step": 1785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001137532256986461,
+      "loss": 0.872,
+      "step": 1786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000113668339616926,
+      "loss": 0.8494,
+      "step": 1787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011358344349759031,
+      "loss": 0.7847,
+      "step": 1788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011349853740298419,
+      "loss": 0.8101,
+      "step": 1789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011341362139546019,
+      "loss": 0.8244,
+      "step": 1790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001133286955373779,
+      "loss": 0.8561,
+      "step": 1791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001132437598911044,
+      "loss": 0.8056,
+      "step": 1792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011315881451901384,
+      "loss": 0.8559,
+      "step": 1793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011307385948348762,
+      "loss": 0.869,
+      "step": 1794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000112988894846914,
+      "loss": 0.8572,
+      "step": 1795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011290392067168852,
+      "loss": 0.3221,
+      "step": 1796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011281893702021365,
+      "loss": 0.8006,
+      "step": 1797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011273394395489882,
+      "loss": 0.8743,
+      "step": 1798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011264894153816033,
+      "loss": 0.7828,
+      "step": 1799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011256392983242143,
+      "loss": 0.8263,
+      "step": 1800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011247890890011212,
+      "loss": 0.8622,
+      "step": 1801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011239387880366925,
+      "loss": 0.8448,
+      "step": 1802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011230883960553627,
+      "loss": 0.8378,
+      "step": 1803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011222379136816345,
+      "loss": 0.8189,
+      "step": 1804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011213873415400766,
+      "loss": 0.8101,
+      "step": 1805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001120536680255323,
+      "loss": 0.8738,
+      "step": 1806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011196859304520742,
+      "loss": 0.8594,
+      "step": 1807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001118835092755095,
+      "loss": 0.8927,
+      "step": 1808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011179841677892142,
+      "loss": 0.3219,
+      "step": 1809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011171331561793263,
+      "loss": 0.8077,
+      "step": 1810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011162820585503881,
+      "loss": 0.3127,
+      "step": 1811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000111543087552742,
+      "loss": 0.8051,
+      "step": 1812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011145796077355051,
+      "loss": 0.8613,
+      "step": 1813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011137282557997888,
+      "loss": 0.3611,
+      "step": 1814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011128768203454782,
+      "loss": 0.3273,
+      "step": 1815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011120253019978417,
+      "loss": 0.8676,
+      "step": 1816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011111737013822088,
+      "loss": 0.8238,
+      "step": 1817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001110322019123969,
+      "loss": 0.8229,
+      "step": 1818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011094702558485721,
+      "loss": 0.8531,
+      "step": 1819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011086184121815277,
+      "loss": 0.8672,
+      "step": 1820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011077664887484036,
+      "loss": 0.8727,
+      "step": 1821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011069144861748264,
+      "loss": 0.8124,
+      "step": 1822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011060624050864821,
+      "loss": 0.8264,
+      "step": 1823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001105210246109112,
+      "loss": 0.8593,
+      "step": 1824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011043580098685167,
+      "loss": 0.8382,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011035056969905526,
+      "loss": 0.8377,
+      "step": 1826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011026533081011323,
+      "loss": 0.8906,
+      "step": 1827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011018008438262248,
+      "loss": 0.8209,
+      "step": 1828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011009483047918539,
+      "loss": 0.835,
+      "step": 1829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011000956916240985,
+      "loss": 0.8347,
+      "step": 1830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010992430049490919,
+      "loss": 0.8378,
+      "step": 1831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010983902453930215,
+      "loss": 0.8651,
+      "step": 1832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010975374135821279,
+      "loss": 0.3568,
+      "step": 1833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010966845101427056,
+      "loss": 0.8558,
+      "step": 1834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010958315357011006,
+      "loss": 0.8949,
+      "step": 1835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010949784908837117,
+      "loss": 0.8407,
+      "step": 1836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010941253763169894,
+      "loss": 0.8225,
+      "step": 1837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010932721926274349,
+      "loss": 0.8531,
+      "step": 1838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010924189404416008,
+      "loss": 0.8463,
+      "step": 1839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010915656203860898,
+      "loss": 0.792,
+      "step": 1840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010907122330875541,
+      "loss": 0.8442,
+      "step": 1841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010898587791726955,
+      "loss": 0.8299,
+      "step": 1842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001089005259268265,
+      "loss": 0.8178,
+      "step": 1843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010881516740010618,
+      "loss": 0.8305,
+      "step": 1844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001087298023997933,
+      "loss": 0.8592,
+      "step": 1845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010864443098857733,
+      "loss": 0.8424,
+      "step": 1846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010855905322915248,
+      "loss": 0.8433,
+      "step": 1847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010847366918421759,
+      "loss": 0.8162,
+      "step": 1848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010838827891647612,
+      "loss": 0.8665,
+      "step": 1849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010830288248863609,
+      "loss": 0.8323,
+      "step": 1850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010821747996341005,
+      "loss": 0.8389,
+      "step": 1851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010813207140351507,
+      "loss": 0.8637,
+      "step": 1852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010804665687167262,
+      "loss": 0.8218,
+      "step": 1853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010796123643060847,
+      "loss": 0.8337,
+      "step": 1854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001078758101430529,
+      "loss": 0.8661,
+      "step": 1855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010779037807174033,
+      "loss": 0.8624,
+      "step": 1856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010770494027940956,
+      "loss": 0.86,
+      "step": 1857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010761949682880345,
+      "loss": 0.8101,
+      "step": 1858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010753404778266916,
+      "loss": 0.8186,
+      "step": 1859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001074485932037578,
+      "loss": 0.7877,
+      "step": 1860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010736313315482471,
+      "loss": 0.8476,
+      "step": 1861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010727766769862913,
+      "loss": 0.8153,
+      "step": 1862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010719219689793434,
+      "loss": 0.8141,
+      "step": 1863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010710672081550745,
+      "loss": 0.8385,
+      "step": 1864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010702123951411955,
+      "loss": 0.3106,
+      "step": 1865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010693575305654558,
+      "loss": 0.8181,
+      "step": 1866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010685026150556412,
+      "loss": 0.7989,
+      "step": 1867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001067647649239576,
+      "loss": 0.8512,
+      "step": 1868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010667926337451217,
+      "loss": 0.8309,
+      "step": 1869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010659375692001756,
+      "loss": 0.8016,
+      "step": 1870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001065082456232671,
+      "loss": 0.8734,
+      "step": 1871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010642272954705774,
+      "loss": 0.8498,
+      "step": 1872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010633720875418989,
+      "loss": 0.8681,
+      "step": 1873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010625168330746745,
+      "loss": 0.8725,
+      "step": 1874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010616615326969767,
+      "loss": 0.8475,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010608061870369128,
+      "loss": 0.8113,
+      "step": 1876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010599507967226228,
+      "loss": 0.2969,
+      "step": 1877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010590953623822794,
+      "loss": 0.8203,
+      "step": 1878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010582398846440875,
+      "loss": 0.7916,
+      "step": 1879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010573843641362842,
+      "loss": 0.789,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001056528801487138,
+      "loss": 0.8637,
+      "step": 1881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010556731973249485,
+      "loss": 0.8159,
+      "step": 1882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010548175522780448,
+      "loss": 0.3373,
+      "step": 1883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010539618669747877,
+      "loss": 0.8471,
+      "step": 1884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010531061420435657,
+      "loss": 0.8234,
+      "step": 1885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010522503781127979,
+      "loss": 0.3367,
+      "step": 1886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010513945758109319,
+      "loss": 0.8709,
+      "step": 1887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010505387357664421,
+      "loss": 0.8402,
+      "step": 1888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010496828586078318,
+      "loss": 0.8292,
+      "step": 1889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010488269449636316,
+      "loss": 0.8159,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010479709954623988,
+      "loss": 0.7889,
+      "step": 1891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010471150107327163,
+      "loss": 0.8324,
+      "step": 1892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010462589914031938,
+      "loss": 0.8164,
+      "step": 1893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010454029381024657,
+      "loss": 0.8842,
+      "step": 1894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010445468514591925,
+      "loss": 0.8183,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010436907321020572,
+      "loss": 0.8273,
+      "step": 1896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010428345806597686,
+      "loss": 0.837,
+      "step": 1897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010419783977610584,
+      "loss": 0.8629,
+      "step": 1898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010411221840346817,
+      "loss": 0.8455,
+      "step": 1899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010402659401094152,
+      "loss": 0.7763,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010394096666140593,
+      "loss": 0.8378,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010385533641774354,
+      "loss": 0.8051,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010376970334283858,
+      "loss": 0.8533,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010368406749957737,
+      "loss": 0.8515,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010359842895084833,
+      "loss": 0.8688,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010351278775954182,
+      "loss": 0.8236,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010342714398855014,
+      "loss": 0.8566,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010334149770076747,
+      "loss": 0.8602,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010325584895908986,
+      "loss": 0.8125,
+      "step": 1909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010317019782641518,
+      "loss": 0.8249,
+      "step": 1910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000103084544365643,
+      "loss": 0.8032,
+      "step": 1911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010299888863967464,
+      "loss": 0.8401,
+      "step": 1912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010291323071141312,
+      "loss": 0.7979,
+      "step": 1913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010282757064376297,
+      "loss": 0.8343,
+      "step": 1914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010274190849963037,
+      "loss": 0.8187,
+      "step": 1915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010265624434192307,
+      "loss": 0.7979,
+      "step": 1916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010257057823355018,
+      "loss": 0.8583,
+      "step": 1917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001024849102374223,
+      "loss": 0.7929,
+      "step": 1918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010239924041645144,
+      "loss": 0.8334,
+      "step": 1919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010231356883355092,
+      "loss": 0.8581,
+      "step": 1920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001022278955516354,
+      "loss": 0.7786,
+      "step": 1921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010214222063362067,
+      "loss": 0.8048,
+      "step": 1922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010205654414242383,
+      "loss": 0.7609,
+      "step": 1923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010197086614096317,
+      "loss": 0.8593,
+      "step": 1924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010188518669215791,
+      "loss": 0.8137,
+      "step": 1925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010179950585892855,
+      "loss": 0.8164,
+      "step": 1926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010171382370419646,
+      "loss": 0.8927,
+      "step": 1927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010162814029088404,
+      "loss": 0.8186,
+      "step": 1928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010154245568191459,
+      "loss": 0.8124,
+      "step": 1929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010145676994021227,
+      "loss": 0.8531,
+      "step": 1930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010137108312870217,
+      "loss": 0.8726,
+      "step": 1931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010128539531031,
+      "loss": 0.8127,
+      "step": 1932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010119970654796242,
+      "loss": 0.7954,
+      "step": 1933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010111401690458654,
+      "loss": 0.3237,
+      "step": 1934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010102832644311037,
+      "loss": 0.826,
+      "step": 1935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001009426352264623,
+      "loss": 0.8213,
+      "step": 1936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010085694331757143,
+      "loss": 0.8195,
+      "step": 1937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010077125077936727,
+      "loss": 0.8368,
+      "step": 1938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010068555767477983,
+      "loss": 0.7647,
+      "step": 1939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010059986406673956,
+      "loss": 0.8251,
+      "step": 1940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010051417001817726,
+      "loss": 0.8108,
+      "step": 1941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010042847559202404,
+      "loss": 0.8402,
+      "step": 1942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010034278085121127,
+      "loss": 0.854,
+      "step": 1943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001002570858586706,
+      "loss": 0.8352,
+      "step": 1944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010017139067733385,
+      "loss": 0.8874,
+      "step": 1945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010008569537013299,
+      "loss": 0.8671,
+      "step": 1946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001,
+      "loss": 0.8196,
+      "step": 1947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991430462986703e-05,
+      "loss": 0.8027,
+      "step": 1948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982860932266617e-05,
+      "loss": 0.8345,
+      "step": 1949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97429141413294e-05,
+      "loss": 0.8158,
+      "step": 1950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965721914878874e-05,
+      "loss": 0.8633,
+      "step": 1951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9571524407976e-05,
+      "loss": 0.8395,
+      "step": 1952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948582998182279e-05,
+      "loss": 0.8492,
+      "step": 1953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940013593326046e-05,
+      "loss": 0.8529,
+      "step": 1954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931444232522018e-05,
+      "loss": 0.8894,
+      "step": 1955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922874922063277e-05,
+      "loss": 0.8222,
+      "step": 1956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914305668242859e-05,
+      "loss": 0.864,
+      "step": 1957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90573647735377e-05,
+      "loss": 0.8189,
+      "step": 1958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897167355688964e-05,
+      "loss": 0.7979,
+      "step": 1959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888598309541347e-05,
+      "loss": 0.8262,
+      "step": 1960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880029345203763e-05,
+      "loss": 0.7643,
+      "step": 1961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871460468969001e-05,
+      "loss": 0.3393,
+      "step": 1962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862891687129786e-05,
+      "loss": 0.8581,
+      "step": 1963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854323005978774e-05,
+      "loss": 0.846,
+      "step": 1964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845754431808543e-05,
+      "loss": 0.7889,
+      "step": 1965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837185970911596e-05,
+      "loss": 0.8417,
+      "step": 1966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828617629580356e-05,
+      "loss": 0.8036,
+      "step": 1967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820049414107145e-05,
+      "loss": 0.759,
+      "step": 1968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811481330784211e-05,
+      "loss": 0.8381,
+      "step": 1969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802913385903688e-05,
+      "loss": 0.8195,
+      "step": 1970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79434558575762e-05,
+      "loss": 0.8457,
+      "step": 1971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785777936637937e-05,
+      "loss": 0.3219,
+      "step": 1972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777210444836463e-05,
+      "loss": 0.8368,
+      "step": 1973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768643116644909e-05,
+      "loss": 0.8401,
+      "step": 1974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760075958354857e-05,
+      "loss": 0.844,
+      "step": 1975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751508976257769e-05,
+      "loss": 0.851,
+      "step": 1976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742942176644985e-05,
+      "loss": 0.8602,
+      "step": 1977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734375565807696e-05,
+      "loss": 0.8326,
+      "step": 1978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725809150036964e-05,
+      "loss": 0.793,
+      "step": 1979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717242935623706e-05,
+      "loss": 0.8036,
+      "step": 1980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70867692885869e-05,
+      "loss": 0.8057,
+      "step": 1981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700111136032537e-05,
+      "loss": 0.823,
+      "step": 1982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691545563435702e-05,
+      "loss": 0.8686,
+      "step": 1983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682980217358483e-05,
+      "loss": 0.8026,
+      "step": 1984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.674415104091017e-05,
+      "loss": 0.8055,
+      "step": 1985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665850229923258e-05,
+      "loss": 0.8238,
+      "step": 1986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65728560114499e-05,
+      "loss": 0.9031,
+      "step": 1987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648721224045819e-05,
+      "loss": 0.7518,
+      "step": 1988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640157104915168e-05,
+      "loss": 0.8717,
+      "step": 1989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631593250042265e-05,
+      "loss": 0.8431,
+      "step": 1990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623029665716145e-05,
+      "loss": 0.8699,
+      "step": 1991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614466358225649e-05,
+      "loss": 0.8508,
+      "step": 1992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605903333859406e-05,
+      "loss": 0.8752,
+      "step": 1993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597340598905852e-05,
+      "loss": 0.8653,
+      "step": 1994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588778159653188e-05,
+      "loss": 0.8432,
+      "step": 1995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580216022389417e-05,
+      "loss": 0.3455,
+      "step": 1996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571654193402317e-05,
+      "loss": 0.8437,
+      "step": 1997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56309267897943e-05,
+      "loss": 0.8306,
+      "step": 1998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554531485408078e-05,
+      "loss": 0.8494,
+      "step": 1999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545970618975341e-05,
+      "loss": 0.8576,
+      "step": 2000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537410085968062e-05,
+      "loss": 0.7826,
+      "step": 2001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52884989267284e-05,
+      "loss": 0.8622,
+      "step": 2002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520290045376014e-05,
+      "loss": 0.3401,
+      "step": 2003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511730550363685e-05,
+      "loss": 0.8245,
+      "step": 2004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503171413921684e-05,
+      "loss": 0.8551,
+      "step": 2005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494612642335582e-05,
+      "loss": 0.8435,
+      "step": 2006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486054241890685e-05,
+      "loss": 0.8322,
+      "step": 2007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477496218872019e-05,
+      "loss": 0.8434,
+      "step": 2008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468938579564343e-05,
+      "loss": 0.8227,
+      "step": 2009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460381330252127e-05,
+      "loss": 0.8258,
+      "step": 2010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451824477219556e-05,
+      "loss": 0.8521,
+      "step": 2011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44326802675052e-05,
+      "loss": 0.8262,
+      "step": 2012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434711985128621e-05,
+      "loss": 0.7912,
+      "step": 2013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426156358637162e-05,
+      "loss": 0.8621,
+      "step": 2014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417601153559128e-05,
+      "loss": 0.8125,
+      "step": 2015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409046376177208e-05,
+      "loss": 0.8987,
+      "step": 2016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400492032773773e-05,
+      "loss": 0.8061,
+      "step": 2017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391938129630872e-05,
+      "loss": 0.8486,
+      "step": 2018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383384673030236e-05,
+      "loss": 0.8197,
+      "step": 2019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374831669253259e-05,
+      "loss": 0.7994,
+      "step": 2020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366279124581012e-05,
+      "loss": 0.7859,
+      "step": 2021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357727045294228e-05,
+      "loss": 0.8414,
+      "step": 2022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349175437673292e-05,
+      "loss": 0.3433,
+      "step": 2023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340624307998246e-05,
+      "loss": 0.8327,
+      "step": 2024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332073662548784e-05,
+      "loss": 0.7948,
+      "step": 2025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323523507604241e-05,
+      "loss": 0.8462,
+      "step": 2026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314973849443593e-05,
+      "loss": 0.8386,
+      "step": 2027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306424694345446e-05,
+      "loss": 0.8155,
+      "step": 2028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297876048588046e-05,
+      "loss": 0.8358,
+      "step": 2029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289327918449257e-05,
+      "loss": 0.8602,
+      "step": 2030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280780310206569e-05,
+      "loss": 0.8296,
+      "step": 2031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272233230137088e-05,
+      "loss": 0.8456,
+      "step": 2032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26368668451753e-05,
+      "loss": 0.8181,
+      "step": 2033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255140679624219e-05,
+      "loss": 0.8331,
+      "step": 2034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246595221733088e-05,
+      "loss": 0.7586,
+      "step": 2035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238050317119657e-05,
+      "loss": 0.7949,
+      "step": 2036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229505972059048e-05,
+      "loss": 0.8682,
+      "step": 2037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220962192825968e-05,
+      "loss": 0.8912,
+      "step": 2038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212418985694711e-05,
+      "loss": 0.8121,
+      "step": 2039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203876356939156e-05,
+      "loss": 0.8418,
+      "step": 2040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195334312832742e-05,
+      "loss": 0.8133,
+      "step": 2041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186792859648492e-05,
+      "loss": 0.8408,
+      "step": 2042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178252003658995e-05,
+      "loss": 0.8652,
+      "step": 2043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169711751136396e-05,
+      "loss": 0.8127,
+      "step": 2044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161172108352392e-05,
+      "loss": 0.8425,
+      "step": 2045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152633081578243e-05,
+      "loss": 0.8169,
+      "step": 2046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144094677084754e-05,
+      "loss": 0.8716,
+      "step": 2047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.135556901142269e-05,
+      "loss": 0.8278,
+      "step": 2048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127019760020672e-05,
+      "loss": 0.839,
+      "step": 2049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118483259989386e-05,
+      "loss": 0.8264,
+      "step": 2050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109947407317352e-05,
+      "loss": 0.8516,
+      "step": 2051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101412208273046e-05,
+      "loss": 0.7923,
+      "step": 2052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092877669124464e-05,
+      "loss": 0.8224,
+      "step": 2053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084343796139105e-05,
+      "loss": 0.3367,
+      "step": 2054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075810595583994e-05,
+      "loss": 0.3154,
+      "step": 2055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067278073725652e-05,
+      "loss": 0.8249,
+      "step": 2056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.058746236830107e-05,
+      "loss": 0.7914,
+      "step": 2057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050215091162884e-05,
+      "loss": 0.8885,
+      "step": 2058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041684642988995e-05,
+      "loss": 0.8305,
+      "step": 2059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033154898572943e-05,
+      "loss": 0.3101,
+      "step": 2060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024625864178722e-05,
+      "loss": 0.8342,
+      "step": 2061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01609754606979e-05,
+      "loss": 0.8187,
+      "step": 2062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007569950509085e-05,
+      "loss": 0.8412,
+      "step": 2063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999043083759017e-05,
+      "loss": 0.7957,
+      "step": 2064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990516952081464e-05,
+      "loss": 0.8166,
+      "step": 2065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981991561737753e-05,
+      "loss": 0.8474,
+      "step": 2066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973466918988675e-05,
+      "loss": 0.8157,
+      "step": 2067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964943030094475e-05,
+      "loss": 0.8433,
+      "step": 2068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956419901314837e-05,
+      "loss": 0.8473,
+      "step": 2069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947897538908882e-05,
+      "loss": 0.8082,
+      "step": 2070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939375949135182e-05,
+      "loss": 0.7807,
+      "step": 2071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930855138251737e-05,
+      "loss": 0.8256,
+      "step": 2072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922335112515966e-05,
+      "loss": 0.8333,
+      "step": 2073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913815878184725e-05,
+      "loss": 0.8694,
+      "step": 2074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90529744151428e-05,
+      "loss": 0.8115,
+      "step": 2075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896779808760312e-05,
+      "loss": 0.821,
+      "step": 2076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888262986177913e-05,
+      "loss": 0.8244,
+      "step": 2077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879746980021586e-05,
+      "loss": 0.8129,
+      "step": 2078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871231796545221e-05,
+      "loss": 0.3232,
+      "step": 2079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862717442002115e-05,
+      "loss": 0.8423,
+      "step": 2080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85420392264495e-05,
+      "loss": 0.805,
+      "step": 2081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845691244725802e-05,
+      "loss": 0.8775,
+      "step": 2082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83717941449612e-05,
+      "loss": 0.7565,
+      "step": 2083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828668438206738e-05,
+      "loss": 0.8922,
+      "step": 2084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820158322107856e-05,
+      "loss": 0.7917,
+      "step": 2085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811649072449055e-05,
+      "loss": 0.8164,
+      "step": 2086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803140695479262e-05,
+      "loss": 0.8588,
+      "step": 2087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-05,
+      "loss": 0.3419,
+      "step": 2088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786126584599236e-05,
+      "loss": 0.8018,
+      "step": 2089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777620863183657e-05,
+      "loss": 0.8633,
+      "step": 2090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769116039446375e-05,
+      "loss": 0.8702,
+      "step": 2091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760612119633078e-05,
+      "loss": 0.8014,
+      "step": 2092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752109109988789e-05,
+      "loss": 0.3235,
+      "step": 2093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743607016757858e-05,
+      "loss": 0.3272,
+      "step": 2094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735105846183971e-05,
+      "loss": 0.8195,
+      "step": 2095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726605604510123e-05,
+      "loss": 0.8574,
+      "step": 2096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718106297978636e-05,
+      "loss": 0.7803,
+      "step": 2097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709607932831152e-05,
+      "loss": 0.8058,
+      "step": 2098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701110515308604e-05,
+      "loss": 0.8392,
+      "step": 2099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692614051651242e-05,
+      "loss": 0.8228,
+      "step": 2100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684118548098615e-05,
+      "loss": 0.3346,
+      "step": 2101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675624010889561e-05,
+      "loss": 0.8497,
+      "step": 2102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667130446262214e-05,
+      "loss": 0.829,
+      "step": 2103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658637860453986e-05,
+      "loss": 0.85,
+      "step": 2104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650146259701582e-05,
+      "loss": 0.8398,
+      "step": 2105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641655650240971e-05,
+      "loss": 0.8368,
+      "step": 2106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633166038307401e-05,
+      "loss": 0.7654,
+      "step": 2107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624677430135392e-05,
+      "loss": 0.8622,
+      "step": 2108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616189831958716e-05,
+      "loss": 0.8429,
+      "step": 2109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60770325001041e-05,
+      "loss": 0.8096,
+      "step": 2110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599217690522766e-05,
+      "loss": 0.8225,
+      "step": 2111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590733159727316e-05,
+      "loss": 0.8283,
+      "step": 2112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58224966385485e-05,
+      "loss": 0.8262,
+      "step": 2113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573767209135387e-05,
+      "loss": 0.3306,
+      "step": 2114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565285801798184e-05,
+      "loss": 0.8483,
+      "step": 2115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556805448071735e-05,
+      "loss": 0.855,
+      "step": 2116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548326154183753e-05,
+      "loss": 0.8764,
+      "step": 2117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539847926361173e-05,
+      "loss": 0.857,
+      "step": 2118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.531370770830154e-05,
+      "loss": 0.7708,
+      "step": 2119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522894693816063e-05,
+      "loss": 0.8587,
+      "step": 2120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514419701543468e-05,
+      "loss": 0.8249,
+      "step": 2121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505945800236153e-05,
+      "loss": 0.8345,
+      "step": 2122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497472996117093e-05,
+      "loss": 0.849,
+      "step": 2123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489001295408459e-05,
+      "loss": 0.8477,
+      "step": 2124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480530704331613e-05,
+      "loss": 0.812,
+      "step": 2125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472061229107104e-05,
+      "loss": 0.8584,
+      "step": 2126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463592875954653e-05,
+      "loss": 0.8182,
+      "step": 2127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455125651093169e-05,
+      "loss": 0.8652,
+      "step": 2128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446659560740717e-05,
+      "loss": 0.8453,
+      "step": 2129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438194611114547e-05,
+      "loss": 0.8506,
+      "step": 2130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429730808431055e-05,
+      "loss": 0.7725,
+      "step": 2131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421268158905807e-05,
+      "loss": 0.8362,
+      "step": 2132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412806668753515e-05,
+      "loss": 0.3297,
+      "step": 2133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404346344188045e-05,
+      "loss": 0.8352,
+      "step": 2134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395887191422397e-05,
+      "loss": 0.8509,
+      "step": 2135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.387429216668722e-05,
+      "loss": 0.8457,
+      "step": 2136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378972426138304e-05,
+      "loss": 0.8526,
+      "step": 2137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370516826041549e-05,
+      "loss": 0.854,
+      "step": 2138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362062422587993e-05,
+      "loss": 0.8768,
+      "step": 2139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353609221986297e-05,
+      "loss": 0.8545,
+      "step": 2140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345157230444239e-05,
+      "loss": 0.8484,
+      "step": 2141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336706454168701e-05,
+      "loss": 0.8167,
+      "step": 2142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328256899365679e-05,
+      "loss": 0.8156,
+      "step": 2143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319808572240274e-05,
+      "loss": 0.7937,
+      "step": 2144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311361478996683e-05,
+      "loss": 0.7927,
+      "step": 2145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302915625838188e-05,
+      "loss": 0.3082,
+      "step": 2146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294471018967174e-05,
+      "loss": 0.8093,
+      "step": 2147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286027664585106e-05,
+      "loss": 0.8069,
+      "step": 2148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277585568892525e-05,
+      "loss": 0.825,
+      "step": 2149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269144738089052e-05,
+      "loss": 0.8039,
+      "step": 2150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260705178373381e-05,
+      "loss": 0.8069,
+      "step": 2151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252266895943266e-05,
+      "loss": 0.8115,
+      "step": 2152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243829896995528e-05,
+      "loss": 0.8204,
+      "step": 2153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235394187726046e-05,
+      "loss": 0.8108,
+      "step": 2154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226959774329747e-05,
+      "loss": 0.8146,
+      "step": 2155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218526663000609e-05,
+      "loss": 0.8444,
+      "step": 2156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210094859931656e-05,
+      "loss": 0.8258,
+      "step": 2157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201664371314948e-05,
+      "loss": 0.8206,
+      "step": 2158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193235203341585e-05,
+      "loss": 0.8265,
+      "step": 2159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184807362201687e-05,
+      "loss": 0.8088,
+      "step": 2160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176380854084408e-05,
+      "loss": 0.7909,
+      "step": 2161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.167955685177928e-05,
+      "loss": 0.8187,
+      "step": 2162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159531861669427e-05,
+      "loss": 0.853,
+      "step": 2163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151109389745108e-05,
+      "loss": 0.8247,
+      "step": 2164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142688275590181e-05,
+      "loss": 0.8233,
+      "step": 2165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134268525388862e-05,
+      "loss": 0.8245,
+      "step": 2166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125850145324354e-05,
+      "loss": 0.8731,
+      "step": 2167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117433141578866e-05,
+      "loss": 0.8307,
+      "step": 2168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10901752033359e-05,
+      "loss": 0.7952,
+      "step": 2169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100603287768703e-05,
+      "loss": 0.8224,
+      "step": 2170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092190450063366e-05,
+      "loss": 0.8102,
+      "step": 2171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083779013395707e-05,
+      "loss": 0.8547,
+      "step": 2172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075368983942834e-05,
+      "loss": 0.8126,
+      "step": 2173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066960367880823e-05,
+      "loss": 0.7798,
+      "step": 2174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058553171384699e-05,
+      "loss": 0.7863,
+      "step": 2175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050147400628459e-05,
+      "loss": 0.8366,
+      "step": 2176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041743061785046e-05,
+      "loss": 0.8483,
+      "step": 2177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033340161026351e-05,
+      "loss": 0.7766,
+      "step": 2178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.024938704523215e-05,
+      "loss": 0.8468,
+      "step": 2179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016538698445406e-05,
+      "loss": 0.8143,
+      "step": 2180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008140148961641e-05,
+      "loss": 0.8456,
+      "step": 2181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999743062239557e-05,
+      "loss": 0.8196,
+      "step": 2182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991347444445722e-05,
+      "loss": 0.8207,
+      "step": 2183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.982953301745626e-05,
+      "loss": 0.8189,
+      "step": 2184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974560640303672e-05,
+      "loss": 0.8128,
+      "step": 2185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966169466283176e-05,
+      "loss": 0.8423,
+      "step": 2186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957779785846366e-05,
+      "loss": 0.8436,
+      "step": 2187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949391605154368e-05,
+      "loss": 0.7939,
+      "step": 2188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941004930367204e-05,
+      "loss": 0.8218,
+      "step": 2189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932619767643801e-05,
+      "loss": 0.8393,
+      "step": 2190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924236123141964e-05,
+      "loss": 0.8351,
+      "step": 2191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915854003018393e-05,
+      "loss": 0.3571,
+      "step": 2192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907473413428661e-05,
+      "loss": 0.7835,
+      "step": 2193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89909436052722e-05,
+      "loss": 0.8026,
+      "step": 2194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890716850467394e-05,
+      "loss": 0.8293,
+      "step": 2195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882340889401376e-05,
+      "loss": 0.8223,
+      "step": 2196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873966483480211e-05,
+      "loss": 0.8059,
+      "step": 2197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865593638853815e-05,
+      "loss": 0.8828,
+      "step": 2198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857222361670952e-05,
+      "loss": 0.7677,
+      "step": 2199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848852658079231e-05,
+      "loss": 0.8198,
+      "step": 2200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840484534225114e-05,
+      "loss": 0.8199,
+      "step": 2201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832117996253898e-05,
+      "loss": 0.8321,
+      "step": 2202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823753050309713e-05,
+      "loss": 0.7941,
+      "step": 2203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815389702535528e-05,
+      "loss": 0.8209,
+      "step": 2204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807027959073125e-05,
+      "loss": 0.8452,
+      "step": 2205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798667826063122e-05,
+      "loss": 0.856,
+      "step": 2206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790309309644942e-05,
+      "loss": 0.7874,
+      "step": 2207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781952415956831e-05,
+      "loss": 0.8276,
+      "step": 2208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773597151135841e-05,
+      "loss": 0.3369,
+      "step": 2209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765243521317822e-05,
+      "loss": 0.7928,
+      "step": 2210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756891532637428e-05,
+      "loss": 0.8151,
+      "step": 2211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748541191228113e-05,
+      "loss": 0.8694,
+      "step": 2212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740192503222106e-05,
+      "loss": 0.8649,
+      "step": 2213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731845474750438e-05,
+      "loss": 0.8585,
+      "step": 2214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72350011194291e-05,
+      "loss": 0.7794,
+      "step": 2215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715156420928104e-05,
+      "loss": 0.7753,
+      "step": 2216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706814407833383e-05,
+      "loss": 0.856,
+      "step": 2217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698474078784859e-05,
+      "loss": 0.8082,
+      "step": 2218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690135439907422e-05,
+      "loss": 0.8448,
+      "step": 2219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681798497324716e-05,
+      "loss": 0.8112,
+      "step": 2220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673463257159146e-05,
+      "loss": 0.8731,
+      "step": 2221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.665129725531847e-05,
+      "loss": 0.8606,
+      "step": 2222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656797908562723e-05,
+      "loss": 0.8938,
+      "step": 2223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648467812370406e-05,
+      "loss": 0.8245,
+      "step": 2224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640139443072266e-05,
+      "loss": 0.7956,
+      "step": 2225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631812806784406e-05,
+      "loss": 0.8555,
+      "step": 2226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623487909621659e-05,
+      "loss": 0.8142,
+      "step": 2227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615164757697572e-05,
+      "loss": 0.8427,
+      "step": 2228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606843357124426e-05,
+      "loss": 0.862,
+      "step": 2229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598523714013194e-05,
+      "loss": 0.8179,
+      "step": 2230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590205834473576e-05,
+      "loss": 0.8266,
+      "step": 2231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581889724613976e-05,
+      "loss": 0.8165,
+      "step": 2232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.573575390541485e-05,
+      "loss": 0.8442,
+      "step": 2233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565262838361904e-05,
+      "loss": 0.819,
+      "step": 2234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556952074179721e-05,
+      "loss": 0.8217,
+      "step": 2235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548643104098108e-05,
+      "loss": 0.809,
+      "step": 2236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54033593421892e-05,
+      "loss": 0.8103,
+      "step": 2237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532030570642699e-05,
+      "loss": 0.7863,
+      "step": 2238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523727019468645e-05,
+      "loss": 0.8162,
+      "step": 2239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515425286794636e-05,
+      "loss": 0.3292,
+      "step": 2240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507125378717212e-05,
+      "loss": 0.8148,
+      "step": 2241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498827301331583e-05,
+      "loss": 0.8585,
+      "step": 2242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490531060731598e-05,
+      "loss": 0.815,
+      "step": 2243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48223666300977e-05,
+      "loss": 0.7842,
+      "step": 2244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473944114257252e-05,
+      "loss": 0.8079,
+      "step": 2245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465653420563845e-05,
+      "loss": 0.8191,
+      "step": 2246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457364588017976e-05,
+      "loss": 0.8656,
+      "step": 2247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449077622706719e-05,
+      "loss": 0.8334,
+      "step": 2248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44079253071577e-05,
+      "loss": 0.7899,
+      "step": 2249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43250931812945e-05,
+      "loss": 0.8477,
+      "step": 2250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4242279910307e-05,
+      "loss": 0.8747,
+      "step": 2251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415948555501074e-05,
+      "loss": 0.8014,
+      "step": 2252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407671017620746e-05,
+      "loss": 0.8702,
+      "step": 2253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399395383468485e-05,
+      "loss": 0.8389,
+      "step": 2254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391121659121663e-05,
+      "loss": 0.8265,
+      "step": 2255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382849850656258e-05,
+      "loss": 0.8022,
+      "step": 2256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.374579964146838e-05,
+      "loss": 0.7948,
+      "step": 2257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366312005666551e-05,
+      "loss": 0.8112,
+      "step": 2258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358045981287141e-05,
+      "loss": 0.8285,
+      "step": 2259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349781897078926e-05,
+      "loss": 0.7831,
+      "step": 2260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341519759110797e-05,
+      "loss": 0.7961,
+      "step": 2261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333259573450222e-05,
+      "loss": 0.8645,
+      "step": 2262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325001346163231e-05,
+      "loss": 0.8306,
+      "step": 2263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316745083314419e-05,
+      "loss": 0.828,
+      "step": 2264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308490790966926e-05,
+      "loss": 0.8004,
+      "step": 2265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300238475182464e-05,
+      "loss": 0.8588,
+      "step": 2266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291988142021282e-05,
+      "loss": 0.8106,
+      "step": 2267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.283739797542174e-05,
+      "loss": 0.8164,
+      "step": 2268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275493447802476e-05,
+      "loss": 0.8322,
+      "step": 2269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267249098858061e-05,
+      "loss": 0.7982,
+      "step": 2270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259006756763327e-05,
+      "loss": 0.829,
+      "step": 2271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2507664275712e-05,
+      "loss": 0.8386,
+      "step": 2272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242528117333127e-05,
+      "loss": 0.8135,
+      "step": 2273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234291832099078e-05,
+      "loss": 0.3364,
+      "step": 2274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226057577917533e-05,
+      "loss": 0.8524,
+      "step": 2275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217825360835473e-05,
+      "loss": 0.8424,
+      "step": 2276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209595186898396e-05,
+      "loss": 0.8733,
+      "step": 2277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201367062150293e-05,
+      "loss": 0.8151,
+      "step": 2278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193140992633643e-05,
+      "loss": 0.8233,
+      "step": 2279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.184916984389434e-05,
+      "loss": 0.8163,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176695043457118e-05,
+      "loss": 0.8414,
+      "step": 2281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168475175874647e-05,
+      "loss": 0.8171,
+      "step": 2282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160257387678439e-05,
+      "loss": 0.8205,
+      "step": 2283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152041684903391e-05,
+      "loss": 0.8432,
+      "step": 2284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14382807358287e-05,
+      "loss": 0.8102,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135616559748699e-05,
+      "loss": 0.8381,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127407149431167e-05,
+      "loss": 0.8498,
+      "step": 2287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119199848659023e-05,
+      "loss": 0.7779,
+      "step": 2288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110994663459451e-05,
+      "loss": 0.7971,
+      "step": 2289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102791599858096e-05,
+      "loss": 0.7896,
+      "step": 2290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094590663879039e-05,
+      "loss": 0.8395,
+      "step": 2291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0863918615448e-05,
+      "loss": 0.8605,
+      "step": 2292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078195198876335e-05,
+      "loss": 0.8443,
+      "step": 2293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070000681893017e-05,
+      "loss": 0.8451,
+      "step": 2294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061808316612658e-05,
+      "loss": 0.7952,
+      "step": 2295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053618109051487e-05,
+      "loss": 0.8235,
+      "step": 2296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045430065224138e-05,
+      "loss": 0.8427,
+      "step": 2297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037244191143661e-05,
+      "loss": 0.8546,
+      "step": 2298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029060492821522e-05,
+      "loss": 0.7789,
+      "step": 2299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020878976267578e-05,
+      "loss": 0.8729,
+      "step": 2300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012699647490086e-05,
+      "loss": 0.8265,
+      "step": 2301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004522512495702e-05,
+      "loss": 0.3319,
+      "step": 2302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996347577289464e-05,
+      "loss": 0.8437,
+      "step": 2303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988174847874798e-05,
+      "loss": 0.7948,
+      "step": 2304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980004330253512e-05,
+      "loss": 0.8199,
+      "step": 2305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971836030425781e-05,
+      "loss": 0.8183,
+      "step": 2306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963669954390161e-05,
+      "loss": 0.8554,
+      "step": 2307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955506108143575e-05,
+      "loss": 0.3438,
+      "step": 2308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947344497681298e-05,
+      "loss": 0.7836,
+      "step": 2309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939185128996976e-05,
+      "loss": 0.3323,
+      "step": 2310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931028008082602e-05,
+      "loss": 0.832,
+      "step": 2311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922873140928516e-05,
+      "loss": 0.8324,
+      "step": 2312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914720533523411e-05,
+      "loss": 0.331,
+      "step": 2313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906570191854308e-05,
+      "loss": 0.806,
+      "step": 2314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898422121906579e-05,
+      "loss": 0.7812,
+      "step": 2315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890276329663914e-05,
+      "loss": 0.7933,
+      "step": 2316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88213282110834e-05,
+      "loss": 0.3195,
+      "step": 2317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873991602220207e-05,
+      "loss": 0.8053,
+      "step": 2318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865852678978172e-05,
+      "loss": 0.8394,
+      "step": 2319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857716057359221e-05,
+      "loss": 0.7405,
+      "step": 2320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849581743338642e-05,
+      "loss": 0.8722,
+      "step": 2321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841449742890032e-05,
+      "loss": 0.8111,
+      "step": 2322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833320061985277e-05,
+      "loss": 0.8303,
+      "step": 2323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825192706594575e-05,
+      "loss": 0.8484,
+      "step": 2324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817067682686413e-05,
+      "loss": 0.7866,
+      "step": 2325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808944996227557e-05,
+      "loss": 0.8038,
+      "step": 2326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800824653183065e-05,
+      "loss": 0.8168,
+      "step": 2327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792706659516275e-05,
+      "loss": 0.8475,
+      "step": 2328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784591021188792e-05,
+      "loss": 0.8599,
+      "step": 2329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776477744160499e-05,
+      "loss": 0.839,
+      "step": 2330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768366834389535e-05,
+      "loss": 0.8613,
+      "step": 2331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760258297832311e-05,
+      "loss": 0.8152,
+      "step": 2332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752152140443495e-05,
+      "loss": 0.818,
+      "step": 2333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744048368175999e-05,
+      "loss": 0.8614,
+      "step": 2334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735946986980988e-05,
+      "loss": 0.8257,
+      "step": 2335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.727848002807877e-05,
+      "loss": 0.7968,
+      "step": 2336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719751421604309e-05,
+      "loss": 0.8315,
+      "step": 2337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.711657249316175e-05,
+      "loss": 0.8274,
+      "step": 2338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703565491887583e-05,
+      "loss": 0.8029,
+      "step": 2339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695476155260883e-05,
+      "loss": 0.3267,
+      "step": 2340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687389245376629e-05,
+      "loss": 0.826,
+      "step": 2341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679304768173613e-05,
+      "loss": 0.7953,
+      "step": 2342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671222729588828e-05,
+      "loss": 0.798,
+      "step": 2343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663143135557477e-05,
+      "loss": 0.8613,
+      "step": 2344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655065992012971e-05,
+      "loss": 0.8467,
+      "step": 2345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646991304886922e-05,
+      "loss": 0.2898,
+      "step": 2346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638919080109133e-05,
+      "loss": 0.8636,
+      "step": 2347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630849323607604e-05,
+      "loss": 0.7864,
+      "step": 2348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622782041308518e-05,
+      "loss": 0.7835,
+      "step": 2349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614717239136246e-05,
+      "loss": 0.8265,
+      "step": 2350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606654923013334e-05,
+      "loss": 0.8395,
+      "step": 2351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5985950988605e-05,
+      "loss": 0.8388,
+      "step": 2352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59053777259664e-05,
+      "loss": 0.7931,
+      "step": 2353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582482950138814e-05,
+      "loss": 0.324,
+      "step": 2354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574430637402234e-05,
+      "loss": 0.8152,
+      "step": 2355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566380840300274e-05,
+      "loss": 0.8437,
+      "step": 2356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558333564744465e-05,
+      "loss": 0.8485,
+      "step": 2357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550288816644483e-05,
+      "loss": 0.8639,
+      "step": 2358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542246601908148e-05,
+      "loss": 0.8063,
+      "step": 2359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534206926441416e-05,
+      "loss": 0.804,
+      "step": 2360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526169796148388e-05,
+      "loss": 0.8456,
+      "step": 2361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518135216931285e-05,
+      "loss": 0.8411,
+      "step": 2362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.51010319469046e-05,
+      "loss": 0.841,
+      "step": 2363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.50207373532439e-05,
+      "loss": 0.838,
+      "step": 2364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494046844729662e-05,
+      "loss": 0.8601,
+      "step": 2365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486022528800989e-05,
+      "loss": 0.8228,
+      "step": 2366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478000793431178e-05,
+      "loss": 0.8365,
+      "step": 2367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469981644511152e-05,
+      "loss": 0.8074,
+      "step": 2368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461965087929938e-05,
+      "loss": 0.8284,
+      "step": 2369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-05,
+      "loss": 0.7981,
+      "step": 2370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445939775330483e-05,
+      "loss": 0.7399,
+      "step": 2371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.437931031080754e-05,
+      "loss": 0.8022,
+      "step": 2372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429924902706832e-05,
+      "loss": 0.8377,
+      "step": 2373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421921396088177e-05,
+      "loss": 0.8314,
+      "step": 2374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413920517102321e-05,
+      "loss": 0.786,
+      "step": 2375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405922271624874e-05,
+      "loss": 0.8066,
+      "step": 2376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397926665529496e-05,
+      "loss": 0.835,
+      "step": 2377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389933704687928e-05,
+      "loss": 0.8006,
+      "step": 2378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381943394969953e-05,
+      "loss": 0.8275,
+      "step": 2379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373955742243416e-05,
+      "loss": 0.8212,
+      "step": 2380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365970752374209e-05,
+      "loss": 0.8319,
+      "step": 2381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357988431226261e-05,
+      "loss": 0.8398,
+      "step": 2382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350008784661555e-05,
+      "loss": 0.8336,
+      "step": 2383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342031818540095e-05,
+      "loss": 0.8034,
+      "step": 2384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334057538719927e-05,
+      "loss": 0.8181,
+      "step": 2385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326085951057125e-05,
+      "loss": 0.8464,
+      "step": 2386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318117061405776e-05,
+      "loss": 0.784,
+      "step": 2387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310150875617994e-05,
+      "loss": 0.837,
+      "step": 2388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302187399543911e-05,
+      "loss": 0.8117,
+      "step": 2389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294226639031653e-05,
+      "loss": 0.7926,
+      "step": 2390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286268599927367e-05,
+      "loss": 0.8445,
+      "step": 2391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278313288075197e-05,
+      "loss": 0.8572,
+      "step": 2392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270360709317281e-05,
+      "loss": 0.7955,
+      "step": 2393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262410869493757e-05,
+      "loss": 0.815,
+      "step": 2394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254463774442742e-05,
+      "loss": 0.817,
+      "step": 2395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246519430000346e-05,
+      "loss": 0.8051,
+      "step": 2396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238577842000657e-05,
+      "loss": 0.3176,
+      "step": 2397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230639016275732e-05,
+      "loss": 0.809,
+      "step": 2398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222702958655608e-05,
+      "loss": 0.8324,
+      "step": 2399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214769674968282e-05,
+      "loss": 0.7646,
+      "step": 2400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206839171039725e-05,
+      "loss": 0.8273,
+      "step": 2401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198911452693853e-05,
+      "loss": 0.8285,
+      "step": 2402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190986525752545e-05,
+      "loss": 0.8429,
+      "step": 2403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183064396035631e-05,
+      "loss": 0.8031,
+      "step": 2404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175145069360877e-05,
+      "loss": 0.3197,
+      "step": 2405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167228551544007e-05,
+      "loss": 0.8005,
+      "step": 2406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159314848398661e-05,
+      "loss": 0.8035,
+      "step": 2407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151403965736429e-05,
+      "loss": 0.806,
+      "step": 2408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143495909366826e-05,
+      "loss": 0.8235,
+      "step": 2409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135590685097287e-05,
+      "loss": 0.7834,
+      "step": 2410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127688298733169e-05,
+      "loss": 0.7875,
+      "step": 2411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11978875607775e-05,
+      "loss": 0.7853,
+      "step": 2412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111892062932209e-05,
+      "loss": 0.8008,
+      "step": 2413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103998225095648e-05,
+      "loss": 0.8432,
+      "step": 2414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0961072483650526e-05,
+      "loss": 0.352,
+      "step": 2415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088219138535325e-05,
+      "loss": 0.7578,
+      "step": 2416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080333901399251e-05,
+      "loss": 0.7731,
+      "step": 2417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0724515427475105e-05,
+      "loss": 0.786,
+      "step": 2418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0645720683686725e-05,
+      "loss": 0.8109,
+      "step": 2419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056695484049182e-05,
+      "loss": 0.793,
+      "step": 2420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048821795573364e-05,
+      "loss": 0.823,
+      "step": 2421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0409510087234215e-05,
+      "loss": 0.8221,
+      "step": 2422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03308312927942e-05,
+      "loss": 0.8596,
+      "step": 2423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0252181630192926e-05,
+      "loss": 0.8289,
+      "step": 2424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01735611571883e-05,
+      "loss": 0.7847,
+      "step": 2425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.009496993151684e-05,
+      "loss": 0.3027,
+      "step": 2426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001640801089361e-05,
+      "loss": 0.3198,
+      "step": 2427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993787545301204e-05,
+      "loss": 0.794,
+      "step": 2428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9859372315544115e-05,
+      "loss": 0.8,
+      "step": 2429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978089865614016e-05,
+      "loss": 0.8087,
+      "step": 2430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9702454532428885e-05,
+      "loss": 0.8006,
+      "step": 2431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962404000201721e-05,
+      "loss": 0.8051,
+      "step": 2432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954565512249044e-05,
+      "loss": 0.8319,
+      "step": 2433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946729995141207e-05,
+      "loss": 0.7922,
+      "step": 2434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938897454632375e-05,
+      "loss": 0.8289,
+      "step": 2435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931067896474529e-05,
+      "loss": 0.7929,
+      "step": 2436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9232413264174636e-05,
+      "loss": 0.7976,
+      "step": 2437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915417750208772e-05,
+      "loss": 0.8537,
+      "step": 2438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9075971735938554e-05,
+      "loss": 0.8259,
+      "step": 2439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899779602315902e-05,
+      "loss": 0.8193,
+      "step": 2440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891965042115907e-05,
+      "loss": 0.8219,
+      "step": 2441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884153498732642e-05,
+      "loss": 0.797,
+      "step": 2442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876344977902671e-05,
+      "loss": 0.8222,
+      "step": 2443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8685394853603384e-05,
+      "loss": 0.8281,
+      "step": 2444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860737026837757e-05,
+      "loss": 0.8291,
+      "step": 2445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852937608064819e-05,
+      "loss": 0.8854,
+      "step": 2446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845141234769181e-05,
+      "loss": 0.8071,
+      "step": 2447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837347912676272e-05,
+      "loss": 0.8245,
+      "step": 2448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829557647509256e-05,
+      "loss": 0.7971,
+      "step": 2449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821770444989081e-05,
+      "loss": 0.7827,
+      "step": 2450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8139863108344304e-05,
+      "loss": 0.8467,
+      "step": 2451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8062052507617314e-05,
+      "loss": 0.8138,
+      "step": 2452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7984272704851694e-05,
+      "loss": 0.8546,
+      "step": 2453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790652375716652e-05,
+      "loss": 0.8354,
+      "step": 2454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782880572165823e-05,
+      "loss": 0.3036,
+      "step": 2455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775111865540075e-05,
+      "loss": 0.8395,
+      "step": 2456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767346261544491e-05,
+      "loss": 0.8204,
+      "step": 2457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7595837658819105e-05,
+      "loss": 0.7948,
+      "step": 2458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751824384252873e-05,
+      "loss": 0.8392,
+      "step": 2459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744068122355627e-05,
+      "loss": 0.8261,
+      "step": 2460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7363149858861465e-05,
+      "loss": 0.789,
+      "step": 2461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7285649805380956e-05,
+      "loss": 0.8325,
+      "step": 2462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720818112002842e-05,
+      "loss": 0.8093,
+      "step": 2463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713074385969457e-05,
+      "loss": 0.8023,
+      "step": 2464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705333808124702e-05,
+      "loss": 0.7886,
+      "step": 2465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697596384153009e-05,
+      "loss": 0.8036,
+      "step": 2466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689862119736522e-05,
+      "loss": 0.8254,
+      "step": 2467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682131020555046e-05,
+      "loss": 0.7884,
+      "step": 2468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6744030922860624e-05,
+      "loss": 0.8351,
+      "step": 2469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6666783406047386e-05,
+      "loss": 0.817,
+      "step": 2470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658956771183892e-05,
+      "loss": 0.77,
+      "step": 2471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6512383896940066e-05,
+      "loss": 0.8653,
+      "step": 2472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64352320180324e-05,
+      "loss": 0.7919,
+      "step": 2473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63581121317738e-05,
+      "loss": 0.7971,
+      "step": 2474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6281024294798864e-05,
+      "loss": 0.834,
+      "step": 2475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6203968563718544e-05,
+      "loss": 0.846,
+      "step": 2476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612694499512021e-05,
+      "loss": 0.8148,
+      "step": 2477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60499536455677e-05,
+      "loss": 0.8397,
+      "step": 2478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597299457160112e-05,
+      "loss": 0.8056,
+      "step": 2479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589606782973683e-05,
+      "loss": 0.89,
+      "step": 2480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5819173476467636e-05,
+      "loss": 0.806,
+      "step": 2481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574231156826226e-05,
+      "loss": 0.8371,
+      "step": 2482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5665482161565885e-05,
+      "loss": 0.8304,
+      "step": 2483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558868531279967e-05,
+      "loss": 0.3033,
+      "step": 2484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551192107836085e-05,
+      "loss": 0.8151,
+      "step": 2485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543518951462287e-05,
+      "loss": 0.8235,
+      "step": 2486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535849067793499e-05,
+      "loss": 0.8169,
+      "step": 2487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52818246246225e-05,
+      "loss": 0.8027,
+      "step": 2488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52051914109867e-05,
+      "loss": 0.8542,
+      "step": 2489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5128591093304726e-05,
+      "loss": 0.813,
+      "step": 2490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505202372782941e-05,
+      "loss": 0.7829,
+      "step": 2491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497548937078962e-05,
+      "loss": 0.8233,
+      "step": 2492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48989880783898e-05,
+      "loss": 0.8127,
+      "step": 2493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482251990681029e-05,
+      "loss": 0.7803,
+      "step": 2494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474608491220692e-05,
+      "loss": 0.8067,
+      "step": 2495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466968315071124e-05,
+      "loss": 0.3021,
+      "step": 2496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459331467843045e-05,
+      "loss": 0.8704,
+      "step": 2497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451697955144726e-05,
+      "loss": 0.8241,
+      "step": 2498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4440677825819765e-05,
+      "loss": 0.8302,
+      "step": 2499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4364409557581754e-05,
+      "loss": 0.7924,
+      "step": 2500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4288174802742294e-05,
+      "loss": 0.8086,
+      "step": 2501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4211973617285873e-05,
+      "loss": 0.7997,
+      "step": 2502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413580605717239e-05,
+      "loss": 0.8835,
+      "step": 2503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405967217833697e-05,
+      "loss": 0.8231,
+      "step": 2504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398357203669e-05,
+      "loss": 0.8152,
+      "step": 2505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39075056881172e-05,
+      "loss": 0.8237,
+      "step": 2506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383147318847935e-05,
+      "loss": 0.7799,
+      "step": 2507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375547459361243e-05,
+      "loss": 0.856,
+      "step": 2508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36795099593275e-05,
+      "loss": 0.7829,
+      "step": 2509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360357934141067e-05,
+      "loss": 0.8959,
+      "step": 2510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3527682795623146e-05,
+      "loss": 0.8212,
+      "step": 2511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345182037770105e-05,
+      "loss": 0.7886,
+      "step": 2512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337599214335538e-05,
+      "loss": 0.8551,
+      "step": 2513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3300198148272185e-05,
+      "loss": 0.3519,
+      "step": 2514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3224438448112266e-05,
+      "loss": 0.8294,
+      "step": 2515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314871309851124e-05,
+      "loss": 0.3281,
+      "step": 2516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3073022155079513e-05,
+      "loss": 0.7818,
+      "step": 2517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299736567340219e-05,
+      "loss": 0.8028,
+      "step": 2518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292174370903919e-05,
+      "loss": 0.8533,
+      "step": 2519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284615631752494e-05,
+      "loss": 0.8432,
+      "step": 2520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2770603554368536e-05,
+      "loss": 0.8428,
+      "step": 2521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269508547505369e-05,
+      "loss": 0.8173,
+      "step": 2522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261960213503858e-05,
+      "loss": 0.7933,
+      "step": 2523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2544153589755826e-05,
+      "loss": 0.8221,
+      "step": 2524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246873989461264e-05,
+      "loss": 0.8301,
+      "step": 2525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239336110499053e-05,
+      "loss": 0.7876,
+      "step": 2526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2318017276245365e-05,
+      "loss": 0.8498,
+      "step": 2527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2242708463707444e-05,
+      "loss": 0.8197,
+      "step": 2528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2167434722681264e-05,
+      "loss": 0.8345,
+      "step": 2529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209219610844553e-05,
+      "loss": 0.7781,
+      "step": 2530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201699267625328e-05,
+      "loss": 0.3562,
+      "step": 2531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1941824481331626e-05,
+      "loss": 0.7996,
+      "step": 2532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1866691578881796e-05,
+      "loss": 0.7748,
+      "step": 2533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179159402407915e-05,
+      "loss": 0.8327,
+      "step": 2534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171653187207299e-05,
+      "loss": 0.8271,
+      "step": 2535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164150517798678e-05,
+      "loss": 0.8023,
+      "step": 2536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1566513996917844e-05,
+      "loss": 0.8543,
+      "step": 2537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149155838393737e-05,
+      "loss": 0.7917,
+      "step": 2538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1416638394090585e-05,
+      "loss": 0.8146,
+      "step": 2539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134175408239642e-05,
+      "loss": 0.8251,
+      "step": 2540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126690550384765e-05,
+      "loss": 0.8293,
+      "step": 2541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1192092713410836e-05,
+      "loss": 0.8357,
+      "step": 2542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111731576602615e-05,
+      "loss": 0.8,
+      "step": 2543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104257471660763e-05,
+      "loss": 0.8344,
+      "step": 2544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0967869620042794e-05,
+      "loss": 0.8207,
+      "step": 2545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0893200531192775e-05,
+      "loss": 0.7669,
+      "step": 2546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081856750489237e-05,
+      "loss": 0.8061,
+      "step": 2547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0743970595949775e-05,
+      "loss": 0.7512,
+      "step": 2548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.06694098591467e-05,
+      "loss": 0.8122,
+      "step": 2549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059488534923831e-05,
+      "loss": 0.884,
+      "step": 2550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052039712095311e-05,
+      "loss": 0.8381,
+      "step": 2551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044594522899309e-05,
+      "loss": 0.8137,
+      "step": 2552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037152972803342e-05,
+      "loss": 0.807,
+      "step": 2553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029715067272256e-05,
+      "loss": 0.8369,
+      "step": 2554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022280811768231e-05,
+      "loss": 0.8294,
+      "step": 2555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014850211750757e-05,
+      "loss": 0.7952,
+      "step": 2556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00742327267664e-05,
+      "loss": 0.3262,
+      "step": 2557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.8386,
+      "step": 2558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99258039917227e-05,
+      "loss": 0.795,
+      "step": 2559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985164475642169e-05,
+      "loss": 0.7853,
+      "step": 2560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977752234855738e-05,
+      "loss": 0.828,
+      "step": 2561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9703436822562986e-05,
+      "loss": 0.8122,
+      "step": 2562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962938823284465e-05,
+      "loss": 0.8272,
+      "step": 2563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9555376633781484e-05,
+      "loss": 0.7983,
+      "step": 2564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9481402079725325e-05,
+      "loss": 0.8336,
+      "step": 2565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940746462500088e-05,
+      "loss": 0.8137,
+      "step": 2566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9333564323905534e-05,
+      "loss": 0.8057,
+      "step": 2567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925970123070944e-05,
+      "loss": 0.8139,
+      "step": 2568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918587539965548e-05,
+      "loss": 0.7726,
+      "step": 2569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911208688495907e-05,
+      "loss": 0.8203,
+      "step": 2570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903833574080825e-05,
+      "loss": 0.796,
+      "step": 2571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8964622021363684e-05,
+      "loss": 0.798,
+      "step": 2572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8890945780758467e-05,
+      "loss": 0.7807,
+      "step": 2573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881730707309821e-05,
+      "loss": 0.8269,
+      "step": 2574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874370595246094e-05,
+      "loss": 0.8065,
+      "step": 2575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867014247289707e-05,
+      "loss": 0.7995,
+      "step": 2576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859661668842949e-05,
+      "loss": 0.7911,
+      "step": 2577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852312865305324e-05,
+      "loss": 0.7773,
+      "step": 2578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84496784207357e-05,
+      "loss": 0.7671,
+      "step": 2579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837626604541657e-05,
+      "loss": 0.8043,
+      "step": 2580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830289158100764e-05,
+      "loss": 0.8487,
+      "step": 2581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8229555081392894e-05,
+      "loss": 0.7916,
+      "step": 2582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815625660042845e-05,
+      "loss": 0.8059,
+      "step": 2583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808299619194251e-05,
+      "loss": 0.7992,
+      "step": 2584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8009773909735234e-05,
+      "loss": 0.8145,
+      "step": 2585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7936589807578946e-05,
+      "loss": 0.8017,
+      "step": 2586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.786344393921781e-05,
+      "loss": 0.8286,
+      "step": 2587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779033635836789e-05,
+      "loss": 0.8225,
+      "step": 2588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7717267118717266e-05,
+      "loss": 0.7927,
+      "step": 2589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7644236273925744e-05,
+      "loss": 0.3281,
+      "step": 2590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757124387762496e-05,
+      "loss": 0.8425,
+      "step": 2591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7498289983418346e-05,
+      "loss": 0.8131,
+      "step": 2592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742537464488096e-05,
+      "loss": 0.8138,
+      "step": 2593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735249791555973e-05,
+      "loss": 0.7857,
+      "step": 2594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727965984897309e-05,
+      "loss": 0.8286,
+      "step": 2595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720686049861106e-05,
+      "loss": 0.8493,
+      "step": 2596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713409991793536e-05,
+      "loss": 0.8144,
+      "step": 2597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706137816037913e-05,
+      "loss": 0.8338,
+      "step": 2598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698869527934703e-05,
+      "loss": 0.7834,
+      "step": 2599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691605132821517e-05,
+      "loss": 0.8018,
+      "step": 2600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684344636033103e-05,
+      "loss": 0.7591,
+      "step": 2601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67708804290136e-05,
+      "loss": 0.8353,
+      "step": 2602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669835358755305e-05,
+      "loss": 0.7812,
+      "step": 2603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662586588921086e-05,
+      "loss": 0.7996,
+      "step": 2604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6553417387219886e-05,
+      "loss": 0.781,
+      "step": 2605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64810081347841e-05,
+      "loss": 0.8464,
+      "step": 2606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64086381850786e-05,
+      "loss": 0.8265,
+      "step": 2607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633630759124982e-05,
+      "loss": 0.8206,
+      "step": 2608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6264016406415014e-05,
+      "loss": 0.8225,
+      "step": 2609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6191764683662744e-05,
+      "loss": 0.8278,
+      "step": 2610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6119552476052466e-05,
+      "loss": 0.822,
+      "step": 2611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60473798366146e-05,
+      "loss": 0.8515,
+      "step": 2612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597524681835063e-05,
+      "loss": 0.7965,
+      "step": 2613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590315347423282e-05,
+      "loss": 0.8059,
+      "step": 2614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583109985720431e-05,
+      "loss": 0.8259,
+      "step": 2615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575908602017922e-05,
+      "loss": 0.8133,
+      "step": 2616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568711201604221e-05,
+      "loss": 0.8444,
+      "step": 2617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561517789764882e-05,
+      "loss": 0.7803,
+      "step": 2618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554328371782537e-05,
+      "loss": 0.8247,
+      "step": 2619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547142952936874e-05,
+      "loss": 0.8142,
+      "step": 2620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539961538504641e-05,
+      "loss": 0.7918,
+      "step": 2621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532784133759663e-05,
+      "loss": 0.8154,
+      "step": 2622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5256107439728035e-05,
+      "loss": 0.7745,
+      "step": 2623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518441374411982e-05,
+      "loss": 0.8173,
+      "step": 2624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51127603034217e-05,
+      "loss": 0.8498,
+      "step": 2625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504114717025373e-05,
+      "loss": 0.7895,
+      "step": 2626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496957439720652e-05,
+      "loss": 0.7849,
+      "step": 2627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48980420368409e-05,
+      "loss": 0.7807,
+      "step": 2628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482655014168805e-05,
+      "loss": 0.7925,
+      "step": 2629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475509876424954e-05,
+      "loss": 0.8225,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4683687956997046e-05,
+      "loss": 0.818,
+      "step": 2631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461231777237248e-05,
+      "loss": 0.3224,
+      "step": 2632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454098826278806e-05,
+      "loss": 0.8046,
+      "step": 2633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4469699480625884e-05,
+      "loss": 0.7598,
+      "step": 2634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439845147823837e-05,
+      "loss": 0.8053,
+      "step": 2635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432724430794786e-05,
+      "loss": 0.8457,
+      "step": 2636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4256078022046734e-05,
+      "loss": 0.8008,
+      "step": 2637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418495267279741e-05,
+      "loss": 0.844,
+      "step": 2638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4113868312432186e-05,
+      "loss": 0.3365,
+      "step": 2639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4042824993153195e-05,
+      "loss": 0.7708,
+      "step": 2640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397182276713265e-05,
+      "loss": 0.7665,
+      "step": 2641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3900861686512315e-05,
+      "loss": 0.7929,
+      "step": 2642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382994180340386e-05,
+      "loss": 0.8144,
+      "step": 2643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375906316988879e-05,
+      "loss": 0.8407,
+      "step": 2644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3688225838018173e-05,
+      "loss": 0.7457,
+      "step": 2645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361742985981278e-05,
+      "loss": 0.8137,
+      "step": 2646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3546675287263126e-05,
+      "loss": 0.8188,
+      "step": 2647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3475962172329176e-05,
+      "loss": 0.8245,
+      "step": 2648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340529056694047e-05,
+      "loss": 0.8235,
+      "step": 2649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333466052299622e-05,
+      "loss": 0.7835,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3264072092364824e-05,
+      "loss": 0.7644,
+      "step": 2651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3193525326884435e-05,
+      "loss": 0.8381,
+      "step": 2652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3123020278362404e-05,
+      "loss": 0.7933,
+      "step": 2653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305255699857549e-05,
+      "loss": 0.8194,
+      "step": 2654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2982135539269854e-05,
+      "loss": 0.8457,
+      "step": 2655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291175595216086e-05,
+      "loss": 0.7924,
+      "step": 2656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284141828893311e-05,
+      "loss": 0.8346,
+      "step": 2657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2771122601240576e-05,
+      "loss": 0.7911,
+      "step": 2658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2700868940706127e-05,
+      "loss": 0.7968,
+      "step": 2659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263065735892205e-05,
+      "loss": 0.817,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256048790744956e-05,
+      "loss": 0.8179,
+      "step": 2661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249036063781896e-05,
+      "loss": 0.7718,
+      "step": 2662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242027560152966e-05,
+      "loss": 0.8191,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235023285004995e-05,
+      "loss": 0.7838,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.22802324348171e-05,
+      "loss": 0.8063,
+      "step": 2665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221027440723738e-05,
+      "loss": 0.8327,
+      "step": 2666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214035881868572e-05,
+      "loss": 0.8444,
+      "step": 2667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207048572050612e-05,
+      "loss": 0.8067,
+      "step": 2668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2000655164011236e-05,
+      "loss": 0.812,
+      "step": 2669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193086720048249e-05,
+      "loss": 0.8461,
+      "step": 2670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186112188117012e-05,
+      "loss": 0.8048,
+      "step": 2671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179141925729294e-05,
+      "loss": 0.8076,
+      "step": 2672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172175938003843e-05,
+      "loss": 0.8533,
+      "step": 2673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1652142300562745e-05,
+      "loss": 0.8237,
+      "step": 2674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158256806999059e-05,
+      "loss": 0.8248,
+      "step": 2675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151303673941506e-05,
+      "loss": 0.7732,
+      "step": 2676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144354835989795e-05,
+      "loss": 0.818,
+      "step": 2677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137410298246943e-05,
+      "loss": 0.8451,
+      "step": 2678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130470065812802e-05,
+      "loss": 0.7815,
+      "step": 2679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123534143784078e-05,
+      "loss": 0.8143,
+      "step": 2680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116602537254298e-05,
+      "loss": 0.8713,
+      "step": 2681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1096752513138216e-05,
+      "loss": 0.7795,
+      "step": 2682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102752291049847e-05,
+      "loss": 0.787,
+      "step": 2683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0958336615463744e-05,
+      "loss": 0.807,
+      "step": 2684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0889193678842476e-05,
+      "loss": 0.3513,
+      "step": 2685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082009415141109e-05,
+      "loss": 0.2991,
+      "step": 2686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075103808391417e-05,
+      "loss": 0.8211,
+      "step": 2687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0682025527064486e-05,
+      "loss": 0.3181,
+      "step": 2688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061305653154272e-05,
+      "loss": 0.8317,
+      "step": 2689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054413114799759e-05,
+      "loss": 0.8013,
+      "step": 2690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047524942704589e-05,
+      "loss": 0.8344,
+      "step": 2691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040641141927224e-05,
+      "loss": 0.8373,
+      "step": 2692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0337617175229166e-05,
+      "loss": 0.8165,
+      "step": 2693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026886674543713e-05,
+      "loss": 0.8012,
+      "step": 2694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020016018038429e-05,
+      "loss": 0.7932,
+      "step": 2695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013149753052677e-05,
+      "loss": 0.8212,
+      "step": 2696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0062878846288296e-05,
+      "loss": 0.815,
+      "step": 2697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999430417806034e-05,
+      "loss": 0.7843,
+      "step": 2698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99257735762021e-05,
+      "loss": 0.8263,
+      "step": 2699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985728709104041e-05,
+      "loss": 0.8154,
+      "step": 2700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978884477286956e-05,
+      "loss": 0.7685,
+      "step": 2701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972044667195163e-05,
+      "loss": 0.8443,
+      "step": 2702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9652092838516076e-05,
+      "loss": 0.8373,
+      "step": 2703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958378332275985e-05,
+      "loss": 0.3325,
+      "step": 2704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9515518174847476e-05,
+      "loss": 0.8127,
+      "step": 2705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944729744491078e-05,
+      "loss": 0.7948,
+      "step": 2706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9379121183048964e-05,
+      "loss": 0.8129,
+      "step": 2707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9310989439328715e-05,
+      "loss": 0.7916,
+      "step": 2708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9242902263783795e-05,
+      "loss": 0.7655,
+      "step": 2709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917485970641548e-05,
+      "loss": 0.8222,
+      "step": 2710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910686181719212e-05,
+      "loss": 0.815,
+      "step": 2711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9038908646049286e-05,
+      "loss": 0.8507,
+      "step": 2712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897100024288979e-05,
+      "loss": 0.771,
+      "step": 2713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890313665758348e-05,
+      "loss": 0.8588,
+      "step": 2714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88353179399673e-05,
+      "loss": 0.8188,
+      "step": 2715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876754413984533e-05,
+      "loss": 0.7995,
+      "step": 2716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869981530698854e-05,
+      "loss": 0.7673,
+      "step": 2717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863213149113498e-05,
+      "loss": 0.82,
+      "step": 2718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856449274198958e-05,
+      "loss": 0.8248,
+      "step": 2719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849689910922415e-05,
+      "loss": 0.8355,
+      "step": 2720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8429350642477504e-05,
+      "loss": 0.75,
+      "step": 2721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836184739135513e-05,
+      "loss": 0.7853,
+      "step": 2722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8294389405429366e-05,
+      "loss": 0.3165,
+      "step": 2723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822697673423937e-05,
+      "loss": 0.8123,
+      "step": 2724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815960942729095e-05,
+      "loss": 0.8306,
+      "step": 2725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809228753405662e-05,
+      "loss": 0.831,
+      "step": 2726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802501110397553e-05,
+      "loss": 0.8113,
+      "step": 2727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795778018645344e-05,
+      "loss": 0.7684,
+      "step": 2728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789059483086276e-05,
+      "loss": 0.7549,
+      "step": 2729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782345508654235e-05,
+      "loss": 0.8247,
+      "step": 2730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775636100279759e-05,
+      "loss": 0.8112,
+      "step": 2731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76893126289004e-05,
+      "loss": 0.7655,
+      "step": 2732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7622310014089046e-05,
+      "loss": 0.7886,
+      "step": 2733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7555353207568244e-05,
+      "loss": 0.768,
+      "step": 2734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748844225850903e-05,
+      "loss": 0.822,
+      "step": 2735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7421577216048775e-05,
+      "loss": 0.8033,
+      "step": 2736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735475812929113e-05,
+      "loss": 0.8372,
+      "step": 2737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7287985047306084e-05,
+      "loss": 0.781,
+      "step": 2738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7221258019129725e-05,
+      "loss": 0.7847,
+      "step": 2739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7154577093764334e-05,
+      "loss": 0.3262,
+      "step": 2740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708794232017842e-05,
+      "loss": 0.8014,
+      "step": 2741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702135374730655e-05,
+      "loss": 0.8089,
+      "step": 2742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.695481142404934e-05,
+      "loss": 0.8031,
+      "step": 2743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6888315399273454e-05,
+      "loss": 0.8437,
+      "step": 2744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.682186572181153e-05,
+      "loss": 0.8199,
+      "step": 2745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675546244046228e-05,
+      "loss": 0.8176,
+      "step": 2746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6689105603990234e-05,
+      "loss": 0.8175,
+      "step": 2747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662279526112581e-05,
+      "loss": 0.8088,
+      "step": 2748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655653146056539e-05,
+      "loss": 0.8202,
+      "step": 2749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.649031425097106e-05,
+      "loss": 0.787,
+      "step": 2750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6424143680970756e-05,
+      "loss": 0.7831,
+      "step": 2751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6358019799158135e-05,
+      "loss": 0.8178,
+      "step": 2752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.629194265409255e-05,
+      "loss": 0.7719,
+      "step": 2753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.622591229429911e-05,
+      "loss": 0.7814,
+      "step": 2754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.61599287682685e-05,
+      "loss": 0.8139,
+      "step": 2755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6093992124457e-05,
+      "loss": 0.3427,
+      "step": 2756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602810241128654e-05,
+      "loss": 0.7891,
+      "step": 2757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596225967714449e-05,
+      "loss": 0.7682,
+      "step": 2758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58964639703838e-05,
+      "loss": 0.7982,
+      "step": 2759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.583071533932282e-05,
+      "loss": 0.8124,
+      "step": 2760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.576501383224537e-05,
+      "loss": 0.7915,
+      "step": 2761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569935949740061e-05,
+      "loss": 0.7691,
+      "step": 2762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.563375238300317e-05,
+      "loss": 0.8415,
+      "step": 2763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5568192537232904e-05,
+      "loss": 0.7956,
+      "step": 2764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5502680008234936e-05,
+      "loss": 0.8396,
+      "step": 2765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543721484411976e-05,
+      "loss": 0.8522,
+      "step": 2766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.537179709296299e-05,
+      "loss": 0.808,
+      "step": 2767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5306426802805436e-05,
+      "loss": 0.8042,
+      "step": 2768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524110402165307e-05,
+      "loss": 0.7892,
+      "step": 2769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.517582879747693e-05,
+      "loss": 0.7851,
+      "step": 2770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511060117821323e-05,
+      "loss": 0.7862,
+      "step": 2771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504542121176314e-05,
+      "loss": 0.3353,
+      "step": 2772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4980288945992825e-05,
+      "loss": 0.8156,
+      "step": 2773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4915204428733504e-05,
+      "loss": 0.8456,
+      "step": 2774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4850167707781256e-05,
+      "loss": 0.8427,
+      "step": 2775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478517883089709e-05,
+      "loss": 0.8026,
+      "step": 2776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472023784580686e-05,
+      "loss": 0.8268,
+      "step": 2777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465534480020124e-05,
+      "loss": 0.8191,
+      "step": 2778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459049974173579e-05,
+      "loss": 0.3322,
+      "step": 2779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4525702718030716e-05,
+      "loss": 0.7643,
+      "step": 2780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446095377667097e-05,
+      "loss": 0.7445,
+      "step": 2781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439625296520629e-05,
+      "loss": 0.8566,
+      "step": 2782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433160033115095e-05,
+      "loss": 0.8067,
+      "step": 2783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.426699592198389e-05,
+      "loss": 0.8472,
+      "step": 2784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.420243978514864e-05,
+      "loss": 0.8107,
+      "step": 2785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413793196805326e-05,
+      "loss": 0.335,
+      "step": 2786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.407347251807039e-05,
+      "loss": 0.8396,
+      "step": 2787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4009061482537065e-05,
+      "loss": 0.7961,
+      "step": 2788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394469890875478e-05,
+      "loss": 0.8281,
+      "step": 2789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3880384843989535e-05,
+      "loss": 0.7957,
+      "step": 2790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.381611933547159e-05,
+      "loss": 0.7794,
+      "step": 2791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.375190243039556e-05,
+      "loss": 0.7917,
+      "step": 2792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.36877341759205e-05,
+      "loss": 0.8173,
+      "step": 2793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362361461916955e-05,
+      "loss": 0.8094,
+      "step": 2794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3559543807230165e-05,
+      "loss": 0.7981,
+      "step": 2795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.349552178715407e-05,
+      "loss": 0.856,
+      "step": 2796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.343154860595709e-05,
+      "loss": 0.8359,
+      "step": 2797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336762431061915e-05,
+      "loss": 0.8285,
+      "step": 2798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330374894808438e-05,
+      "loss": 0.7657,
+      "step": 2799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3239922565260905e-05,
+      "loss": 0.8144,
+      "step": 2800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317614520902087e-05,
+      "loss": 0.8536,
+      "step": 2801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311241692620045e-05,
+      "loss": 0.8395,
+      "step": 2802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304873776359972e-05,
+      "loss": 0.7788,
+      "step": 2803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2985107767982835e-05,
+      "loss": 0.7927,
+      "step": 2804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292152698607768e-05,
+      "loss": 0.7844,
+      "step": 2805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285799546457602e-05,
+      "loss": 0.8026,
+      "step": 2806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2794513250133574e-05,
+      "loss": 0.8457,
+      "step": 2807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2731080389369703e-05,
+      "loss": 0.7766,
+      "step": 2808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.266769692886762e-05,
+      "loss": 0.7881,
+      "step": 2809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260436291517418e-05,
+      "loss": 0.8214,
+      "step": 2810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2541078394799975e-05,
+      "loss": 0.8439,
+      "step": 2811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247784341421928e-05,
+      "loss": 0.8467,
+      "step": 2812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2414658019869934e-05,
+      "loss": 0.8578,
+      "step": 2813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2351522258153345e-05,
+      "loss": 0.8233,
+      "step": 2814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2288436175434577e-05,
+      "loss": 0.7973,
+      "step": 2815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2225399818042115e-05,
+      "loss": 0.7693,
+      "step": 2816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216241323226792e-05,
+      "loss": 0.7841,
+      "step": 2817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209947646436752e-05,
+      "loss": 0.8233,
+      "step": 2818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203658956055969e-05,
+      "loss": 0.8373,
+      "step": 2819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197375256702668e-05,
+      "loss": 0.8023,
+      "step": 2820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.191096552991413e-05,
+      "loss": 0.797,
+      "step": 2821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.184822849533091e-05,
+      "loss": 0.8205,
+      "step": 2822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.178554150934919e-05,
+      "loss": 0.837,
+      "step": 2823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172290461800443e-05,
+      "loss": 0.8372,
+      "step": 2824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1660317867295275e-05,
+      "loss": 0.8087,
+      "step": 2825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1597781303183516e-05,
+      "loss": 0.8116,
+      "step": 2826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153529497159412e-05,
+      "loss": 0.8231,
+      "step": 2827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.147285891841514e-05,
+      "loss": 0.3383,
+      "step": 2828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141047318949778e-05,
+      "loss": 0.7601,
+      "step": 2829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134813783065621e-05,
+      "loss": 0.8111,
+      "step": 2830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.12858528876676e-05,
+      "loss": 0.7912,
+      "step": 2831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1223618406272183e-05,
+      "loss": 0.7677,
+      "step": 2832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1161434432173065e-05,
+      "loss": 0.7899,
+      "step": 2833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.109930101103623e-05,
+      "loss": 0.8679,
+      "step": 2834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10372181884907e-05,
+      "loss": 0.7974,
+      "step": 2835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097518601012808e-05,
+      "loss": 0.8461,
+      "step": 2836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0913204521503025e-05,
+      "loss": 0.8079,
+      "step": 2837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085127376813285e-05,
+      "loss": 0.8711,
+      "step": 2838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.078939379549758e-05,
+      "loss": 0.8205,
+      "step": 2839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.072756464904006e-05,
+      "loss": 0.801,
+      "step": 2840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066578637416573e-05,
+      "loss": 0.7607,
+      "step": 2841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0604059016242645e-05,
+      "loss": 0.8348,
+      "step": 2842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0542382620601604e-05,
+      "loss": 0.7899,
+      "step": 2843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0480757232535772e-05,
+      "loss": 0.7654,
+      "step": 2844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041918289730107e-05,
+      "loss": 0.8186,
+      "step": 2845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0357659660115768e-05,
+      "loss": 0.3029,
+      "step": 2846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0296187566160672e-05,
+      "loss": 0.8193,
+      "step": 2847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0234766660579082e-05,
+      "loss": 0.8283,
+      "step": 2848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017339698847663e-05,
+      "loss": 0.8221,
+      "step": 2849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011207859492131e-05,
+      "loss": 0.8002,
+      "step": 2850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005081152494359e-05,
+      "loss": 0.8516,
+      "step": 2851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9989595823536065e-05,
+      "loss": 0.8136,
+      "step": 2852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.992843153565369e-05,
+      "loss": 0.8107,
+      "step": 2853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9867318706213744e-05,
+      "loss": 0.8218,
+      "step": 2854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9806257380095605e-05,
+      "loss": 0.8144,
+      "step": 2855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9745247602140824e-05,
+      "loss": 0.8136,
+      "step": 2856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968428941715321e-05,
+      "loss": 0.8634,
+      "step": 2857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9623382869898587e-05,
+      "loss": 0.7895,
+      "step": 2858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.956252800510485e-05,
+      "loss": 0.808,
+      "step": 2859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9501724867462055e-05,
+      "loss": 0.7293,
+      "step": 2860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944097350162207e-05,
+      "loss": 0.8352,
+      "step": 2861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9380273952198955e-05,
+      "loss": 0.8165,
+      "step": 2862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.931962626376861e-05,
+      "loss": 0.3009,
+      "step": 2863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9259030480868798e-05,
+      "loss": 0.824,
+      "step": 2864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9198486647999334e-05,
+      "loss": 0.8039,
+      "step": 2865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9137994809621716e-05,
+      "loss": 0.8069,
+      "step": 2866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9077555010159308e-05,
+      "loss": 0.8166,
+      "step": 2867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901716729399735e-05,
+      "loss": 0.8201,
+      "step": 2868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8956831705482658e-05,
+      "loss": 0.7869,
+      "step": 2869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889654828892393e-05,
+      "loss": 0.7649,
+      "step": 2870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8836317088591468e-05,
+      "loss": 0.7559,
+      "step": 2871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8776138148717225e-05,
+      "loss": 0.8164,
+      "step": 2872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8716011513494833e-05,
+      "loss": 0.7839,
+      "step": 2873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8655937227079466e-05,
+      "loss": 0.8217,
+      "step": 2874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859591533358782e-05,
+      "loss": 0.8456,
+      "step": 2875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8535945877098235e-05,
+      "loss": 0.8183,
+      "step": 2876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8476028901650455e-05,
+      "loss": 0.8107,
+      "step": 2877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8416164451245607e-05,
+      "loss": 0.7841,
+      "step": 2878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8356352569846423e-05,
+      "loss": 0.7933,
+      "step": 2879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829659330137693e-05,
+      "loss": 0.8374,
+      "step": 2880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8236886689722475e-05,
+      "loss": 0.3232,
+      "step": 2881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8177232778729856e-05,
+      "loss": 0.8034,
+      "step": 2882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8117631612207084e-05,
+      "loss": 0.8,
+      "step": 2883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8058083233923404e-05,
+      "loss": 0.7633,
+      "step": 2884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7998587687609455e-05,
+      "loss": 0.8158,
+      "step": 2885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7939145016956845e-05,
+      "loss": 0.8211,
+      "step": 2886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7879755265618555e-05,
+      "loss": 0.8284,
+      "step": 2887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782041847720861e-05,
+      "loss": 0.8245,
+      "step": 2888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7761134695302103e-05,
+      "loss": 0.8094,
+      "step": 2889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7701903963435327e-05,
+      "loss": 0.807,
+      "step": 2890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7642726325105506e-05,
+      "loss": 0.8225,
+      "step": 2891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.758360182377089e-05,
+      "loss": 0.8122,
+      "step": 2892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7524530502850788e-05,
+      "loss": 0.823,
+      "step": 2893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74655124057253e-05,
+      "loss": 0.8342,
+      "step": 2894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740654757573562e-05,
+      "loss": 0.8029,
+      "step": 2895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73476360561837e-05,
+      "loss": 0.7714,
+      "step": 2896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7288777890332352e-05,
+      "loss": 0.8128,
+      "step": 2897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7229973121405295e-05,
+      "loss": 0.8318,
+      "step": 2898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7171221792586942e-05,
+      "loss": 0.8246,
+      "step": 2899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.711252394702247e-05,
+      "loss": 0.8288,
+      "step": 2900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7053879627817857e-05,
+      "loss": 0.8022,
+      "step": 2901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6995288878039682e-05,
+      "loss": 0.787,
+      "step": 2902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6936751740715237e-05,
+      "loss": 0.834,
+      "step": 2903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6878268258832417e-05,
+      "loss": 0.8224,
+      "step": 2904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.681983847533969e-05,
+      "loss": 0.774,
+      "step": 2905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6761462433146202e-05,
+      "loss": 0.8527,
+      "step": 2906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67031401751215e-05,
+      "loss": 0.8301,
+      "step": 2907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6644871744095666e-05,
+      "loss": 0.7921,
+      "step": 2908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6586657182859343e-05,
+      "loss": 0.794,
+      "step": 2909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6528496534163538e-05,
+      "loss": 0.7983,
+      "step": 2910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.647038984071959e-05,
+      "loss": 0.835,
+      "step": 2911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6412337145199383e-05,
+      "loss": 0.8258,
+      "step": 2912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6354338490235043e-05,
+      "loss": 0.7826,
+      "step": 2913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6296393918418982e-05,
+      "loss": 0.8099,
+      "step": 2914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.623850347230401e-05,
+      "loss": 0.8501,
+      "step": 2915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6180667194403095e-05,
+      "loss": 0.7969,
+      "step": 2916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6122885127189433e-05,
+      "loss": 0.7844,
+      "step": 2917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6065157313096457e-05,
+      "loss": 0.8407,
+      "step": 2918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600748379451773e-05,
+      "loss": 0.8158,
+      "step": 2919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5949864613806916e-05,
+      "loss": 0.7631,
+      "step": 2920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.589229981327781e-05,
+      "loss": 0.8303,
+      "step": 2921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5834789435204243e-05,
+      "loss": 0.7722,
+      "step": 2922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5777333521820148e-05,
+      "loss": 0.8225,
+      "step": 2923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5719932115319377e-05,
+      "loss": 0.7681,
+      "step": 2924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5662585257855775e-05,
+      "loss": 0.7992,
+      "step": 2925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5605292991543184e-05,
+      "loss": 0.83,
+      "step": 2926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554805535845529e-05,
+      "loss": 0.3323,
+      "step": 2927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5490872400625686e-05,
+      "loss": 0.7913,
+      "step": 2928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54337441600478e-05,
+      "loss": 0.8181,
+      "step": 2929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5376670678674864e-05,
+      "loss": 0.7801,
+      "step": 2930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5319651998419968e-05,
+      "loss": 0.8038,
+      "step": 2931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5262688161155877e-05,
+      "loss": 0.766,
+      "step": 2932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5205779208715096e-05,
+      "loss": 0.763,
+      "step": 2933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-05,
+      "loss": 0.8092,
+      "step": 2934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5092126125432136e-05,
+      "loss": 0.7818,
+      "step": 2935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5035382078053247e-05,
+      "loss": 0.8044,
+      "step": 2936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4978693082424454e-05,
+      "loss": 0.8104,
+      "step": 2937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4922059180176394e-05,
+      "loss": 0.8148,
+      "step": 2938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4865480412899254e-05,
+      "loss": 0.8378,
+      "step": 2939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480895682214286e-05,
+      "loss": 0.7951,
+      "step": 2940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4752488449416367e-05,
+      "loss": 0.8102,
+      "step": 2941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4696075336188428e-05,
+      "loss": 0.8145,
+      "step": 2942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4639717523887195e-05,
+      "loss": 0.851,
+      "step": 2943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4583415053900095e-05,
+      "loss": 0.8568,
+      "step": 2944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4527167967573973e-05,
+      "loss": 0.3277,
+      "step": 2945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4470976306214978e-05,
+      "loss": 0.7978,
+      "step": 2946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4414840111088556e-05,
+      "loss": 0.8581,
+      "step": 2947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4358759423419474e-05,
+      "loss": 0.814,
+      "step": 2948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.430273428439168e-05,
+      "loss": 0.7764,
+      "step": 2949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42467647351483e-05,
+      "loss": 0.8129,
+      "step": 2950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4190850816791742e-05,
+      "loss": 0.8391,
+      "step": 2951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4134992570383486e-05,
+      "loss": 0.8178,
+      "step": 2952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4079190036944112e-05,
+      "loss": 0.7717,
+      "step": 2953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4023443257453325e-05,
+      "loss": 0.7944,
+      "step": 2954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3967752272849854e-05,
+      "loss": 0.7558,
+      "step": 2955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391211712403153e-05,
+      "loss": 0.7904,
+      "step": 2956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3856537851855098e-05,
+      "loss": 0.8226,
+      "step": 2957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3801014497136254e-05,
+      "loss": 0.8593,
+      "step": 2958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3745547100649744e-05,
+      "loss": 0.7499,
+      "step": 2959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.369013570312911e-05,
+      "loss": 0.7932,
+      "step": 2960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3634780345266806e-05,
+      "loss": 0.7817,
+      "step": 2961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357948106771415e-05,
+      "loss": 0.7743,
+      "step": 2962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3524237911081214e-05,
+      "loss": 0.781,
+      "step": 2963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3469050915936975e-05,
+      "loss": 0.8264,
+      "step": 2964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341392012280905e-05,
+      "loss": 0.8358,
+      "step": 2965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3358845572183818e-05,
+      "loss": 0.8331,
+      "step": 2966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33038273045064e-05,
+      "loss": 0.8111,
+      "step": 2967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3248865360180538e-05,
+      "loss": 0.8395,
+      "step": 2968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.319395977956861e-05,
+      "loss": 0.8185,
+      "step": 2969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3139110602991633e-05,
+      "loss": 0.7825,
+      "step": 2970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3084317870729167e-05,
+      "loss": 0.7711,
+      "step": 2971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3029581623019315e-05,
+      "loss": 0.7898,
+      "step": 2972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2974901900058777e-05,
+      "loss": 0.7978,
+      "step": 2973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2920278742002676e-05,
+      "loss": 0.3325,
+      "step": 2974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.286571218896456e-05,
+      "loss": 0.7812,
+      "step": 2975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281120228101653e-05,
+      "loss": 0.8269,
+      "step": 2976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2756749058188987e-05,
+      "loss": 0.8011,
+      "step": 2977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270235256047072e-05,
+      "loss": 0.3198,
+      "step": 2978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2648012827808874e-05,
+      "loss": 0.842,
+      "step": 2979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2593729900108885e-05,
+      "loss": 0.8179,
+      "step": 2980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2539503817234553e-05,
+      "loss": 0.7746,
+      "step": 2981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2485334619007835e-05,
+      "loss": 0.8267,
+      "step": 2982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2431222345208935e-05,
+      "loss": 0.8125,
+      "step": 2983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237716703557633e-05,
+      "loss": 0.7718,
+      "step": 2984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2323168729806575e-05,
+      "loss": 0.3348,
+      "step": 2985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2269227467554388e-05,
+      "loss": 0.3298,
+      "step": 2986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.221534328843261e-05,
+      "loss": 0.8129,
+      "step": 2987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2161516232012114e-05,
+      "loss": 0.8007,
+      "step": 2988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2107746337821934e-05,
+      "loss": 0.781,
+      "step": 2989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2054033645349015e-05,
+      "loss": 0.3189,
+      "step": 2990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2000378194038328e-05,
+      "loss": 0.8296,
+      "step": 2991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1946780023292847e-05,
+      "loss": 0.8535,
+      "step": 2992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1893239172473424e-05,
+      "loss": 0.8377,
+      "step": 2993
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1839755680898853e-05,
+      "loss": 0.8062,
+      "step": 2994
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178632958784579e-05,
+      "loss": 0.8399,
+      "step": 2995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173296093254875e-05,
+      "loss": 0.8478,
+      "step": 2996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1679649754200016e-05,
+      "loss": 0.8633,
+      "step": 2997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1626396091949776e-05,
+      "loss": 0.8523,
+      "step": 2998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1573199984905868e-05,
+      "loss": 0.8009,
+      "step": 2999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1520061472133902e-05,
+      "loss": 0.7824,
+      "step": 3000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1466980592657225e-05,
+      "loss": 0.8286,
+      "step": 3001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1413957385456828e-05,
+      "loss": 0.8077,
+      "step": 3002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1360991889471327e-05,
+      "loss": 0.7994,
+      "step": 3003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1308084143596997e-05,
+      "loss": 0.7973,
+      "step": 3004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1255234186687667e-05,
+      "loss": 0.8327,
+      "step": 3005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1202442057554785e-05,
+      "loss": 0.8076,
+      "step": 3006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1149707794967266e-05,
+      "loss": 0.313,
+      "step": 3007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109703143765156e-05,
+      "loss": 0.801,
+      "step": 3008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104441302429162e-05,
+      "loss": 0.8047,
+      "step": 3009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0991852593528793e-05,
+      "loss": 0.786,
+      "step": 3010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0939350183961893e-05,
+      "loss": 0.8256,
+      "step": 3011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0886905834147074e-05,
+      "loss": 0.8312,
+      "step": 3012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0834519582597878e-05,
+      "loss": 0.7923,
+      "step": 3013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.078219146778523e-05,
+      "loss": 0.8073,
+      "step": 3014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0729921528137288e-05,
+      "loss": 0.8064,
+      "step": 3015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067770980203948e-05,
+      "loss": 0.8117,
+      "step": 3016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0625556327834593e-05,
+      "loss": 0.8116,
+      "step": 3017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0573461143822524e-05,
+      "loss": 0.8125,
+      "step": 3018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052142428826038e-05,
+      "loss": 0.814,
+      "step": 3019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046944579936253e-05,
+      "loss": 0.7936,
+      "step": 3020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04175257153003e-05,
+      "loss": 0.8196,
+      "step": 3021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0365664074202318e-05,
+      "loss": 0.8026,
+      "step": 3022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0313860914154172e-05,
+      "loss": 0.8062,
+      "step": 3023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0262116273198504e-05,
+      "loss": 0.7681,
+      "step": 3024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0210430189335083e-05,
+      "loss": 0.8069,
+      "step": 3025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0158802700520574e-05,
+      "loss": 0.8129,
+      "step": 3026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0107233844668617e-05,
+      "loss": 0.8236,
+      "step": 3027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0055723659649904e-05,
+      "loss": 0.8448,
+      "step": 3028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0004272183291873e-05,
+      "loss": 0.8041,
+      "step": 3029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9952879453378938e-05,
+      "loss": 0.9015,
+      "step": 3030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9901545507652418e-05,
+      "loss": 0.8001,
+      "step": 3031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9850270383810388e-05,
+      "loss": 0.8004,
+      "step": 3032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9799054119507722e-05,
+      "loss": 0.8458,
+      "step": 3033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9747896752356145e-05,
+      "loss": 0.8274,
+      "step": 3034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9696798319924058e-05,
+      "loss": 0.7759,
+      "step": 3035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9645758859736608e-05,
+      "loss": 0.7742,
+      "step": 3036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9594778409275638e-05,
+      "loss": 0.8271,
+      "step": 3037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9543857005979605e-05,
+      "loss": 0.8126,
+      "step": 3038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9492994687243714e-05,
+      "loss": 0.3004,
+      "step": 3039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9442191490419693e-05,
+      "loss": 0.8533,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9391447452815835e-05,
+      "loss": 0.7942,
+      "step": 3041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9340762611697093e-05,
+      "loss": 0.3018,
+      "step": 3042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929013700428486e-05,
+      "loss": 0.8276,
+      "step": 3043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9239570667757e-05,
+      "loss": 0.7395,
+      "step": 3044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9189063639248008e-05,
+      "loss": 0.734,
+      "step": 3045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9138615955848605e-05,
+      "loss": 0.3246,
+      "step": 3046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908822765460613e-05,
+      "loss": 0.7814,
+      "step": 3047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9037898772524176e-05,
+      "loss": 0.8672,
+      "step": 3048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.898762934656276e-05,
+      "loss": 0.7808,
+      "step": 3049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.893741941363828e-05,
+      "loss": 0.8244,
+      "step": 3050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.888726901062334e-05,
+      "loss": 0.8279,
+      "step": 3051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.883717817434688e-05,
+      "loss": 0.804,
+      "step": 3052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8787146941594158e-05,
+      "loss": 0.8316,
+      "step": 3053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.873717534910653e-05,
+      "loss": 0.7786,
+      "step": 3054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8687263433581626e-05,
+      "loss": 0.8408,
+      "step": 3055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8637411231673295e-05,
+      "loss": 0.8214,
+      "step": 3056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8587618779991466e-05,
+      "loss": 0.7677,
+      "step": 3057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8537886115102178e-05,
+      "loss": 0.7981,
+      "step": 3058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8488213273527656e-05,
+      "loss": 0.7975,
+      "step": 3059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84386002917461e-05,
+      "loss": 0.7973,
+      "step": 3060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838904720619178e-05,
+      "loss": 0.8056,
+      "step": 3061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8339554053255025e-05,
+      "loss": 0.7955,
+      "step": 3062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8290120869282034e-05,
+      "loss": 0.8661,
+      "step": 3063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.824074769057512e-05,
+      "loss": 0.8179,
+      "step": 3064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8191434553392428e-05,
+      "loss": 0.8331,
+      "step": 3065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8142181493947997e-05,
+      "loss": 0.8306,
+      "step": 3066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809298854841185e-05,
+      "loss": 0.7869,
+      "step": 3067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8043855752909767e-05,
+      "loss": 0.8494,
+      "step": 3068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7994783143523363e-05,
+      "loss": 0.8523,
+      "step": 3069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794577075629016e-05,
+      "loss": 0.7561,
+      "step": 3070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.789681862720326e-05,
+      "loss": 0.8268,
+      "step": 3071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784792679221172e-05,
+      "loss": 0.3176,
+      "step": 3072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7799095287220202e-05,
+      "loss": 0.8126,
+      "step": 3073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7750324148089048e-05,
+      "loss": 0.7721,
+      "step": 3074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7701613410634365e-05,
+      "loss": 0.8199,
+      "step": 3075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.765296311062782e-05,
+      "loss": 0.8508,
+      "step": 3076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7604373283796705e-05,
+      "loss": 0.8171,
+      "step": 3077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7555843965823992e-05,
+      "loss": 0.8026,
+      "step": 3078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7507375192348043e-05,
+      "loss": 0.7923,
+      "step": 3079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7458966998962933e-05,
+      "loss": 0.7675,
+      "step": 3080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7410619421218156e-05,
+      "loss": 0.7703,
+      "step": 3081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7362332494618695e-05,
+      "loss": 0.8367,
+      "step": 3082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731410625462504e-05,
+      "loss": 0.8291,
+      "step": 3083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7265940736653075e-05,
+      "loss": 0.8061,
+      "step": 3084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7217835976074072e-05,
+      "loss": 0.7843,
+      "step": 3085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7169792008214757e-05,
+      "loss": 0.7972,
+      "step": 3086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7121808868357182e-05,
+      "loss": 0.7898,
+      "step": 3087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707388659173863e-05,
+      "loss": 0.7714,
+      "step": 3088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026025213551854e-05,
+      "loss": 0.8047,
+      "step": 3089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697822476894477e-05,
+      "loss": 0.3176,
+      "step": 3090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.693048529302056e-05,
+      "loss": 0.8368,
+      "step": 3091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6882806820837705e-05,
+      "loss": 0.8296,
+      "step": 3092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.68351893874098e-05,
+      "loss": 0.7927,
+      "step": 3093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.678763302770563e-05,
+      "loss": 0.791,
+      "step": 3094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6740137776649223e-05,
+      "loss": 0.8485,
+      "step": 3095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.669270366911957e-05,
+      "loss": 0.7941,
+      "step": 3096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6645330739950893e-05,
+      "loss": 0.7757,
+      "step": 3097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6598019023932443e-05,
+      "loss": 0.7753,
+      "step": 3098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655076855580846e-05,
+      "loss": 0.7971,
+      "step": 3099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6503579370278344e-05,
+      "loss": 0.7779,
+      "step": 3100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645645150199635e-05,
+      "loss": 0.8546,
+      "step": 3101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640938498557175e-05,
+      "loss": 0.7999,
+      "step": 3102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6362379855568834e-05,
+      "loss": 0.8401,
+      "step": 3103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6315436146506703e-05,
+      "loss": 0.8227,
+      "step": 3104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6268553892859405e-05,
+      "loss": 0.7792,
+      "step": 3105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.622173312905585e-05,
+      "loss": 0.7832,
+      "step": 3106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6174973889479774e-05,
+      "loss": 0.765,
+      "step": 3107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6128276208469794e-05,
+      "loss": 0.8068,
+      "step": 3108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6081640120319243e-05,
+      "loss": 0.7931,
+      "step": 3109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603506565927624e-05,
+      "loss": 0.7765,
+      "step": 3110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5988552859543715e-05,
+      "loss": 0.8293,
+      "step": 3111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.594210175527925e-05,
+      "loss": 0.7834,
+      "step": 3112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5895712380595075e-05,
+      "loss": 0.8124,
+      "step": 3113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5849384769558195e-05,
+      "loss": 0.822,
+      "step": 3114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.580311895619021e-05,
+      "loss": 0.8341,
+      "step": 3115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5756914974467285e-05,
+      "loss": 0.8049,
+      "step": 3116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.57107728583203e-05,
+      "loss": 0.8837,
+      "step": 3117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5664692641634594e-05,
+      "loss": 0.8469,
+      "step": 3118
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5618674358250063e-05,
+      "loss": 0.805,
+      "step": 3119
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.557271804196121e-05,
+      "loss": 0.793,
+      "step": 3120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.552682372651688e-05,
+      "loss": 0.7862,
+      "step": 3121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5480991445620542e-05,
+      "loss": 0.7792,
+      "step": 3122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5435221232929996e-05,
+      "loss": 0.8392,
+      "step": 3123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538951312205751e-05,
+      "loss": 0.7595,
+      "step": 3124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5343867146569758e-05,
+      "loss": 0.7961,
+      "step": 3125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5298283339987762e-05,
+      "loss": 0.8014,
+      "step": 3126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5252761735786858e-05,
+      "loss": 0.7673,
+      "step": 3127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5207302367396792e-05,
+      "loss": 0.8378,
+      "step": 3128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5161905268201516e-05,
+      "loss": 0.7622,
+      "step": 3129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5116570471539293e-05,
+      "loss": 0.8355,
+      "step": 3130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5071298010702617e-05,
+      "loss": 0.7901,
+      "step": 3131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5026087918938214e-05,
+      "loss": 0.8206,
+      "step": 3132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4980940229447038e-05,
+      "loss": 0.7863,
+      "step": 3133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4935854975384178e-05,
+      "loss": 0.7806,
+      "step": 3134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489083218985885e-05,
+      "loss": 0.7826,
+      "step": 3135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4845871905934472e-05,
+      "loss": 0.8359,
+      "step": 3136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48009741566285e-05,
+      "loss": 0.7902,
+      "step": 3137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.475613897491248e-05,
+      "loss": 0.7525,
+      "step": 3138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4711366393712011e-05,
+      "loss": 0.7887,
+      "step": 3139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4666656445906713e-05,
+      "loss": 0.8148,
+      "step": 3140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4622009164330263e-05,
+      "loss": 0.8215,
+      "step": 3141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4577424581770239e-05,
+      "loss": 0.7958,
+      "step": 3142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4532902730968212e-05,
+      "loss": 0.8114,
+      "step": 3143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448844364461971e-05,
+      "loss": 0.7673,
+      "step": 3144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4444047355374124e-05,
+      "loss": 0.7599,
+      "step": 3145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4399713895834743e-05,
+      "loss": 0.8124,
+      "step": 3146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4355443298558735e-05,
+      "loss": 0.7681,
+      "step": 3147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4311235596057082e-05,
+      "loss": 0.7559,
+      "step": 3148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4267090820794549e-05,
+      "loss": 0.8409,
+      "step": 3149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4223009005189792e-05,
+      "loss": 0.839,
+      "step": 3150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4178990181615126e-05,
+      "loss": 0.8382,
+      "step": 3151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4135034382396628e-05,
+      "loss": 0.842,
+      "step": 3152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4091141639814143e-05,
+      "loss": 0.8046,
+      "step": 3153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4047311986101175e-05,
+      "loss": 0.8252,
+      "step": 3154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4003545453444877e-05,
+      "loss": 0.8436,
+      "step": 3155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3959842073986085e-05,
+      "loss": 0.8577,
+      "step": 3156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3916201879819202e-05,
+      "loss": 0.8132,
+      "step": 3157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3872624902992315e-05,
+      "loss": 0.7969,
+      "step": 3158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3829111175507015e-05,
+      "loss": 0.8248,
+      "step": 3159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.378566072931846e-05,
+      "loss": 0.7908,
+      "step": 3160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3742273596335386e-05,
+      "loss": 0.833,
+      "step": 3161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3698949808419959e-05,
+      "loss": 0.8009,
+      "step": 3162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.365568939738786e-05,
+      "loss": 0.7608,
+      "step": 3163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3612492395008237e-05,
+      "loss": 0.8178,
+      "step": 3164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3569358833003643e-05,
+      "loss": 0.8217,
+      "step": 3165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3526288743050097e-05,
+      "loss": 0.811,
+      "step": 3166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3483282156776945e-05,
+      "loss": 0.8138,
+      "step": 3167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344033910576692e-05,
+      "loss": 0.8038,
+      "step": 3168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 0.8049,
+      "step": 3169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3354643735633964e-05,
+      "loss": 0.7586,
+      "step": 3170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3311891479443095e-05,
+      "loss": 0.3147,
+      "step": 3171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3269202884379506e-05,
+      "loss": 0.8172,
+      "step": 3172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.322657798179241e-05,
+      "loss": 0.7975,
+      "step": 3173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3184016802984212e-05,
+      "loss": 0.792,
+      "step": 3174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.314151937921061e-05,
+      "loss": 0.7821,
+      "step": 3175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3099085741680406e-05,
+      "loss": 0.8112,
+      "step": 3176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305671592155555e-05,
+      "loss": 0.7816,
+      "step": 3177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3014409949951223e-05,
+      "loss": 0.8337,
+      "step": 3178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2972167857935602e-05,
+      "loss": 0.7983,
+      "step": 3179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2929989676530019e-05,
+      "loss": 0.8236,
+      "step": 3180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2887875436708852e-05,
+      "loss": 0.7914,
+      "step": 3181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2845825169399507e-05,
+      "loss": 0.803,
+      "step": 3182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2803838905482468e-05,
+      "loss": 0.7872,
+      "step": 3183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2761916675791164e-05,
+      "loss": 0.8221,
+      "step": 3184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2720058511111987e-05,
+      "loss": 0.816,
+      "step": 3185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2678264442184362e-05,
+      "loss": 0.8549,
+      "step": 3186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2636534499700548e-05,
+      "loss": 0.3103,
+      "step": 3187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2594868714305764e-05,
+      "loss": 0.844,
+      "step": 3188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255326711659811e-05,
+      "loss": 0.8506,
+      "step": 3189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2511729737128518e-05,
+      "loss": 0.8147,
+      "step": 3190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247025660640082e-05,
+      "loss": 0.8492,
+      "step": 3191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.242884775487161e-05,
+      "loss": 0.7438,
+      "step": 3192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238750321295028e-05,
+      "loss": 0.8137,
+      "step": 3193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2346223010999058e-05,
+      "loss": 0.7734,
+      "step": 3194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.230500717933285e-05,
+      "loss": 0.7516,
+      "step": 3195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2263855748219322e-05,
+      "loss": 0.7886,
+      "step": 3196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2222768747878832e-05,
+      "loss": 0.7662,
+      "step": 3197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2181746208484424e-05,
+      "loss": 0.785,
+      "step": 3198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2140788160161843e-05,
+      "loss": 0.7598,
+      "step": 3199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2099894632989417e-05,
+      "loss": 0.8126,
+      "step": 3200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2059065656998102e-05,
+      "loss": 0.3157,
+      "step": 3201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2018301262171516e-05,
+      "loss": 0.8634,
+      "step": 3202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977601478445765e-05,
+      "loss": 0.8126,
+      "step": 3203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1936966335709521e-05,
+      "loss": 0.8212,
+      "step": 3204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1896395863804066e-05,
+      "loss": 0.8102,
+      "step": 3205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1855890092523081e-05,
+      "loss": 0.7979,
+      "step": 3206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.181544905161276e-05,
+      "loss": 0.7708,
+      "step": 3207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1775072770771834e-05,
+      "loss": 0.8387,
+      "step": 3208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1734761279651397e-05,
+      "loss": 0.8143,
+      "step": 3209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1694514607854968e-05,
+      "loss": 0.8413,
+      "step": 3210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1654332784938538e-05,
+      "loss": 0.3517,
+      "step": 3211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1614215840410402e-05,
+      "loss": 0.8356,
+      "step": 3212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1574163803731219e-05,
+      "loss": 0.7704,
+      "step": 3213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1534176704314004e-05,
+      "loss": 0.815,
+      "step": 3214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1494254571524055e-05,
+      "loss": 0.7886,
+      "step": 3215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1454397434679021e-05,
+      "loss": 0.788,
+      "step": 3216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1414605323048743e-05,
+      "loss": 0.8305,
+      "step": 3217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.137487826585535e-05,
+      "loss": 0.7597,
+      "step": 3218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1335216292273221e-05,
+      "loss": 0.8639,
+      "step": 3219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1295619431428884e-05,
+      "loss": 0.752,
+      "step": 3220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1256087712401087e-05,
+      "loss": 0.8182,
+      "step": 3221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.121662116422072e-05,
+      "loss": 0.8025,
+      "step": 3222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1177219815870809e-05,
+      "loss": 0.82,
+      "step": 3223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1137883696286565e-05,
+      "loss": 0.7622,
+      "step": 3224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1098612834355204e-05,
+      "loss": 0.7779,
+      "step": 3225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1059407258916054e-05,
+      "loss": 0.8227,
+      "step": 3226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1020266998760542e-05,
+      "loss": 0.787,
+      "step": 3227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0981192082632074e-05,
+      "loss": 0.7561,
+      "step": 3228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0942182539226064e-05,
+      "loss": 0.7479,
+      "step": 3229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0903238397189997e-05,
+      "loss": 0.8295,
+      "step": 3230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0864359685123226e-05,
+      "loss": 0.7251,
+      "step": 3231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0825546431577094e-05,
+      "loss": 0.7865,
+      "step": 3232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0786798665054931e-05,
+      "loss": 0.7827,
+      "step": 3233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0748116414011888e-05,
+      "loss": 0.8199,
+      "step": 3234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0709499706855043e-05,
+      "loss": 0.7932,
+      "step": 3235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0670948571943362e-05,
+      "loss": 0.7822,
+      "step": 3236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0632463037587603e-05,
+      "loss": 0.8249,
+      "step": 3237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0594043132050424e-05,
+      "loss": 0.8577,
+      "step": 3238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055568888354621e-05,
+      "loss": 0.8346,
+      "step": 3239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0517400320241133e-05,
+      "loss": 0.753,
+      "step": 3240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.047917747025322e-05,
+      "loss": 0.8355,
+      "step": 3241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0441020361652144e-05,
+      "loss": 0.7739,
+      "step": 3242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0402929022459318e-05,
+      "loss": 0.7821,
+      "step": 3243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0364903480647914e-05,
+      "loss": 0.7859,
+      "step": 3244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0326943764142705e-05,
+      "loss": 0.3239,
+      "step": 3245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0289049900820148e-05,
+      "loss": 0.7878,
+      "step": 3246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0251221918508425e-05,
+      "loss": 0.8194,
+      "step": 3247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021345984498716e-05,
+      "loss": 0.3175,
+      "step": 3248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0175763707987751e-05,
+      "loss": 0.8277,
+      "step": 3249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0138133535193084e-05,
+      "loss": 0.7701,
+      "step": 3250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01005693542376e-05,
+      "loss": 0.8139,
+      "step": 3251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006307119270733e-05,
+      "loss": 0.8328,
+      "step": 3252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.002563907813977e-05,
+      "loss": 0.7842,
+      "step": 3253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.988273038023932e-06,
+      "loss": 0.8103,
+      "step": 3254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.950973099800354e-06,
+      "loss": 0.8111,
+      "step": 3255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.913739290860902e-06,
+      "loss": 0.8314,
+      "step": 3256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.876571638549015e-06,
+      "loss": 0.8003,
+      "step": 3257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.839470170159492e-06,
+      "loss": 0.7915,
+      "step": 3258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.802434912938496e-06,
+      "loss": 0.8315,
+      "step": 3259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.765465894083636e-06,
+      "loss": 0.8019,
+      "step": 3260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.72856314074384e-06,
+      "loss": 0.3141,
+      "step": 3261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.691726680019341e-06,
+      "loss": 0.812,
+      "step": 3262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65495653896179e-06,
+      "loss": 0.7826,
+      "step": 3263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.618252744573997e-06,
+      "loss": 0.8052,
+      "step": 3264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.581615323810134e-06,
+      "loss": 0.8027,
+      "step": 3265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.545044303575646e-06,
+      "loss": 0.3317,
+      "step": 3266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.508539710727171e-06,
+      "loss": 0.8063,
+      "step": 3267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.472101572072546e-06,
+      "loss": 0.8535,
+      "step": 3268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.435729914370894e-06,
+      "loss": 0.8303,
+      "step": 3269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.399424764332432e-06,
+      "loss": 0.837,
+      "step": 3270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.363186148618553e-06,
+      "loss": 0.8585,
+      "step": 3271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.327014093841858e-06,
+      "loss": 0.7937,
+      "step": 3272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.29090862656593e-06,
+      "loss": 0.7577,
+      "step": 3273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.254869773305585e-06,
+      "loss": 0.7749,
+      "step": 3274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.218897560526663e-06,
+      "loss": 0.8345,
+      "step": 3275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.182992014646041e-06,
+      "loss": 0.7482,
+      "step": 3276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.147153162031708e-06,
+      "loss": 0.816,
+      "step": 3277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.111381029002618e-06,
+      "loss": 0.8406,
+      "step": 3278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.075675641828718e-06,
+      "loss": 0.8364,
+      "step": 3279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.04003702673103e-06,
+      "loss": 0.7951,
+      "step": 3280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.004465209881396e-06,
+      "loss": 0.8147,
+      "step": 3281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.968960217402745e-06,
+      "loss": 0.8475,
+      "step": 3282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.93352207536884e-06,
+      "loss": 0.7882,
+      "step": 3283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.898150809804373e-06,
+      "loss": 0.7844,
+      "step": 3284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.86284644668497e-06,
+      "loss": 0.7567,
+      "step": 3285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827609011937066e-06,
+      "loss": 0.7794,
+      "step": 3286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79243853143794e-06,
+      "loss": 0.7993,
+      "step": 3287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.75733503101579e-06,
+      "loss": 0.7788,
+      "step": 3288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.722298536449558e-06,
+      "loss": 0.8032,
+      "step": 3289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.687329073468919e-06,
+      "loss": 0.807,
+      "step": 3290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.652426667754455e-06,
+      "loss": 0.8191,
+      "step": 3291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617591344937426e-06,
+      "loss": 0.8221,
+      "step": 3292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.582823130599804e-06,
+      "loss": 0.8138,
+      "step": 3293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.548122050274366e-06,
+      "loss": 0.8362,
+      "step": 3294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.513488129444491e-06,
+      "loss": 0.7888,
+      "step": 3295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.478921393544337e-06,
+      "loss": 0.3234,
+      "step": 3296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.444421867958651e-06,
+      "loss": 0.8239,
+      "step": 3297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.409989578022792e-06,
+      "loss": 0.8494,
+      "step": 3298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.375624549022854e-06,
+      "loss": 0.8062,
+      "step": 3299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34132680619546e-06,
+      "loss": 0.7885,
+      "step": 3300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.307096374727807e-06,
+      "loss": 0.7821,
+      "step": 3301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.272933279757722e-06,
+      "loss": 0.8589,
+      "step": 3302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.238837546373557e-06,
+      "loss": 0.7988,
+      "step": 3303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20480919961415e-06,
+      "loss": 0.8025,
+      "step": 3304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.170848264468934e-06,
+      "loss": 0.8137,
+      "step": 3305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.136954765877748e-06,
+      "loss": 0.7978,
+      "step": 3306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.103128728731002e-06,
+      "loss": 0.8177,
+      "step": 3307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.069370177869472e-06,
+      "loss": 0.7989,
+      "step": 3308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.035679138084428e-06,
+      "loss": 0.7564,
+      "step": 3309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.002055634117578e-06,
+      "loss": 0.8125,
+      "step": 3310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.968499690660991e-06,
+      "loss": 0.7702,
+      "step": 3311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.935011332357112e-06,
+      "loss": 0.8061,
+      "step": 3312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.901590583798824e-06,
+      "loss": 0.813,
+      "step": 3313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.868237469529294e-06,
+      "loss": 0.7841,
+      "step": 3314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83495201404204e-06,
+      "loss": 0.7646,
+      "step": 3315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.801734241780901e-06,
+      "loss": 0.8152,
+      "step": 3316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.768584177139982e-06,
+      "loss": 0.8012,
+      "step": 3317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.73550184446371e-06,
+      "loss": 0.7726,
+      "step": 3318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.702487268046765e-06,
+      "loss": 0.3047,
+      "step": 3319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.669540472134008e-06,
+      "loss": 0.8112,
+      "step": 3320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.636661480920616e-06,
+      "loss": 0.8057,
+      "step": 3321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.603850318551931e-06,
+      "loss": 0.7981,
+      "step": 3322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5711070091234234e-06,
+      "loss": 0.7723,
+      "step": 3323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.538431576680838e-06,
+      "loss": 0.8104,
+      "step": 3324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.505824045220011e-06,
+      "loss": 0.7815,
+      "step": 3325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.473284438686911e-06,
+      "loss": 0.8533,
+      "step": 3326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4408127809776685e-06,
+      "loss": 0.8099,
+      "step": 3327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.408409095938484e-06,
+      "loss": 0.7997,
+      "step": 3328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3760734073656245e-06,
+      "loss": 0.7571,
+      "step": 3329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3438057390054715e-06,
+      "loss": 0.2956,
+      "step": 3330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.311606114554415e-06,
+      "loss": 0.8084,
+      "step": 3331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.279474557658872e-06,
+      "loss": 0.8314,
+      "step": 3332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.247411091915301e-06,
+      "loss": 0.8205,
+      "step": 3333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.215415740870124e-06,
+      "loss": 0.8013,
+      "step": 3334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.183488528019799e-06,
+      "loss": 0.861,
+      "step": 3335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.151629476810661e-06,
+      "loss": 0.8111,
+      "step": 3336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119838610639052e-06,
+      "loss": 0.8038,
+      "step": 3337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.088115952851238e-06,
+      "loss": 0.8127,
+      "step": 3338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0564615267433744e-06,
+      "loss": 0.8031,
+      "step": 3339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0248753555615e-06,
+      "loss": 0.8014,
+      "step": 3340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9933574625015594e-06,
+      "loss": 0.8119,
+      "step": 3341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9619078707093166e-06,
+      "loss": 0.8356,
+      "step": 3342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.930526603280441e-06,
+      "loss": 0.8167,
+      "step": 3343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.899213683260364e-06,
+      "loss": 0.7841,
+      "step": 3344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.867969133644325e-06,
+      "loss": 0.7977,
+      "step": 3345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.836792977377427e-06,
+      "loss": 0.7836,
+      "step": 3346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.805685237354486e-06,
+      "loss": 0.8209,
+      "step": 3347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.774645936420043e-06,
+      "loss": 0.8431,
+      "step": 3348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.743675097368463e-06,
+      "loss": 0.8193,
+      "step": 3349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.712772742943784e-06,
+      "loss": 0.7789,
+      "step": 3350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.681938895839746e-06,
+      "loss": 0.7584,
+      "step": 3351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.651173578699821e-06,
+      "loss": 0.2984,
+      "step": 3352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620476814117104e-06,
+      "loss": 0.7918,
+      "step": 3353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.589848624634398e-06,
+      "loss": 0.8292,
+      "step": 3354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.559289032744098e-06,
+      "loss": 0.7709,
+      "step": 3355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.528798060888263e-06,
+      "loss": 0.7771,
+      "step": 3356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.498375731458528e-06,
+      "loss": 0.8149,
+      "step": 3357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.468022066796131e-06,
+      "loss": 0.7813,
+      "step": 3358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.437737089191875e-06,
+      "loss": 0.7889,
+      "step": 3359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.407520820886159e-06,
+      "loss": 0.7938,
+      "step": 3360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3773732840688996e-06,
+      "loss": 0.8143,
+      "step": 3361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.347294500879497e-06,
+      "loss": 0.8162,
+      "step": 3362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.317284493406938e-06,
+      "loss": 0.7867,
+      "step": 3363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.287343283689661e-06,
+      "loss": 0.7997,
+      "step": 3364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.257470893715567e-06,
+      "loss": 0.7762,
+      "step": 3365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.227667345422061e-06,
+      "loss": 0.3108,
+      "step": 3366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.197932660695937e-06,
+      "loss": 0.895,
+      "step": 3367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.168266861373473e-06,
+      "loss": 0.7862,
+      "step": 3368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1386699692403405e-06,
+      "loss": 0.8118,
+      "step": 3369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.109142006031565e-06,
+      "loss": 0.8228,
+      "step": 3370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0796829934316455e-06,
+      "loss": 0.8004,
+      "step": 3371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.050292953074355e-06,
+      "loss": 0.8099,
+      "step": 3372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.020971906542849e-06,
+      "loss": 0.7783,
+      "step": 3373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.991719875369639e-06,
+      "loss": 0.8336,
+      "step": 3374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.962536881036507e-06,
+      "loss": 0.7989,
+      "step": 3375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933422944974587e-06,
+      "loss": 0.3298,
+      "step": 3376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.904378088564255e-06,
+      "loss": 0.3153,
+      "step": 3377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.875402333135183e-06,
+      "loss": 0.804,
+      "step": 3378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.846495699966292e-06,
+      "loss": 0.7587,
+      "step": 3379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.817658210285748e-06,
+      "loss": 0.7873,
+      "step": 3380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.78888988527092e-06,
+      "loss": 0.7917,
+      "step": 3381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.760190746048399e-06,
+      "loss": 0.8104,
+      "step": 3382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.73156081369397e-06,
+      "loss": 0.7794,
+      "step": 3383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.703000109232559e-06,
+      "loss": 0.7801,
+      "step": 3384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.674508653638344e-06,
+      "loss": 0.832,
+      "step": 3385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.646086467834544e-06,
+      "loss": 0.8387,
+      "step": 3386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.617733572693551e-06,
+      "loss": 0.7787,
+      "step": 3387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5894499890369105e-06,
+      "loss": 0.8147,
+      "step": 3388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.561235737635196e-06,
+      "loss": 0.8205,
+      "step": 3389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533090839208133e-06,
+      "loss": 0.8361,
+      "step": 3390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.505015314424455e-06,
+      "loss": 0.8259,
+      "step": 3391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.477009183901982e-06,
+      "loss": 0.2955,
+      "step": 3392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.449072468207583e-06,
+      "loss": 0.8147,
+      "step": 3393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4212051878571475e-06,
+      "loss": 0.8051,
+      "step": 3394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.393407363315528e-06,
+      "loss": 0.7635,
+      "step": 3395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.365679014996639e-06,
+      "loss": 0.74,
+      "step": 3396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.338020163263336e-06,
+      "loss": 0.7612,
+      "step": 3397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.310430828427437e-06,
+      "loss": 0.8618,
+      "step": 3398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282911030749726e-06,
+      "loss": 0.7965,
+      "step": 3399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.255460790439881e-06,
+      "loss": 0.8225,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.228080127656565e-06,
+      "loss": 0.8002,
+      "step": 3401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.200769062507305e-06,
+      "loss": 0.8427,
+      "step": 3402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1735276150484905e-06,
+      "loss": 0.8267,
+      "step": 3403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146355805285452e-06,
+      "loss": 0.8085,
+      "step": 3404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.119253653172329e-06,
+      "loss": 0.3266,
+      "step": 3405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0922211786121334e-06,
+      "loss": 0.3197,
+      "step": 3406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.065258401456685e-06,
+      "loss": 0.7474,
+      "step": 3407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.038365341506634e-06,
+      "loss": 0.8272,
+      "step": 3408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.011542018511428e-06,
+      "loss": 0.7643,
+      "step": 3409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984788452169331e-06,
+      "loss": 0.8105,
+      "step": 3410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958104662127316e-06,
+      "loss": 0.3249,
+      "step": 3411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.931490667981175e-06,
+      "loss": 0.7817,
+      "step": 3412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.90494648927543e-06,
+      "loss": 0.7876,
+      "step": 3413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8784721455032875e-06,
+      "loss": 0.8013,
+      "step": 3414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852067656106763e-06,
+      "loss": 0.861,
+      "step": 3415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.825733040476465e-06,
+      "loss": 0.8094,
+      "step": 3416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.799468317951739e-06,
+      "loss": 0.8246,
+      "step": 3417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.773273507820641e-06,
+      "loss": 0.8477,
+      "step": 3418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.74714862931982e-06,
+      "loss": 0.8143,
+      "step": 3419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.721093701634593e-06,
+      "loss": 0.7821,
+      "step": 3420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.695108743898935e-06,
+      "loss": 0.7941,
+      "step": 3421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.669193775195402e-06,
+      "loss": 0.8315,
+      "step": 3422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.643348814555159e-06,
+      "loss": 0.8445,
+      "step": 3423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.61757388095797e-06,
+      "loss": 0.7628,
+      "step": 3424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.59186899333216e-06,
+      "loss": 0.3475,
+      "step": 3425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.566234170554639e-06,
+      "loss": 0.7947,
+      "step": 3426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.540669431450828e-06,
+      "loss": 0.8118,
+      "step": 3427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.515174794794697e-06,
+      "loss": 0.7946,
+      "step": 3428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.489750279308757e-06,
+      "loss": 0.8229,
+      "step": 3429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4643959036639985e-06,
+      "loss": 0.3179,
+      "step": 3430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4391116864798796e-06,
+      "loss": 0.8279,
+      "step": 3431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4138976463244165e-06,
+      "loss": 0.7703,
+      "step": 3432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.388753801713985e-06,
+      "loss": 0.8013,
+      "step": 3433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3636801711135045e-06,
+      "loss": 0.7776,
+      "step": 3434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3386767729362785e-06,
+      "loss": 0.8516,
+      "step": 3435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3137436255440315e-06,
+      "loss": 0.821,
+      "step": 3436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2888807472469375e-06,
+      "loss": 0.8189,
+      "step": 3437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.264088156303536e-06,
+      "loss": 0.8169,
+      "step": 3438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.239365870920741e-06,
+      "loss": 0.3273,
+      "step": 3439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.214713909253887e-06,
+      "loss": 0.8248,
+      "step": 3440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.19013228940659e-06,
+      "loss": 0.8495,
+      "step": 3441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.165621029430855e-06,
+      "loss": 0.7978,
+      "step": 3442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.14118014732704e-06,
+      "loss": 0.8216,
+      "step": 3443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.116809661043774e-06,
+      "loss": 0.8345,
+      "step": 3444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.092509588477988e-06,
+      "loss": 0.7943,
+      "step": 3445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.06827994747494e-06,
+      "loss": 0.7814,
+      "step": 3446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.044120755828152e-06,
+      "loss": 0.3023,
+      "step": 3447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0200320312793975e-06,
+      "loss": 0.8231,
+      "step": 3448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9960137915186934e-06,
+      "loss": 0.7885,
+      "step": 3449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.97206605418432e-06,
+      "loss": 0.8056,
+      "step": 3450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.948188836862776e-06,
+      "loss": 0.7662,
+      "step": 3451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.924382157088769e-06,
+      "loss": 0.7818,
+      "step": 3452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.90064603234519e-06,
+      "loss": 0.8134,
+      "step": 3453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.876980480063153e-06,
+      "loss": 0.7755,
+      "step": 3454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.85338551762191e-06,
+      "loss": 0.3387,
+      "step": 3455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.829861162348891e-06,
+      "loss": 0.8076,
+      "step": 3456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8064074315196984e-06,
+      "loss": 0.8619,
+      "step": 3457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.783024342357988e-06,
+      "loss": 0.7978,
+      "step": 3458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.759711912035635e-06,
+      "loss": 0.7569,
+      "step": 3459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.736470157672578e-06,
+      "loss": 0.3552,
+      "step": 3460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.713299096336842e-06,
+      "loss": 0.7867,
+      "step": 3461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6901987450445707e-06,
+      "loss": 0.856,
+      "step": 3462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6671691207599614e-06,
+      "loss": 0.7898,
+      "step": 3463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.644210240395263e-06,
+      "loss": 0.8136,
+      "step": 3464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6213221208108216e-06,
+      "loss": 0.7778,
+      "step": 3465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5985047788149373e-06,
+      "loss": 0.7663,
+      "step": 3466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.575758231163984e-06,
+      "loss": 0.8112,
+      "step": 3467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5530824945623542e-06,
+      "loss": 0.8292,
+      "step": 3468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5304775856623952e-06,
+      "loss": 0.8326,
+      "step": 3469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.507943521064505e-06,
+      "loss": 0.3189,
+      "step": 3470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4854803173170135e-06,
+      "loss": 0.8032,
+      "step": 3471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4630879909161917e-06,
+      "loss": 0.8318,
+      "step": 3472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4407665583063186e-06,
+      "loss": 0.7515,
+      "step": 3473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.418516035879571e-06,
+      "loss": 0.3167,
+      "step": 3474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3963364399760332e-06,
+      "loss": 0.3364,
+      "step": 3475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.374227786883777e-06,
+      "loss": 0.8195,
+      "step": 3476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3521900928386917e-06,
+      "loss": 0.83,
+      "step": 3477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3302233740245993e-06,
+      "loss": 0.806,
+      "step": 3478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.308327646573206e-06,
+      "loss": 0.7705,
+      "step": 3479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2865029265640723e-06,
+      "loss": 0.8286,
+      "step": 3480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2647492300245994e-06,
+      "loss": 0.8014,
+      "step": 3481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.243066572930087e-06,
+      "loss": 0.8039,
+      "step": 3482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2214549712035524e-06,
+      "loss": 0.3164,
+      "step": 3483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1999144407159566e-06,
+      "loss": 0.8146,
+      "step": 3484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1784449972860007e-06,
+      "loss": 0.8277,
+      "step": 3485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1570466566801737e-06,
+      "loss": 0.7752,
+      "step": 3486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.135719434612805e-06,
+      "loss": 0.854,
+      "step": 3487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.114463346745955e-06,
+      "loss": 0.803,
+      "step": 3488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.093278408689426e-06,
+      "loss": 0.7798,
+      "step": 3489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0721646360008293e-06,
+      "loss": 0.7926,
+      "step": 3490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0511220441854505e-06,
+      "loss": 0.7991,
+      "step": 3491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0301506486963614e-06,
+      "loss": 0.7868,
+      "step": 3492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.00925046493431e-06,
+      "loss": 0.3256,
+      "step": 3493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9884215082477408e-06,
+      "loss": 0.8013,
+      "step": 3494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.96766379393284e-06,
+      "loss": 0.8104,
+      "step": 3495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9469773372334365e-06,
+      "loss": 0.7891,
+      "step": 3496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9263621533410225e-06,
+      "loss": 0.7476,
+      "step": 3497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.905818257394799e-06,
+      "loss": 0.8387,
+      "step": 3498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8853456644815647e-06,
+      "loss": 0.8073,
+      "step": 3499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8649443896357485e-06,
+      "loss": 0.8715,
+      "step": 3500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8446144478394667e-06,
+      "loss": 0.7984,
+      "step": 3501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.824355854022398e-06,
+      "loss": 0.7955,
+      "step": 3502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8041686230618423e-06,
+      "loss": 0.8322,
+      "step": 3503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.784052769782697e-06,
+      "loss": 0.802,
+      "step": 3504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7640083089574333e-06,
+      "loss": 0.3379,
+      "step": 3505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7440352553060876e-06,
+      "loss": 0.8215,
+      "step": 3506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7241336234962944e-06,
+      "loss": 0.7864,
+      "step": 3507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7043034281431734e-06,
+      "loss": 0.8042,
+      "step": 3508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6845446838094533e-06,
+      "loss": 0.8075,
+      "step": 3509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.66485740500535e-06,
+      "loss": 0.7437,
+      "step": 3510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6452416061885864e-06,
+      "loss": 0.778,
+      "step": 3511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6256973017644404e-06,
+      "loss": 0.7984,
+      "step": 3512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6062245060856527e-06,
+      "loss": 0.8297,
+      "step": 3513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.586823233452429e-06,
+      "loss": 0.79,
+      "step": 3514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5674934981125164e-06,
+      "loss": 0.8128,
+      "step": 3515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5482353142610716e-06,
+      "loss": 0.8317,
+      "step": 3516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5290486960407146e-06,
+      "loss": 0.7916,
+      "step": 3517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.50993365754153e-06,
+      "loss": 0.3475,
+      "step": 3518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4908902128010113e-06,
+      "loss": 0.7541,
+      "step": 3519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.471918375804105e-06,
+      "loss": 0.7877,
+      "step": 3520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.453018160483145e-06,
+      "loss": 0.804,
+      "step": 3521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4341895807178717e-06,
+      "loss": 0.7977,
+      "step": 3522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4154326503354363e-06,
+      "loss": 0.82,
+      "step": 3523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.396747383110365e-06,
+      "loss": 0.8098,
+      "step": 3524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3781337927645585e-06,
+      "loss": 0.8076,
+      "step": 3525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.35959189296725e-06,
+      "loss": 0.7716,
+      "step": 3526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.341121697335058e-06,
+      "loss": 0.8056,
+      "step": 3527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.322723219431955e-06,
+      "loss": 0.8176,
+      "step": 3528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3043964727692103e-06,
+      "loss": 0.7656,
+      "step": 3529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2861414708054363e-06,
+      "loss": 0.8349,
+      "step": 3530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.267958226946554e-06,
+      "loss": 0.8279,
+      "step": 3531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2498467545458035e-06,
+      "loss": 0.7972,
+      "step": 3532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2318070669036685e-06,
+      "loss": 0.7745,
+      "step": 3533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2138391772679847e-06,
+      "loss": 0.8052,
+      "step": 3534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.195943098833819e-06,
+      "loss": 0.334,
+      "step": 3535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.178118844743482e-06,
+      "loss": 0.7523,
+      "step": 3536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1603664280866132e-06,
+      "loss": 0.8258,
+      "step": 3537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.142685861900018e-06,
+      "loss": 0.3208,
+      "step": 3538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1250771591677765e-06,
+      "loss": 0.8146,
+      "step": 3539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1075403328212005e-06,
+      "loss": 0.8274,
+      "step": 3540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0900753957388e-06,
+      "loss": 0.8055,
+      "step": 3541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.072682360746281e-06,
+      "loss": 0.7771,
+      "step": 3542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0553612406165932e-06,
+      "loss": 0.7754,
+      "step": 3543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0381120480698066e-06,
+      "loss": 0.7471,
+      "step": 3544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0209347957732328e-06,
+      "loss": 0.8189,
+      "step": 3545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.003829496341325e-06,
+      "loss": 0.8361,
+      "step": 3546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.98679616233568e-06,
+      "loss": 0.8044,
+      "step": 3547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9698348062650917e-06,
+      "loss": 0.8351,
+      "step": 3548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952945440585452e-06,
+      "loss": 0.7802,
+      "step": 3549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9361280776997946e-06,
+      "loss": 0.8094,
+      "step": 3550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9193827299582968e-06,
+      "loss": 0.8325,
+      "step": 3551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.90270940965821e-06,
+      "loss": 0.7482,
+      "step": 3552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8861081290439508e-06,
+      "loss": 0.8112,
+      "step": 3553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8695789003069897e-06,
+      "loss": 0.7649,
+      "step": 3554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.853121735585872e-06,
+      "loss": 0.8084,
+      "step": 3555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.836736646966264e-06,
+      "loss": 0.7891,
+      "step": 3556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820423646480862e-06,
+      "loss": 0.7958,
+      "step": 3557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8041827461094618e-06,
+      "loss": 0.8409,
+      "step": 3558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.788013957778878e-06,
+      "loss": 0.8148,
+      "step": 3559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7719172933629791e-06,
+      "loss": 0.7432,
+      "step": 3560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.755892764682654e-06,
+      "loss": 0.8191,
+      "step": 3561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.739940383505867e-06,
+      "loss": 0.7863,
+      "step": 3562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.724060161547536e-06,
+      "loss": 0.8146,
+      "step": 3563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7082521104696215e-06,
+      "loss": 0.3287,
+      "step": 3564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6925162418811036e-06,
+      "loss": 0.7617,
+      "step": 3565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6768525673378943e-06,
+      "loss": 0.7975,
+      "step": 3566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6612610983429365e-06,
+      "loss": 0.819,
+      "step": 3567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6457418463461493e-06,
+      "loss": 0.8485,
+      "step": 3568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6302948227443604e-06,
+      "loss": 0.7434,
+      "step": 3569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6149200388814511e-06,
+      "loss": 0.8165,
+      "step": 3570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5996175060481567e-06,
+      "loss": 0.8033,
+      "step": 3571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5843872354822097e-06,
+      "loss": 0.8109,
+      "step": 3572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5692292383682638e-06,
+      "loss": 0.8004,
+      "step": 3573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5541435258378923e-06,
+      "loss": 0.7979,
+      "step": 3574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5391301089695776e-06,
+      "loss": 0.7951,
+      "step": 3575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5241889987887336e-06,
+      "loss": 0.7957,
+      "step": 3576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5093202062676503e-06,
+      "loss": 0.824,
+      "step": 3577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4945237423255154e-06,
+      "loss": 0.7845,
+      "step": 3578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4797996178284257e-06,
+      "loss": 0.8103,
+      "step": 3579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4651478435893096e-06,
+      "loss": 0.7957,
+      "step": 3580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4505684303679934e-06,
+      "loss": 0.795,
+      "step": 3581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4360613888711793e-06,
+      "loss": 0.2961,
+      "step": 3582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4216267297523679e-06,
+      "loss": 0.8027,
+      "step": 3583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.407264463611957e-06,
+      "loss": 0.7967,
+      "step": 3584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3929746009971433e-06,
+      "loss": 0.8195,
+      "step": 3585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3787571524019883e-06,
+      "loss": 0.8006,
+      "step": 3586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.364612128267362e-06,
+      "loss": 0.8302,
+      "step": 3587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3505395389809106e-06,
+      "loss": 0.8073,
+      "step": 3588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3365393948771454e-06,
+      "loss": 0.7941,
+      "step": 3589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3226117062373534e-06,
+      "loss": 0.794,
+      "step": 3590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.308756483289586e-06,
+      "loss": 0.7866,
+      "step": 3591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2949737362087156e-06,
+      "loss": 0.8266,
+      "step": 3592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2812634751163676e-06,
+      "loss": 0.7876,
+      "step": 3593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2676257100809442e-06,
+      "loss": 0.8075,
+      "step": 3594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2540604511176224e-06,
+      "loss": 0.8126,
+      "step": 3595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2405677081883005e-06,
+      "loss": 0.785,
+      "step": 3596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2271474912016411e-06,
+      "loss": 0.8014,
+      "step": 3597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.21379981001305e-06,
+      "loss": 0.7809,
+      "step": 3598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2005246744246745e-06,
+      "loss": 0.8056,
+      "step": 3599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1873220941853502e-06,
+      "loss": 0.7802,
+      "step": 3600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1741920789906768e-06,
+      "loss": 0.7713,
+      "step": 3601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1611346384829192e-06,
+      "loss": 0.7896,
+      "step": 3602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1481497822510733e-06,
+      "loss": 0.7994,
+      "step": 3603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.135237519830834e-06,
+      "loss": 0.7568,
+      "step": 3604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1223978607045714e-06,
+      "loss": 0.7824,
+      "step": 3605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1096308143013323e-06,
+      "loss": 0.8251,
+      "step": 3606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.096936389996872e-06,
+      "loss": 0.7321,
+      "step": 3607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0843145971135783e-06,
+      "loss": 0.8231,
+      "step": 3608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.071765444920514e-06,
+      "loss": 0.7969,
+      "step": 3609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0592889426333963e-06,
+      "loss": 0.7949,
+      "step": 3610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0468850994145963e-06,
+      "loss": 0.7984,
+      "step": 3611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0345539243731273e-06,
+      "loss": 0.8262,
+      "step": 3612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0222954265646123e-06,
+      "loss": 0.8187,
+      "step": 3613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0101096149913392e-06,
+      "loss": 0.7982,
+      "step": 3614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.979964986021939e-07,
+      "loss": 0.8091,
+      "step": 3615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.85956086292672e-07,
+      "loss": 0.8455,
+      "step": 3616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.739883869049227e-07,
+      "loss": 0.7916,
+      "step": 3617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.620934092276267e-07,
+      "loss": 0.8126,
+      "step": 3618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.50271161996108e-07,
+      "loss": 0.8227,
+      "step": 3619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.385216538922659e-07,
+      "loss": 0.7827,
+      "step": 3620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.268448935445983e-07,
+      "loss": 0.7786,
+      "step": 3621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.15240889528135e-07,
+      "loss": 0.8019,
+      "step": 3622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.03709650364537e-07,
+      "loss": 0.7623,
+      "step": 3623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.922511845219971e-07,
+      "loss": 0.8292,
+      "step": 3624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.8086550041524e-07,
+      "loss": 0.8047,
+      "step": 3625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.695526064055881e-07,
+      "loss": 0.8173,
+      "step": 3626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.583125108008738e-07,
+      "loss": 0.8923,
+      "step": 3627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.471452218555053e-07,
+      "loss": 0.8169,
+      "step": 3628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.360507477703894e-07,
+      "loss": 0.8139,
+      "step": 3629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.250290966929641e-07,
+      "loss": 0.3131,
+      "step": 3630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.140802767172217e-07,
+      "loss": 0.786,
+      "step": 3631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.032042958836195e-07,
+      "loss": 0.8203,
+      "step": 3632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.924011621791571e-07,
+      "loss": 0.8042,
+      "step": 3633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.816708835373443e-07,
+      "loss": 0.322,
+      "step": 3634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.710134678381442e-07,
+      "loss": 0.8145,
+      "step": 3635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.60428922908063e-07,
+      "loss": 0.3268,
+      "step": 3636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.499172565200718e-07,
+      "loss": 0.7756,
+      "step": 3637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.394784763936069e-07,
+      "loss": 0.7964,
+      "step": 3638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.291125901946027e-07,
+      "loss": 0.8183,
+      "step": 3639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.188196055354479e-07,
+      "loss": 0.8419,
+      "step": 3640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.085995299749959e-07,
+      "loss": 0.8024,
+      "step": 3641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.984523710185875e-07,
+      "loss": 0.7614,
+      "step": 3642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.883781361179509e-07,
+      "loss": 0.741,
+      "step": 3643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.783768326713235e-07,
+      "loss": 0.7803,
+      "step": 3644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.684484680233527e-07,
+      "loss": 0.8556,
+      "step": 3645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.585930494651283e-07,
+      "loss": 0.3071,
+      "step": 3646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.488105842341829e-07,
+      "loss": 0.8398,
+      "step": 3647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.39101079514448e-07,
+      "loss": 0.757,
+      "step": 3648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.294645424362978e-07,
+      "loss": 0.7785,
+      "step": 3649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.199009800765265e-07,
+      "loss": 0.8656,
+      "step": 3650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.104103994583054e-07,
+      "loss": 0.8334,
+      "step": 3651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.009928075512372e-07,
+      "loss": 0.8341,
+      "step": 3652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.916482112713118e-07,
+      "loss": 0.8028,
+      "step": 3653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.823766174809397e-07,
+      "loss": 0.7858,
+      "step": 3654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.731780329888636e-07,
+      "loss": 0.75,
+      "step": 3655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.640524645502798e-07,
+      "loss": 0.8298,
+      "step": 3656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.549999188667054e-07,
+      "loss": 0.7562,
+      "step": 3657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.460204025860671e-07,
+      "loss": 0.7815,
+      "step": 3658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.371139223026678e-07,
+      "loss": 0.7816,
+      "step": 3659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.282804845571199e-07,
+      "loss": 0.8085,
+      "step": 3660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.195200958364566e-07,
+      "loss": 0.8307,
+      "step": 3661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.108327625740539e-07,
+      "loss": 0.807,
+      "step": 3662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.022184911495864e-07,
+      "loss": 0.7729,
+      "step": 3663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.936772878891605e-07,
+      "loss": 0.7676,
+      "step": 3664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.852091590651475e-07,
+      "loss": 0.765,
+      "step": 3665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7681411089630647e-07,
+      "loss": 0.8294,
+      "step": 3666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.684921495476835e-07,
+      "loss": 0.7999,
+      "step": 3667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6024328113069e-07,
+      "loss": 0.8447,
+      "step": 3668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5206751170305817e-07,
+      "loss": 0.8051,
+      "step": 3669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.439648472687963e-07,
+      "loss": 0.8028,
+      "step": 3670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.359352937782668e-07,
+      "loss": 0.8339,
+      "step": 3671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.279788571281529e-07,
+      "loss": 0.8261,
+      "step": 3672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.200955431613918e-07,
+      "loss": 0.7847,
+      "step": 3673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1228535766726364e-07,
+      "loss": 0.3071,
+      "step": 3674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.045483063813471e-07,
+      "loss": 0.7767,
+      "step": 3675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.96884394985475e-07,
+      "loss": 0.8076,
+      "step": 3676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8929362910778976e-07,
+      "loss": 0.8145,
+      "step": 3677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8177601432274334e-07,
+      "loss": 0.7959,
+      "step": 3678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7433155615103076e-07,
+      "loss": 0.8268,
+      "step": 3679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6696026005964556e-07,
+      "loss": 0.8274,
+      "step": 3680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.596621314618354e-07,
+      "loss": 0.7925,
+      "step": 3681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.524371757171241e-07,
+      "loss": 0.8516,
+      "step": 3682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.45285398131312e-07,
+      "loss": 0.7732,
+      "step": 3683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3820680395644235e-07,
+      "loss": 0.859,
+      "step": 3684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3120139839082353e-07,
+      "loss": 0.793,
+      "step": 3685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2426918657900704e-07,
+      "loss": 0.7335,
+      "step": 3686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.174101736117985e-07,
+      "loss": 0.7865,
+      "step": 3687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1062436452625745e-07,
+      "loss": 0.8077,
+      "step": 3688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0391176430567547e-07,
+      "loss": 0.8,
+      "step": 3689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.972723778795872e-07,
+      "loss": 0.3088,
+      "step": 3690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9070621012374786e-07,
+      "loss": 0.7832,
+      "step": 3691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8421326586016705e-07,
+      "loss": 0.765,
+      "step": 3692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7779354985704166e-07,
+      "loss": 0.7665,
+      "step": 3693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.714470668288449e-07,
+      "loss": 0.8493,
+      "step": 3694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.651738214362265e-07,
+      "loss": 0.8419,
+      "step": 3695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5897381828606794e-07,
+      "loss": 0.7676,
+      "step": 3696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5284706193147157e-07,
+      "loss": 0.7775,
+      "step": 3697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4679355687173834e-07,
+      "loss": 0.8123,
+      "step": 3698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4081330755237885e-07,
+      "loss": 0.7832,
+      "step": 3699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3490631836510234e-07,
+      "loss": 0.7732,
+      "step": 3700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2907259364783884e-07,
+      "loss": 0.8293,
+      "step": 3701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2331213768468363e-07,
+      "loss": 0.8107,
+      "step": 3702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.176249547059528e-07,
+      "loss": 0.7657,
+      "step": 3703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1201104888812772e-07,
+      "loss": 0.7814,
+      "step": 3704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.064704243539106e-07,
+      "loss": 0.7776,
+      "step": 3705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0100308517214672e-07,
+      "loss": 0.7486,
+      "step": 3706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9560903535790208e-07,
+      "loss": 0.805,
+      "step": 3707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9028827887239697e-07,
+      "loss": 0.7784,
+      "step": 3708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.850408196230391e-07,
+      "loss": 0.8095,
+      "step": 3709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7986666146339038e-07,
+      "loss": 0.7935,
+      "step": 3710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7476580819321132e-07,
+      "loss": 0.7703,
+      "step": 3711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.697382635584055e-07,
+      "loss": 0.7878,
+      "step": 3712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6478403125106401e-07,
+      "loss": 0.8379,
+      "step": 3713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.599031149093988e-07,
+      "loss": 0.8123,
+      "step": 3714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5509551811783152e-07,
+      "loss": 0.3326,
+      "step": 3715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.503612444069047e-07,
+      "loss": 0.8223,
+      "step": 3716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4570029725333724e-07,
+      "loss": 0.7621,
+      "step": 3717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4111268007996891e-07,
+      "loss": 0.8086,
+      "step": 3718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3659839625581594e-07,
+      "loss": 0.3304,
+      "step": 3719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3215744909603755e-07,
+      "loss": 0.7586,
+      "step": 3720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.27789841861925e-07,
+      "loss": 0.8149,
+      "step": 3721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.234955777609237e-07,
+      "loss": 0.2999,
+      "step": 3722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1927465994659992e-07,
+      "loss": 0.791,
+      "step": 3723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1512709151868528e-07,
+      "loss": 0.776,
+      "step": 3724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1105287552301003e-07,
+      "loss": 0.7898,
+      "step": 3725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0705201495155858e-07,
+      "loss": 0.3173,
+      "step": 3726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0312451274245849e-07,
+      "loss": 0.7464,
+      "step": 3727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.927037177993592e-08,
+      "loss": 0.8376,
+      "step": 3728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.548959489434905e-08,
+      "loss": 0.8322,
+      "step": 3729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.178218486219914e-08,
+      "loss": 0.823,
+      "step": 3730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.81481444060972e-08,
+      "loss": 0.7437,
+      "step": 3731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.458747619475293e-08,
+      "loss": 0.8523,
+      "step": 3732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.110018284304133e-08,
+      "loss": 0.8091,
+      "step": 3733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.768626691190272e-08,
+      "loss": 0.8046,
+      "step": 3734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.434573090842056e-08,
+      "loss": 0.8015,
+      "step": 3735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.107857728578804e-08,
+      "loss": 0.8043,
+      "step": 3736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.788480844329703e-08,
+      "loss": 0.784,
+      "step": 3737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.47644267263603e-08,
+      "loss": 0.8162,
+      "step": 3738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.171743442650035e-08,
+      "loss": 0.8262,
+      "step": 3739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.87438337813162e-08,
+      "loss": 0.7561,
+      "step": 3740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.584362697453882e-08,
+      "loss": 0.7858,
+      "step": 3741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.301681613599785e-08,
+      "loss": 0.3059,
+      "step": 3742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.02634033416105e-08,
+      "loss": 0.7534,
+      "step": 3743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.758339061341488e-08,
+      "loss": 0.8481,
+      "step": 3744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.497677991950333e-08,
+      "loss": 0.7975,
+      "step": 3745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.24435731741224e-08,
+      "loss": 0.8096,
+      "step": 3746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.998377223756178e-08,
+      "loss": 0.7858,
+      "step": 3747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.759737891623205e-08,
+      "loss": 0.804,
+      "step": 3748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5284394962631365e-08,
+      "loss": 0.8462,
+      "step": 3749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.304482207533433e-08,
+      "loss": 0.7606,
+      "step": 3750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.087866189903643e-08,
+      "loss": 0.7966,
+      "step": 3751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8785916024476334e-08,
+      "loss": 0.7637,
+      "step": 3752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.676658598851356e-08,
+      "loss": 0.8265,
+      "step": 3753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.482067327409521e-08,
+      "loss": 0.8457,
+      "step": 3754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.294817931022264e-08,
+      "loss": 0.8258,
+      "step": 3755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1149105471995888e-08,
+      "loss": 0.8227,
+      "step": 3756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9423453080624765e-08,
+      "loss": 0.8425,
+      "step": 3757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7771223403351133e-08,
+      "loss": 0.8037,
+      "step": 3758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6192417653537738e-08,
+      "loss": 0.8013,
+      "step": 3759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4687036990601588e-08,
+      "loss": 0.7912,
+      "step": 3760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3255082520058359e-08,
+      "loss": 0.7735,
+      "step": 3761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1896555293489097e-08,
+      "loss": 0.8055,
+      "step": 3762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0611456308551315e-08,
+      "loss": 0.8369,
+      "step": 3763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.399786508990095e-09,
+      "loss": 0.7699,
+      "step": 3764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.26154678461588e-09,
+      "loss": 0.8179,
+      "step": 3765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.19673797130449e-09,
+      "loss": 0.8082,
+      "step": 3766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.205360851041508e-09,
+      "loss": 0.8176,
+      "step": 3767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.287416151844582e-09,
+      "loss": 0.8243,
+      "step": 3768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.442904547841131e-09,
+      "loss": 0.8127,
+      "step": 3769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.671826659201738e-09,
+      "loss": 0.7837,
+      "step": 3770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.974183052195656e-09,
+      "loss": 0.2948,
+      "step": 3771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.349974239146402e-09,
+      "loss": 0.7936,
+      "step": 3772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7992006784428584e-09,
+      "loss": 0.79,
+      "step": 3773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3218627745725797e-09,
+      "loss": 0.7696,
+      "step": 3774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.179608780662819e-10,
+      "loss": 0.8703,
+      "step": 3775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8749528554225e-10,
+      "loss": 0.7375,
+      "step": 3776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3046623968413514e-10,
+      "loss": 0.8101,
+      "step": 3777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4687392924095378e-10,
+      "loss": 0.8031,
+      "step": 3778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.671848904929221e-11,
+      "loss": 0.7891,
+      "step": 3779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5778,
+      "step": 3780
+    },
+    {
+      "epoch": 1.0,
+      "step": 3780,
+      "total_flos": 8791679674089472.0,
+      "train_loss": 0.8149038041079486,
+      "train_runtime": 120337.8265,
+      "train_samples_per_second": 5.529,
+      "train_steps_per_second": 0.031
+    }
+  ],
+  "max_steps": 3780,
+  "num_train_epochs": 1,
+  "total_flos": 8791679674089472.0,
+  "trial_name": null,
+  "trial_params": null
+}