diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31201 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5196,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 1.9121,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 1.8052,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1.9185,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.6846,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 1.8999,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.7012,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.6997,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 1.7114,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 1.6631,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 1.6396,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 1.6694,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.4072,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.4321,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.4541,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.3589,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 0.5197,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 1.2705,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 1.2019,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 1.2495,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 1.2524,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.5291,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 1.2004,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 1.2014,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 1.1709,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 1.1411,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.1406,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 1.209,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 1.1204,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 1.0979,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 1.0994,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 1.0957,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 1.0955,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 0.5317,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 1.0491,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 1.0811,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 1.0486,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 1.0491,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 1.0977,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 1.0598,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 1.0457,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.256410256410257e-05,
+      "loss": 1.082,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 0.9895,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.512820512820514e-05,
+      "loss": 1.0425,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 1.0208,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769230769230769e-05,
+      "loss": 1.0122,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 0.9482,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.025641025641026e-05,
+      "loss": 1.0181,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 1.0166,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.282051282051282e-05,
+      "loss": 0.9878,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 0.4897,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.538461538461539e-05,
+      "loss": 0.947,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.9819,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794871794871795e-05,
+      "loss": 0.9834,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 1.0066,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.051282051282052e-05,
+      "loss": 1.0388,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 1.0071,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.307692307692307e-05,
+      "loss": 0.9753,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 0.9778,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.564102564102564e-05,
+      "loss": 0.9697,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.9805,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.820512820512821e-05,
+      "loss": 0.9316,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 0.946,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.076923076923078e-05,
+      "loss": 0.9331,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 0.483,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 0.9209,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 0.9392,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.58974358974359e-05,
+      "loss": 0.9685,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 0.9436,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 0.936,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 0.9043,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.102564102564103e-05,
+      "loss": 0.9392,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 0.9106,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.35897435897436e-05,
+      "loss": 0.9436,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 0.9543,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 0.8994,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 0.925,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871794871794872e-05,
+      "loss": 0.9197,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.9136,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010128205128205129,
+      "loss": 0.9141,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010256410256410256,
+      "loss": 0.9221,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 0.9895,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010512820512820514,
+      "loss": 0.9397,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010641025641025641,
+      "loss": 0.9521,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 0.9417,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010897435897435896,
+      "loss": 0.9443,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011025641025641027,
+      "loss": 0.9441,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011153846153846154,
+      "loss": 0.8948,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011282051282051283,
+      "loss": 0.9141,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001141025641025641,
+      "loss": 0.8694,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.9612,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.9219,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011794871794871796,
+      "loss": 0.9219,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011923076923076923,
+      "loss": 0.8606,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012051282051282052,
+      "loss": 0.8733,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012179487179487179,
+      "loss": 0.8833,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.9224,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012435897435897437,
+      "loss": 0.8989,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.9292,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 0.8762,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.8823,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001294871794871795,
+      "loss": 0.8845,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.9148,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013205128205128204,
+      "loss": 0.9314,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.8855,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 0.9202,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.8948,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013717948717948718,
+      "loss": 0.8926,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 0.9194,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013974358974358974,
+      "loss": 0.9373,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.8567,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 0.9563,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.9441,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014487179487179488,
+      "loss": 0.8857,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.9185,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014743589743589745,
+      "loss": 0.9321,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.877,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.8667,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.854,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015256410256410255,
+      "loss": 0.9348,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.8577,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015512820512820515,
+      "loss": 0.9114,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.8613,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 0.9058,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.9236,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016025641025641028,
+      "loss": 0.8748,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.8831,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016282051282051282,
+      "loss": 0.9116,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.8103,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 0.875,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.8564,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 0.8723,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.8384,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017051282051282053,
+      "loss": 0.9341,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.9121,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 0.8423,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.8655,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017564102564102566,
+      "loss": 0.9504,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.9031,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017820512820512823,
+      "loss": 0.8557,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.9036,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 0.8752,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.8877,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.9048,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.865,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001858974358974359,
+      "loss": 0.8606,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 0.8735,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 0.8472,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.8894,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019102564102564104,
+      "loss": 0.8989,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 0.9043,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001935897435897436,
+      "loss": 0.8687,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 0.8875,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 0.9009,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 0.8689,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019871794871794874,
+      "loss": 0.8721,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8569,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999998057286926,
+      "loss": 0.9202,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999999222914846,
+      "loss": 0.9001,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999982515586868,
+      "loss": 0.8879,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999996891660592,
+      "loss": 0.8818,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999951432210904,
+      "loss": 0.8953,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999993006240861,
+      "loss": 0.9294,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999904807207347,
+      "loss": 0.8792,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999875666616918,
+      "loss": 0.9241,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999842640648654,
+      "loss": 0.8398,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999805729315381,
+      "loss": 0.8359,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999976493263145,
+      "loss": 0.8669,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999720250612698,
+      "loss": 0.8982,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999671683276497,
+      "loss": 0.8779,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999619230641713,
+      "loss": 0.8533,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999956289272873,
+      "loss": 0.8564,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999502669559432,
+      "loss": 0.8674,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999943856115722,
+      "loss": 0.8779,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999370567547008,
+      "loss": 0.8833,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999298688755209,
+      "loss": 0.9304,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999922292480975,
+      "loss": 0.8428,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999143275740072,
+      "loss": 0.8958,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999905974157712,
+      "loss": 0.8777,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998972322353352,
+      "loss": 0.8171,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998881018102737,
+      "loss": 0.8047,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998785828860742,
+      "loss": 0.4299,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999868675466436,
+      "loss": 0.8899,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998583795552083,
+      "loss": 0.8811,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998476951563915,
+      "loss": 0.8655,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998366222741368,
+      "loss": 0.8984,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998251609127464,
+      "loss": 0.8723,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998133110766742,
+      "loss": 0.8352,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998010727705236,
+      "loss": 0.8816,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199978844599905,
+      "loss": 0.8813,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997754307671598,
+      "loss": 0.9255,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997620270799092,
+      "loss": 0.8657,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997482349425066,
+      "loss": 0.8374,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999734054360311,
+      "loss": 0.8552,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997194853388318,
+      "loss": 0.8804,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997045278837296,
+      "loss": 0.895,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996891820008164,
+      "loss": 0.8457,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996734476960545,
+      "loss": 0.8855,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996573249755572,
+      "loss": 0.9163,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996408138455896,
+      "loss": 0.8242,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996239143125658,
+      "loss": 0.856,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996066263830531,
+      "loss": 0.8223,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999588950063768,
+      "loss": 0.9082,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995708853615783,
+      "loss": 0.8999,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995524322835034,
+      "loss": 0.8596,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999533590836713,
+      "loss": 0.823,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995143610285277,
+      "loss": 0.8853,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994947428664194,
+      "loss": 0.8489,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199947473635801,
+      "loss": 0.874,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994543415110733,
+      "loss": 0.8462,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994335583335335,
+      "loss": 0.844,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994123868334655,
+      "loss": 0.9048,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999390827019096,
+      "loss": 0.8704,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993688788988009,
+      "loss": 0.8921,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999346542481109,
+      "loss": 0.8425,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993238177746982,
+      "loss": 0.8403,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993007047883988,
+      "loss": 0.8538,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992772035311903,
+      "loss": 0.8274,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992533140122042,
+      "loss": 0.8796,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999229036240723,
+      "loss": 0.8704,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992043702261793,
+      "loss": 0.896,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991793159781569,
+      "loss": 0.845,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991538735063906,
+      "loss": 0.8257,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999128042820766,
+      "loss": 0.8979,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991018239313188,
+      "loss": 0.8313,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990752168482369,
+      "loss": 0.8621,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999048221581858,
+      "loss": 0.9084,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990208381426706,
+      "loss": 0.4559,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989930665413147,
+      "loss": 0.8232,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989649067885808,
+      "loss": 0.8691,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199893635889541,
+      "loss": 0.8511,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998907422872894,
+      "loss": 0.8396,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988780987322766,
+      "loss": 0.8796,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988483864849508,
+      "loss": 0.8933,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988182861424613,
+      "loss": 0.845,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998787797716503,
+      "loss": 0.8535,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987569212189224,
+      "loss": 0.8413,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987256566617162,
+      "loss": 0.9023,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998694004057032,
+      "loss": 0.8489,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986619634171682,
+      "loss": 0.3568,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998629534754574,
+      "loss": 0.8567,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998596718081849,
+      "loss": 0.8982,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985635134117443,
+      "loss": 0.8545,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985299207571613,
+      "loss": 0.8379,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998495940131152,
+      "loss": 0.874,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998461571546919,
+      "loss": 0.8855,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984268150178167,
+      "loss": 0.8003,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983916705573488,
+      "loss": 0.8279,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998356138179171,
+      "loss": 0.8586,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983202178970887,
+      "loss": 0.8469,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982839097250587,
+      "loss": 0.8301,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982472136771882,
+      "loss": 0.368,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998210129767735,
+      "loss": 0.855,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981726580111083,
+      "loss": 0.8362,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998134798421867,
+      "loss": 0.7957,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980965510147213,
+      "loss": 0.3776,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998057915804532,
+      "loss": 0.8818,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980188928063103,
+      "loss": 0.8633,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979794820352187,
+      "loss": 0.8362,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979396835065696,
+      "loss": 0.886,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978994972358265,
+      "loss": 0.8967,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978589232386035,
+      "loss": 0.8857,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978179615306655,
+      "loss": 0.8635,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977766121279277,
+      "loss": 0.8933,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997734875046456,
+      "loss": 0.8149,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976927503024676,
+      "loss": 0.8716,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997650237912329,
+      "loss": 0.8369,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976073378925584,
+      "loss": 0.8147,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975640502598244,
+      "loss": 0.8845,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997520375030946,
+      "loss": 0.8528,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974763122228927,
+      "loss": 0.834,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974318618527849,
+      "loss": 0.8789,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973870239378938,
+      "loss": 0.8718,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973417984956403,
+      "loss": 0.8516,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997296185543597,
+      "loss": 0.3734,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972501850994858,
+      "loss": 0.8779,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972037971811802,
+      "loss": 0.3873,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997157021806704,
+      "loss": 0.8657,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971098589942313,
+      "loss": 0.8391,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970623087620868,
+      "loss": 0.8418,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970143711287456,
+      "loss": 0.8594,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969660461128342,
+      "loss": 0.8311,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996917333733128,
+      "loss": 0.8228,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968682340085545,
+      "loss": 0.8606,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996818746958191,
+      "loss": 0.9099,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996768872601265,
+      "loss": 0.8633,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967186109571552,
+      "loss": 0.8523,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199666796204539,
+      "loss": 0.8303,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966169258856488,
+      "loss": 0.9194,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965655024977618,
+      "loss": 0.8191,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965136919017083,
+      "loss": 0.8474,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964614941176195,
+      "loss": 0.375,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996408909165776,
+      "loss": 0.8687,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963559370666102,
+      "loss": 0.8435,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996302577840703,
+      "loss": 0.8586,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962488315087873,
+      "loss": 0.8674,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961946980917456,
+      "loss": 0.8484,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961401776106113,
+      "loss": 0.9031,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960852700865678,
+      "loss": 0.9155,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960299755409492,
+      "loss": 0.8694,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959742939952392,
+      "loss": 0.8286,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995918225471073,
+      "loss": 0.8147,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958617699902353,
+      "loss": 0.8752,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995804927574662,
+      "loss": 0.8801,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995747698246438,
+      "loss": 0.8413,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956900820278,
+      "loss": 0.8552,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956320789411337,
+      "loss": 0.8882,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955736890089768,
+      "loss": 0.8276,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955149122540152,
+      "loss": 0.8245,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954557486990868,
+      "loss": 0.8345,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953961983671788,
+      "loss": 0.9055,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953362612814296,
+      "loss": 0.8577,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952759374651264,
+      "loss": 0.8562,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952152269417085,
+      "loss": 0.845,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951541297347642,
+      "loss": 0.8899,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995092645868032,
+      "loss": 0.7896,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950307753654017,
+      "loss": 0.8604,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949685182509117,
+      "loss": 0.854,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949058745487522,
+      "loss": 0.8428,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994842844283263,
+      "loss": 0.8037,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994779427478934,
+      "loss": 0.8484,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994715624160405,
+      "loss": 0.8894,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946514343524666,
+      "loss": 0.822,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945868580800594,
+      "loss": 0.8462,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019945218953682734,
+      "loss": 0.7722,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199445654624235,
+      "loss": 0.8359,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943908107276798,
+      "loss": 0.8157,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943246888498041,
+      "loss": 0.8586,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994258180634414,
+      "loss": 0.8628,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941912861073507,
+      "loss": 0.8303,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941240052946053,
+      "loss": 0.8435,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940563382223197,
+      "loss": 0.8125,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939882849167852,
+      "loss": 0.8127,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939198454044439,
+      "loss": 0.8298,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938510197118866,
+      "loss": 0.8203,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937818078658557,
+      "loss": 0.8486,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937122098932428,
+      "loss": 0.8235,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936422258210894,
+      "loss": 0.7781,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935718556765876,
+      "loss": 0.8682,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993501099487079,
+      "loss": 0.8433,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934299572800556,
+      "loss": 0.8401,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993358429083159,
+      "loss": 0.825,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932865149241812,
+      "loss": 0.8259,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932142148310636,
+      "loss": 0.8118,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931415288318986,
+      "loss": 0.8391,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930684569549264,
+      "loss": 0.8384,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929949992285396,
+      "loss": 0.7891,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929211556812796,
+      "loss": 0.853,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928469263418374,
+      "loss": 0.8762,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927723112390545,
+      "loss": 0.8267,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992697310401922,
+      "loss": 0.8101,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992621923859581,
+      "loss": 0.8174,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925461516413223,
+      "loss": 0.8188,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924699937765863,
+      "loss": 0.8677,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923934502949644,
+      "loss": 0.8743,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992316521226196,
+      "loss": 0.8213,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922392066001722,
+      "loss": 0.8535,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921615064469325,
+      "loss": 0.8247,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992083420796667,
+      "loss": 0.8604,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920049496797152,
+      "loss": 0.8398,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919260931265664,
+      "loss": 0.8352,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918468511678596,
+      "loss": 0.8447,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991767223834384,
+      "loss": 0.8513,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916872111570784,
+      "loss": 0.8225,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916068131670302,
+      "loss": 0.8606,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915260298954785,
+      "loss": 0.3713,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914448613738106,
+      "loss": 0.8079,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991363307633564,
+      "loss": 0.8765,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912813687064253,
+      "loss": 0.8328,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911990446242319,
+      "loss": 0.8713,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199111633541897,
+      "loss": 0.832,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910332411227758,
+      "loss": 0.908,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909497617679348,
+      "loss": 0.8325,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908658973868823,
+      "loss": 0.8623,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907816480122033,
+      "loss": 0.8257,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990697013676632,
+      "loss": 0.8184,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990611994413053,
+      "loss": 0.873,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905265902544993,
+      "loss": 0.8337,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904408012341542,
+      "loss": 0.8989,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903546273853504,
+      "loss": 0.8494,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019902680687415705,
+      "loss": 0.8564,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901811253364456,
+      "loss": 0.8271,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900937972037577,
+      "loss": 0.7847,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900060843774364,
+      "loss": 0.8491,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989917986891563,
+      "loss": 0.8032,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898295047803664,
+      "loss": 0.8245,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897406380782261,
+      "loss": 0.8379,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896513868196704,
+      "loss": 0.8479,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019895617510393772,
+      "loss": 0.8467,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019894717307721734,
+      "loss": 0.7849,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893813260530368,
+      "loss": 0.8145,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892905369170926,
+      "loss": 0.8696,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891993633996163,
+      "loss": 0.8408,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891078055360334,
+      "loss": 0.8059,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989015863361917,
+      "loss": 0.8928,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889235369129911,
+      "loss": 0.8342,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888308262251285,
+      "loss": 0.428,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887377313343514,
+      "loss": 0.8486,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988644252276831,
+      "loss": 0.8655,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885503890888876,
+      "loss": 0.8228,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988456141806991,
+      "loss": 0.8323,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883615104677608,
+      "loss": 0.8662,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988266495107965,
+      "loss": 0.4093,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988171095764521,
+      "loss": 0.8494,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019880753124744963,
+      "loss": 0.8188,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879791452751057,
+      "loss": 0.8323,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878825942037148,
+      "loss": 0.8418,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019877856592978376,
+      "loss": 0.863,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019876883405951377,
+      "loss": 0.8457,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019875906381334275,
+      "loss": 0.8452,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874925519506684,
+      "loss": 0.8699,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873940820849714,
+      "loss": 0.8176,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872952285745959,
+      "loss": 0.8435,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871959914579508,
+      "loss": 0.8306,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019870963707735938,
+      "loss": 0.8655,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986996366560232,
+      "loss": 0.8521,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868959788567212,
+      "loss": 0.793,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019867952077020666,
+      "loss": 0.8123,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866940531354215,
+      "loss": 0.8428,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865925151960894,
+      "loss": 0.8396,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019864905939235214,
+      "loss": 0.8345,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863882893573188,
+      "loss": 0.8286,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862856015372317,
+      "loss": 0.8293,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019861825305031574,
+      "loss": 0.8308,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860790762951447,
+      "loss": 0.7107,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985975238953389,
+      "loss": 0.895,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985871018518236,
+      "loss": 0.3485,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857664150301795,
+      "loss": 0.8154,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985661428529863,
+      "loss": 0.9031,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019855560590580778,
+      "loss": 0.8086,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854503066557645,
+      "loss": 0.8367,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985344171364012,
+      "loss": 0.886,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852376532240594,
+      "loss": 0.7886,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851307522772923,
+      "loss": 0.812,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019850234685652476,
+      "loss": 0.8408,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849158021296081,
+      "loss": 0.8982,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848077530122083,
+      "loss": 0.8354,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846993212550287,
+      "loss": 0.8015,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845905069002003,
+      "loss": 0.8503,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984481309990002,
+      "loss": 0.8147,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843717305668613,
+      "loss": 0.8687,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019842617686733545,
+      "loss": 0.8428,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841514243522064,
+      "loss": 0.8196,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840406976462908,
+      "loss": 0.8716,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839295885986296,
+      "loss": 0.3313,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838180972523936,
+      "loss": 0.8271,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837062236509014,
+      "loss": 0.844,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019835939678376213,
+      "loss": 0.8198,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019834813298561684,
+      "loss": 0.8159,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833683097503086,
+      "loss": 0.8691,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983254907563955,
+      "loss": 0.814,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831411233411678,
+      "loss": 0.8135,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830269571261583,
+      "loss": 0.8582,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019829124089632845,
+      "loss": 0.8267,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827974788970534,
+      "loss": 0.8064,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826821669721202,
+      "loss": 0.8193,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019825664732332884,
+      "loss": 0.8455,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019824503977255097,
+      "loss": 0.8206,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982333940493885,
+      "loss": 0.8699,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822171015836625,
+      "loss": 0.866,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982099881040239,
+      "loss": 0.8076,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019819822789091598,
+      "loss": 0.4022,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818642952361187,
+      "loss": 0.7856,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019817459300669566,
+      "loss": 0.7871,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816271834476642,
+      "loss": 0.8623,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981508055424379,
+      "loss": 0.7986,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813885460433879,
+      "loss": 0.8398,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019812686553511246,
+      "loss": 0.7966,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019811483833941728,
+      "loss": 0.8074,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810277302192622,
+      "loss": 0.8159,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019809066958732726,
+      "loss": 0.8477,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807852804032305,
+      "loss": 0.7664,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980663483856311,
+      "loss": 0.877,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019805413062798377,
+      "loss": 0.8093,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980418747721281,
+      "loss": 0.8564,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802958082282607,
+      "loss": 0.8726,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801724878485438,
+      "loss": 0.8462,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980048786630046,
+      "loss": 0.8796,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799247046208297,
+      "loss": 0.8896,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019798002418691067,
+      "loss": 0.7932,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796753984232358,
+      "loss": 0.811,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979550174331724,
+      "loss": 0.8425,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979424569643226,
+      "loss": 0.8059,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979298584406545,
+      "loss": 0.8315,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019791722186706317,
+      "loss": 0.8274,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979045472484584,
+      "loss": 0.7993,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019789183458976484,
+      "loss": 0.8259,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019787908389592196,
+      "loss": 0.389,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019786629517188384,
+      "loss": 0.8013,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785346842261957,
+      "loss": 0.8123,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019784060365311276,
+      "loss": 0.7964,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197827700868362,
+      "loss": 0.8677,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781476007338058,
+      "loss": 0.3603,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019780178127319647,
+      "loss": 0.8845,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019778876447285263,
+      "loss": 0.8372,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977757096774065,
+      "loss": 0.8372,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776261689193048,
+      "loss": 0.7759,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774948612151166,
+      "loss": 0.8667,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019773631737125192,
+      "loss": 0.8606,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772311064626788,
+      "loss": 0.8003,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019770986595169096,
+      "loss": 0.8518,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019769658329266718,
+      "loss": 0.7952,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976832626743575,
+      "loss": 0.7908,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019766990410193753,
+      "loss": 0.7939,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019765650758059766,
+      "loss": 0.8083,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764307311554298,
+      "loss": 0.3959,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019762960071199333,
+      "loss": 0.803,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976160903751834,
+      "loss": 0.8289,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019760254211036244,
+      "loss": 0.8191,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975889559227946,
+      "loss": 0.854,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757533181775865,
+      "loss": 0.3842,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019756166980054813,
+      "loss": 0.8228,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754796987647133,
+      "loss": 0.832,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019753423205085127,
+      "loss": 0.8259,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019752045632902567,
+      "loss": 0.7903,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019750664271634698,
+      "loss": 0.8796,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749279121818235,
+      "loss": 0.8511,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019747890183991376,
+      "loss": 0.8723,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019746497458693774,
+      "loss": 0.8308,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745100946466564,
+      "loss": 0.8076,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743700647852354,
+      "loss": 0.8005,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019742296563395216,
+      "loss": 0.843,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197408886936407,
+      "loss": 0.8574,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973947703913582,
+      "loss": 0.8521,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019738061600429064,
+      "loss": 0.8428,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736642378070392,
+      "loss": 0.8271,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735219372611233,
+      "loss": 0.802,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019733792584604486,
+      "loss": 0.7949,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019732362014604515,
+      "loss": 0.8264,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973092766316716,
+      "loss": 0.8245,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972948953084973,
+      "loss": 0.8547,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019728047618210995,
+      "loss": 0.8345,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726601925811204,
+      "loss": 0.8118,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019725152454212066,
+      "loss": 0.9148,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019723699203976766,
+      "loss": 0.7913,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019722242175669954,
+      "loss": 0.8582,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019720781369857746,
+      "loss": 0.8367,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019719316787107728,
+      "loss": 0.9014,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717848427988952,
+      "loss": 0.3704,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971637629307194,
+      "loss": 0.8691,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019714900382928675,
+      "loss": 0.8269,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019713420698132614,
+      "loss": 0.8157,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019711937239258679,
+      "loss": 0.8044,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019710450006883256,
+      "loss": 0.8062,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019708959001584194,
+      "loss": 0.3631,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019707464223940818,
+      "loss": 0.36,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970596567453391,
+      "loss": 0.7896,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970446335394572,
+      "loss": 0.8269,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702957262759965,
+      "loss": 0.8914,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701447401561826,
+      "loss": 0.8511,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019699933770937945,
+      "loss": 0.7839,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019698416371476433,
+      "loss": 0.8218,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969689520376687,
+      "loss": 0.7915,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019695370268400287,
+      "loss": 0.8159,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019693841565969188,
+      "loss": 0.9004,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019692309097067546,
+      "loss": 0.8452,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969077286229078,
+      "loss": 0.8394,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968923286223579,
+      "loss": 0.8369,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019687689097500928,
+      "loss": 0.7834,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019686141568686015,
+      "loss": 0.8142,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684590276392328,
+      "loss": 0.8362,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019683035221222618,
+      "loss": 0.8162,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968147640378108,
+      "loss": 0.774,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019679913824673386,
+      "loss": 0.8328,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019678347484506669,
+      "loss": 0.7996,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019676777383889507,
+      "loss": 0.876,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019675203523431966,
+      "loss": 0.7751,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673625903745545,
+      "loss": 0.8373,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019672044525443223,
+      "loss": 0.8416,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019670459389139433,
+      "loss": 0.8594,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019668870495450066,
+      "loss": 0.8596,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667277844992475,
+      "loss": 0.7932,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665681438385473,
+      "loss": 0.8674,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019664081276249334,
+      "loss": 0.8472,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019662477359205785,
+      "loss": 0.8162,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019660869687878019,
+      "loss": 0.8137,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019659258262890683,
+      "loss": 0.7986,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019657643084869888,
+      "loss": 0.801,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019656024154443195,
+      "loss": 0.8167,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965440147223963,
+      "loss": 0.7881,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019652775038889674,
+      "loss": 0.8042,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019651144855025265,
+      "loss": 0.8164,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019649510921279802,
+      "loss": 0.8547,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964787323828813,
+      "loss": 0.7808,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019646231806686566,
+      "loss": 0.7925,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644586627112876,
+      "loss": 0.7959,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964293770020628,
+      "loss": 0.7825,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964128502660745,
+      "loss": 0.8823,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019639628606958533,
+      "loss": 0.8159,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963796844190311,
+      "loss": 0.8477,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963630453208623,
+      "loss": 0.8013,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963463687815439,
+      "loss": 0.8262,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019632965480755548,
+      "loss": 0.8103,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019631290340539108,
+      "loss": 0.8296,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019629611458155937,
+      "loss": 0.8096,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962792883425835,
+      "loss": 0.8306,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962624246950012,
+      "loss": 0.79,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019624552364536473,
+      "loss": 0.8352,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622858520024086,
+      "loss": 0.8396,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019621160936621085,
+      "loss": 0.7793,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961945961498706,
+      "loss": 0.8323,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019617754555783043,
+      "loss": 0.8372,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019616045759671523,
+      "loss": 0.8267,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961433322731644,
+      "loss": 0.8564,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961261695938319,
+      "loss": 0.8796,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019610896956538608,
+      "loss": 0.8516,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019609173219450998,
+      "loss": 0.855,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019607445748790094,
+      "loss": 0.8604,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019605714545227102,
+      "loss": 0.8135,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019603979609434665,
+      "loss": 0.8451,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019602240942086878,
+      "loss": 0.3881,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960049854385929,
+      "loss": 0.8186,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959875241542889,
+      "loss": 0.8262,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019597002557474134,
+      "loss": 0.7959,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019595248970674908,
+      "loss": 0.8396,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959349165571256,
+      "loss": 0.8403,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959173061326988,
+      "loss": 0.8447,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958996584403111,
+      "loss": 0.3514,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958819734868193,
+      "loss": 0.8,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019586425127909486,
+      "loss": 0.8235,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019584649182402357,
+      "loss": 0.3507,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019582869512850575,
+      "loss": 0.8279,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019581086119945616,
+      "loss": 0.8257,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019579299004380402,
+      "loss": 0.8638,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019577508166849304,
+      "loss": 0.8635,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019575713608048145,
+      "loss": 0.8264,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957391532867418,
+      "loss": 0.8916,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019572113329426123,
+      "loss": 0.8308,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019570307611004124,
+      "loss": 0.3672,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019568498174109782,
+      "loss": 0.8276,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956668501944614,
+      "loss": 0.7681,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956486814771769,
+      "loss": 0.8413,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019563047559630357,
+      "loss": 0.8293,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956122325589152,
+      "loss": 0.8259,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955939523721,
+      "loss": 0.8572,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019557563504296062,
+      "loss": 0.3726,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955572805786141,
+      "loss": 0.8354,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955388889861919,
+      "loss": 0.3837,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019552046027284,
+      "loss": 0.8113,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955019944457187,
+      "loss": 0.8523,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019548349151200274,
+      "loss": 0.7986,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019546495147888132,
+      "loss": 0.3799,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544637435355808,
+      "loss": 0.8108,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195427760143251,
+      "loss": 0.7947,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019540910885519242,
+      "loss": 0.8494,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019539042049662925,
+      "loss": 0.8091,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953716950748227,
+      "loss": 0.8047,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019535293259704837,
+      "loss": 0.8447,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019533413307059626,
+      "loss": 0.8455,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019531529650277085,
+      "loss": 0.8218,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952964229008909,
+      "loss": 0.8403,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527751227228963,
+      "loss": 0.8391,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952585646243146,
+      "loss": 0.8379,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019523957996432786,
+      "loss": 0.7913,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019522055829970564,
+      "loss": 0.8235,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520149963783874,
+      "loss": 0.8528,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019518240398613227,
+      "loss": 0.8325,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019516327135200566,
+      "loss": 0.7905,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514410174289277,
+      "loss": 0.7429,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951248951662418,
+      "loss": 0.8047,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019510565162951537,
+      "loss": 0.3732,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019508637114019038,
+      "loss": 0.8364,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950670537057581,
+      "loss": 0.8188,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019504769933372424,
+      "loss": 0.8384,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019502830803160874,
+      "loss": 0.7727,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019500887980694595,
+      "loss": 0.8357,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949894146672846,
+      "loss": 0.8396,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019496991262018772,
+      "loss": 0.8098,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019495037367323262,
+      "loss": 0.7871,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019493079783401113,
+      "loss": 0.8452,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019491118511012922,
+      "loss": 0.8669,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019489153550920728,
+      "loss": 0.8477,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019487184903887997,
+      "loss": 0.8604,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019485212570679642,
+      "loss": 0.3352,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019483236552061994,
+      "loss": 0.3651,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948125684880282,
+      "loss": 0.8083,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947927346167132,
+      "loss": 0.8411,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947728639143812,
+      "loss": 0.7991,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947529563887529,
+      "loss": 0.3439,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019473301204756314,
+      "loss": 0.824,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947130308985612,
+      "loss": 0.8154,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946930129495106,
+      "loss": 0.8467,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019467295820818913,
+      "loss": 0.8445,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019465286668238893,
+      "loss": 0.8022,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019463273837991643,
+      "loss": 0.842,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019461257330859232,
+      "loss": 0.8149,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945923714762516,
+      "loss": 0.3896,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019457213289074355,
+      "loss": 0.7844,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945518575599317,
+      "loss": 0.8376,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019453154549169387,
+      "loss": 0.8096,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019451119669392219,
+      "loss": 0.856,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019449081117452302,
+      "loss": 0.8062,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019447038894141705,
+      "loss": 0.8179,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019444993000253914,
+      "loss": 0.7832,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019442943436583844,
+      "loss": 0.8511,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019440890203927845,
+      "loss": 0.8789,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019438833303083678,
+      "loss": 0.823,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943677273485054,
+      "loss": 0.8005,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019434708500029049,
+      "loss": 0.7778,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943264059942125,
+      "loss": 0.8325,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019430569033830605,
+      "loss": 0.9028,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001942849380406201,
+      "loss": 0.8088,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019426414910921787,
+      "loss": 0.811,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942433235521766,
+      "loss": 0.873,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000194222461377588,
+      "loss": 0.8528,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019420156259355791,
+      "loss": 0.8479,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019418062720820637,
+      "loss": 0.8196,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019415965522966765,
+      "loss": 0.823,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019413864666609034,
+      "loss": 0.7766,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019411760152563708,
+      "loss": 0.8066,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019409651981648485,
+      "loss": 0.8088,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019407540154682472,
+      "loss": 0.8723,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019405424672486218,
+      "loss": 0.8313,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019403305535881663,
+      "loss": 0.8198,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940118274569219,
+      "loss": 0.7905,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019399056302742596,
+      "loss": 0.7893,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019396926207859084,
+      "loss": 0.7908,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019394792461869295,
+      "loss": 0.8208,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392655065602281,
+      "loss": 0.8425,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019390514019888507,
+      "loss": 0.7874,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938836932555986,
+      "loss": 0.8394,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019386220983449653,
+      "loss": 0.8357,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193840689943926,
+      "loss": 0.8267,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019381913359224842,
+      "loss": 0.782,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019379754078783937,
+      "loss": 0.8367,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019377591153908858,
+      "loss": 0.7808,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019375424585439994,
+      "loss": 0.8135,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373254374219144,
+      "loss": 0.8079,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019371080521089534,
+      "loss": 0.3677,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019368903026895795,
+      "loss": 0.8286,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019366721892483978,
+      "loss": 0.854,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019364537118701542,
+      "loss": 0.8206,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019362348706397373,
+      "loss": 0.8125,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019360156656421756,
+      "loss": 0.7869,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019357960969626397,
+      "loss": 0.8289,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019355761646864416,
+      "loss": 0.781,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935355868899034,
+      "loss": 0.7734,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935135209686012,
+      "loss": 0.8142,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193491418713311,
+      "loss": 0.8303,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001934692801326205,
+      "loss": 0.8181,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344710523513156,
+      "loss": 0.7881,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019342489402945998,
+      "loss": 0.8301,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019340264652423583,
+      "loss": 0.8584,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019338036272810314,
+      "loss": 0.7783,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019335804264972018,
+      "loss": 0.8042,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019333568629775923,
+      "loss": 0.825,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019331329368090666,
+      "loss": 0.7544,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019329086480786298,
+      "loss": 0.7711,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019326839968734279,
+      "loss": 0.7666,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019324589832807467,
+      "loss": 0.399,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019322336073880142,
+      "loss": 0.8206,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019320078692827987,
+      "loss": 0.7986,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019317817690528083,
+      "loss": 0.825,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019315553067858935,
+      "loss": 0.8186,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019313284825700437,
+      "loss": 0.7915,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000193110129649339,
+      "loss": 0.8201,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019308737486442045,
+      "loss": 0.8239,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019306458391108984,
+      "loss": 0.821,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019304175679820247,
+      "loss": 0.8601,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019301889353462762,
+      "loss": 0.8416,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019299599412924868,
+      "loss": 0.7766,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019297305859096304,
+      "loss": 0.8293,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019295008692868208,
+      "loss": 0.8298,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019292707915133134,
+      "loss": 0.8115,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019290403526785025,
+      "loss": 0.8196,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019288095528719243,
+      "loss": 0.8083,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019285783921832537,
+      "loss": 0.7888,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019283468707023067,
+      "loss": 0.8164,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928114988519039,
+      "loss": 0.3577,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927882745723547,
+      "loss": 0.8962,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927650142406067,
+      "loss": 0.7971,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927417178656975,
+      "loss": 0.791,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019271838545667876,
+      "loss": 0.8237,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926950170226161,
+      "loss": 0.7827,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019267161257258914,
+      "loss": 0.8418,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019264817211569153,
+      "loss": 0.8167,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926246956610309,
+      "loss": 0.864,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926011832177288,
+      "loss": 0.7715,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019257763479492085,
+      "loss": 0.78,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019255405040175663,
+      "loss": 0.8379,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019253043004739968,
+      "loss": 0.7878,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925067737410275,
+      "loss": 0.8079,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019248308149183153,
+      "loss": 0.8811,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019245935330901732,
+      "loss": 0.8079,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019243558920180418,
+      "loss": 0.7546,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019241178917942553,
+      "loss": 0.7778,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 0.7786,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923640814261749,
+      "loss": 0.8923,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019234017371383945,
+      "loss": 0.8284,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019231623012341145,
+      "loss": 0.8018,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192292250664194,
+      "loss": 0.8159,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019226823534550418,
+      "loss": 0.8235,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019224418417667293,
+      "loss": 0.8416,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922200971670452,
+      "loss": 0.8164,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019219597432597975,
+      "loss": 0.3802,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921718156628494,
+      "loss": 0.8159,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019214762118704076,
+      "loss": 0.8179,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921233909079545,
+      "loss": 0.8098,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192099124835005,
+      "loss": 0.8628,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019207482297762076,
+      "loss": 0.8237,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019205048534524406,
+      "loss": 0.803,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019202611194733108,
+      "loss": 0.8284,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001920017027933519,
+      "loss": 0.7795,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019197725789279064,
+      "loss": 0.7893,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001919527772551451,
+      "loss": 0.8623,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019192826088992702,
+      "loss": 0.8259,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019190370880666207,
+      "loss": 0.79,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019187912101488984,
+      "loss": 0.7935,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019185449752416367,
+      "loss": 0.8217,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918298383440509,
+      "loss": 0.8589,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019180514348413259,
+      "loss": 0.8406,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019178041295400382,
+      "loss": 0.8374,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019175564676327339,
+      "loss": 0.824,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019173084492156407,
+      "loss": 0.8359,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917060074385124,
+      "loss": 0.8474,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019168113432376885,
+      "loss": 0.8157,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019165622558699763,
+      "loss": 0.8003,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019163128123787687,
+      "loss": 0.8279,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916063012860985,
+      "loss": 0.8088,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915812857413683,
+      "loss": 0.8081,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019155623461340594,
+      "loss": 0.8022,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019153114791194473,
+      "loss": 0.7776,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019150602564673198,
+      "loss": 0.8469,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019148086782752878,
+      "loss": 0.3883,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019145567446411,
+      "loss": 0.7594,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019143044556626436,
+      "loss": 0.8049,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019140518114379434,
+      "loss": 0.7573,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913798812065162,
+      "loss": 0.8523,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913545457642601,
+      "loss": 0.8433,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019132917482686992,
+      "loss": 0.8525,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019130376840420337,
+      "loss": 0.8281,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019127832650613189,
+      "loss": 0.803,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912528491425408,
+      "loss": 0.8374,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019122733632332908,
+      "loss": 0.8123,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019120178805840955,
+      "loss": 0.7874,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019117620435770886,
+      "loss": 0.8218,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019115058523116733,
+      "loss": 0.865,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019112493068873906,
+      "loss": 0.804,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019109924074039197,
+      "loss": 0.7761,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910735153961077,
+      "loss": 0.835,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019104775466588161,
+      "loss": 0.8079,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019102195855972287,
+      "loss": 0.8291,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019099612708765434,
+      "loss": 0.7585,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019097026025971268,
+      "loss": 0.8359,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019094435808594823,
+      "loss": 0.8237,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019091842057642508,
+      "loss": 0.7898,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908924477412211,
+      "loss": 0.7932,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908664395904278,
+      "loss": 0.8279,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908403961341505,
+      "loss": 0.8567,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019081431738250814,
+      "loss": 0.7593,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019078820334563347,
+      "loss": 0.7896,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019076205403367285,
+      "loss": 0.771,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907358694567865,
+      "loss": 0.3859,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070964962514814,
+      "loss": 0.8162,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019068339454894536,
+      "loss": 0.8521,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019065710423837935,
+      "loss": 0.7737,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000190630778703665,
+      "loss": 0.8225,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019060441795503094,
+      "loss": 0.7537,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019057802200271942,
+      "loss": 0.3844,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055159085698637,
+      "loss": 0.8357,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905251245281015,
+      "loss": 0.8325,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000190498623026348,
+      "loss": 0.8257,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019047208636202288,
+      "loss": 0.7927,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019044551454543683,
+      "loss": 0.7908,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019041890758691402,
+      "loss": 0.7521,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019039226549679244,
+      "loss": 0.7991,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019036558828542367,
+      "loss": 0.7903,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019033887596317298,
+      "loss": 0.832,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019031212854041918,
+      "loss": 0.8433,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019028534602755483,
+      "loss": 0.865,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019025852843498607,
+      "loss": 0.8459,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019023167577313264,
+      "loss": 0.803,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019020478805242803,
+      "loss": 0.8162,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901778652833192,
+      "loss": 0.8345,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901509074762668,
+      "loss": 0.8232,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019012391464174507,
+      "loss": 0.7603,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900968867902419,
+      "loss": 0.8071,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019006982393225878,
+      "loss": 0.7595,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019004272607831078,
+      "loss": 0.8086,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019001559323892652,
+      "loss": 0.7703,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018998842542464832,
+      "loss": 0.8064,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018996122264603203,
+      "loss": 0.8064,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018993398491364702,
+      "loss": 0.7832,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001899067122380764,
+      "loss": 0.774,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898794046299167,
+      "loss": 0.8293,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018985206209977813,
+      "loss": 0.821,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018982468465828442,
+      "loss": 0.8513,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897972723160729,
+      "loss": 0.8435,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018976982508379434,
+      "loss": 0.3811,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018974234297211326,
+      "loss": 0.8049,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897148259917076,
+      "loss": 0.7866,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018968727415326884,
+      "loss": 0.3571,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896596874675021,
+      "loss": 0.8547,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018963206594512592,
+      "loss": 0.8379,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018960440959687254,
+      "loss": 0.8245,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895767184334875,
+      "loss": 0.813,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895489924657301,
+      "loss": 0.8271,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018952123170437303,
+      "loss": 0.8647,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018949343616020252,
+      "loss": 0.8054,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894656058440183,
+      "loss": 0.8284,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894377407666337,
+      "loss": 0.7688,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940984093887542,
+      "loss": 0.8174,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018938190637158377,
+      "loss": 0.8105,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018935393707561251,
+      "loss": 0.7925,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893259330618289,
+      "loss": 0.8003,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018929789434111373,
+      "loss": 0.8479,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018926982092436117,
+      "loss": 0.8291,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018924171282247898,
+      "loss": 0.7737,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018921357004638835,
+      "loss": 0.7651,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018918539260702396,
+      "loss": 0.8313,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891571805153339,
+      "loss": 0.7932,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018912893378227985,
+      "loss": 0.8237,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891006524188368,
+      "loss": 0.8008,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890723364359933,
+      "loss": 0.7407,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890439858447513,
+      "loss": 0.8638,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890156006561262,
+      "loss": 0.8167,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889871808811469,
+      "loss": 0.8218,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018895872653085563,
+      "loss": 0.7529,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018893023761630814,
+      "loss": 0.8115,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018890171414857364,
+      "loss": 0.8406,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888731561387347,
+      "loss": 0.7766,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018884456359788724,
+      "loss": 0.7411,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018881593653714075,
+      "loss": 0.7727,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018878727496761809,
+      "loss": 0.8149,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018875857890045543,
+      "loss": 0.7979,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018872984834680242,
+      "loss": 0.8533,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018870108331782217,
+      "loss": 0.3506,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886722838246911,
+      "loss": 0.8374,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018864344987859898,
+      "loss": 0.8152,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886145814907491,
+      "loss": 0.7954,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188585678672358,
+      "loss": 0.824,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018855674143465566,
+      "loss": 0.8228,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018852776978888551,
+      "loss": 0.7869,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884987637463042,
+      "loss": 0.7988,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884697233181818,
+      "loss": 0.8037,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884406485158018,
+      "loss": 0.8201,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018841153935046098,
+      "loss": 0.8552,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018838239583346948,
+      "loss": 0.3953,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018835321797615082,
+      "loss": 0.8567,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018832400578984183,
+      "loss": 0.8337,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018829475928589271,
+      "loss": 0.7888,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018826547847566693,
+      "loss": 0.824,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018823616337054134,
+      "loss": 0.8643,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018820681398190616,
+      "loss": 0.874,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881774303211648,
+      "loss": 0.7651,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018814801239973414,
+      "loss": 0.8481,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018811856022904423,
+      "loss": 0.8083,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018808907382053858,
+      "loss": 0.7466,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880595531856738,
+      "loss": 0.7664,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018802999833592002,
+      "loss": 0.8079,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018800040928276048,
+      "loss": 0.8315,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018797078603769184,
+      "loss": 0.7881,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000187941128612224,
+      "loss": 0.7703,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018791143701788005,
+      "loss": 0.3621,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018788171126619653,
+      "loss": 0.7974,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018785195136872317,
+      "loss": 0.8171,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018782215733702286,
+      "loss": 0.7571,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018779232918267195,
+      "loss": 0.802,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018776246691725988,
+      "loss": 0.413,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018773257055238943,
+      "loss": 0.7795,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018770264009967667,
+      "loss": 0.3813,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001876726755707508,
+      "loss": 0.8174,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876426769772543,
+      "loss": 0.8337,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018761264433084294,
+      "loss": 0.8535,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018758257764318567,
+      "loss": 0.8152,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018755247692596466,
+      "loss": 0.7888,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018752234219087538,
+      "loss": 0.7419,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874921734496264,
+      "loss": 0.8059,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018746197071393958,
+      "loss": 0.7959,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018743173399555,
+      "loss": 0.825,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018740146330620584,
+      "loss": 0.7847,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018737115865766863,
+      "loss": 0.7942,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018734082006171299,
+      "loss": 0.834,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018731044753012675,
+      "loss": 0.3588,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018728004107471092,
+      "loss": 0.8616,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018724960070727972,
+      "loss": 0.8562,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018721912643966055,
+      "loss": 0.8015,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871886182836939,
+      "loss": 0.8103,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018715807625123358,
+      "loss": 0.8179,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871275003541464,
+      "loss": 0.8303,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018709689060431242,
+      "loss": 0.7842,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018706624701362483,
+      "loss": 0.7905,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018703556959398998,
+      "loss": 0.7493,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018700485835732735,
+      "loss": 0.814,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018697411331556956,
+      "loss": 0.8066,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018694333448066237,
+      "loss": 0.3761,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018691252186456465,
+      "loss": 0.8386,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018688167547924845,
+      "loss": 0.8079,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001868507953366989,
+      "loss": 0.822,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018681988144891425,
+      "loss": 0.8633,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867889338279058,
+      "loss": 0.7771,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018675795248569813,
+      "loss": 0.8291,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018672693743432874,
+      "loss": 0.7854,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001866958886858483,
+      "loss": 0.3455,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018666480625232062,
+      "loss": 0.8059,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001866336901458225,
+      "loss": 0.7615,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.8196,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018657135696228779,
+      "loss": 0.8335,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865401399094703,
+      "loss": 0.7683,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018650888923212058,
+      "loss": 0.8035,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018647760494238083,
+      "loss": 0.783,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018644628705240636,
+      "loss": 0.8206,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001864149355743655,
+      "loss": 0.8687,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018638355052043957,
+      "loss": 0.3944,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001863521319028231,
+      "loss": 0.7822,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018632067973372348,
+      "loss": 0.7737,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862891940253613,
+      "loss": 0.8169,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018625767478997002,
+      "loss": 0.8657,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018622612203979627,
+      "loss": 0.365,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018619453578709954,
+      "loss": 0.8362,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018616291604415258,
+      "loss": 0.8088,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018613126282324092,
+      "loss": 0.7834,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018609957613666315,
+      "loss": 0.76,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018606785599673098,
+      "loss": 0.8276,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000186036102415769,
+      "loss": 0.8323,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001860043154061148,
+      "loss": 0.8027,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018597249498011903,
+      "loss": 0.8147,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859406411501453,
+      "loss": 0.8606,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018590875392857008,
+      "loss": 0.8413,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018587683332778302,
+      "loss": 0.7661,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018584487936018661,
+      "loss": 0.3527,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858128920381963,
+      "loss": 0.7688,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018578087137424056,
+      "loss": 0.7712,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018574881738076074,
+      "loss": 0.7722,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018571673007021123,
+      "loss": 0.8162,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856846094550593,
+      "loss": 0.833,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018565245554778517,
+      "loss": 0.8547,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018562026836088197,
+      "loss": 0.7937,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018558804790685588,
+      "loss": 0.8035,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018555579419822583,
+      "loss": 0.8259,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018552350724752382,
+      "loss": 0.3632,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018549118706729468,
+      "loss": 0.8062,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018545883367009617,
+      "loss": 0.7876,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018542644706849898,
+      "loss": 0.8052,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853940272750867,
+      "loss": 0.7917,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853615743024558,
+      "loss": 0.783,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018532908816321558,
+      "loss": 0.8123,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018529656886998836,
+      "loss": 0.832,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018526401643540922,
+      "loss": 0.8274,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018523143087212623,
+      "loss": 0.8267,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851988121928002,
+      "loss": 0.8396,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018516616041010494,
+      "loss": 0.8311,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018513347553672703,
+      "loss": 0.8181,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018510075758536591,
+      "loss": 0.7649,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018506800656873398,
+      "loss": 0.8257,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850352224995563,
+      "loss": 0.7917,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850024053905709,
+      "loss": 0.8298,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018496955525452874,
+      "loss": 0.7761,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018493667210419336,
+      "loss": 0.3566,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018490375595234132,
+      "loss": 0.8174,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018487080681176194,
+      "loss": 0.8398,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018483782469525738,
+      "loss": 0.8455,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848048096156426,
+      "loss": 0.8259,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018477176158574534,
+      "loss": 0.8491,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847386806184062,
+      "loss": 0.8113,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847055667264785,
+      "loss": 0.781,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018467241992282843,
+      "loss": 0.38,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018463924022033492,
+      "loss": 0.8164,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018460602763188972,
+      "loss": 0.76,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018457278217039736,
+      "loss": 0.8657,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018453950384877504,
+      "loss": 0.7854,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018450619267995283,
+      "loss": 0.76,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001844728486768736,
+      "loss": 0.8013,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018443947185249287,
+      "loss": 0.7686,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018440606221977894,
+      "loss": 0.7491,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001843726197917129,
+      "loss": 0.3866,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001843391445812886,
+      "loss": 0.8455,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001843056366015125,
+      "loss": 0.8225,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842720958654039,
+      "loss": 0.8389,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018423852238599487,
+      "loss": 0.7783,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842049161763301,
+      "loss": 0.8237,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018417127724946702,
+      "loss": 0.3928,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018413760561847577,
+      "loss": 0.8315,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018410390129643925,
+      "loss": 0.7412,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018407016429645303,
+      "loss": 0.8054,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018403639463162534,
+      "loss": 0.7776,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018400259231507717,
+      "loss": 0.7827,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839687573599421,
+      "loss": 0.825,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018393488977936652,
+      "loss": 0.8308,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018390098958650941,
+      "loss": 0.7893,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018386705679454242,
+      "loss": 0.7961,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001838330914166499,
+      "loss": 0.8142,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018379909346602881,
+      "loss": 0.7991,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018376506295588886,
+      "loss": 0.8086,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018373099989945236,
+      "loss": 0.8713,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836969043099542,
+      "loss": 0.8086,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018366277620064197,
+      "loss": 0.804,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018362861558477594,
+      "loss": 0.8059,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018359442247562897,
+      "loss": 0.7666,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018356019688648648,
+      "loss": 0.8699,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018352593883064664,
+      "loss": 0.7769,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018349164832142013,
+      "loss": 0.7871,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018345732537213027,
+      "loss": 0.7905,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018342296999611298,
+      "loss": 0.8306,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018338858220671682,
+      "loss": 0.8174,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833541620173029,
+      "loss": 0.7771,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001833197094412449,
+      "loss": 0.8225,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018328522449192916,
+      "loss": 0.8115,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832507071827545,
+      "loss": 0.8865,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832161575271324,
+      "loss": 0.8093,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831815755384869,
+      "loss": 0.79,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018314696123025454,
+      "loss": 0.7622,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018311231461588446,
+      "loss": 0.7932,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018307763570883832,
+      "loss": 0.8218,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830429245225904,
+      "loss": 0.7983,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018300818107062743,
+      "loss": 0.8293,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018297340536644875,
+      "loss": 0.8169,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018293859742356621,
+      "loss": 0.7769,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018290375725550417,
+      "loss": 0.7715,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018286888487579952,
+      "loss": 0.7866,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018283398029800166,
+      "loss": 0.7942,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018279904353567253,
+      "loss": 0.8079,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827640746023865,
+      "loss": 0.8218,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018272907351173052,
+      "loss": 0.8003,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018269404027730402,
+      "loss": 0.8162,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018265897491271887,
+      "loss": 0.8384,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001826238774315995,
+      "loss": 0.8081,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018258874784758276,
+      "loss": 0.8704,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018255358617431797,
+      "loss": 0.7832,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018251839242546696,
+      "loss": 0.8337,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000182483166614704,
+      "loss": 0.8022,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824479087557158,
+      "loss": 0.8279,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018241261886220154,
+      "loss": 0.8054,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018237729694787293,
+      "loss": 0.8503,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018234194302645394,
+      "loss": 0.8132,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001823065571116811,
+      "loss": 0.8411,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018227113921730334,
+      "loss": 0.8333,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018223568935708207,
+      "loss": 0.7993,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018220020754479102,
+      "loss": 0.7949,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821646937942164,
+      "loss": 0.7952,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821291481191568,
+      "loss": 0.7998,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018209357053342325,
+      "loss": 0.3834,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018205796105083915,
+      "loss": 0.7565,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018202231968524033,
+      "loss": 0.7864,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018198664645047497,
+      "loss": 0.8013,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001819509413604036,
+      "loss": 0.7942,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001819152044288992,
+      "loss": 0.7808,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818794356698471,
+      "loss": 0.8259,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018184363509714496,
+      "loss": 0.7678,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018180780272470282,
+      "loss": 0.8516,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018177193856644316,
+      "loss": 0.7866,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018173604263630063,
+      "loss": 0.7681,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001817001149482224,
+      "loss": 0.7725,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018166415551616792,
+      "loss": 0.8374,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001816281643541089,
+      "loss": 0.8047,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018159214147602946,
+      "loss": 0.7981,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018155608689592604,
+      "loss": 0.7275,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018152000062780736,
+      "loss": 0.7913,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018148388268569453,
+      "loss": 0.3888,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814477330836208,
+      "loss": 0.8625,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018141155183563193,
+      "loss": 0.8127,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018137533895578583,
+      "loss": 0.7764,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018133909445815279,
+      "loss": 0.8384,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018130281835681528,
+      "loss": 0.8413,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018126651066586818,
+      "loss": 0.8347,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812301713994185,
+      "loss": 0.7639,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018119380057158568,
+      "loss": 0.7786,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018115739819650123,
+      "loss": 0.8269,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018112096428830911,
+      "loss": 0.8665,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018108449886116543,
+      "loss": 0.8533,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018104800192923854,
+      "loss": 0.7544,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018101147350670907,
+      "loss": 0.7913,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018097491360776985,
+      "loss": 0.7922,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018093832224662596,
+      "loss": 0.3649,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 0.8491,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808650451946057,
+      "loss": 0.7622,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018082835953220054,
+      "loss": 0.3925,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807916424645332,
+      "loss": 0.801,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018075489400586991,
+      "loss": 0.8308,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807181141704889,
+      "loss": 0.856,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001806813029726808,
+      "loss": 0.8276,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018064446042674828,
+      "loss": 0.822,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018060758654700622,
+      "loss": 0.8208,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018057068134778173,
+      "loss": 0.7781,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018053374484341403,
+      "loss": 0.7944,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018049677704825453,
+      "loss": 0.843,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018045977797666684,
+      "loss": 0.8237,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018042274764302663,
+      "loss": 0.7759,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018038568606172173,
+      "loss": 0.8408,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001803485932471522,
+      "loss": 0.8064,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018031146921373018,
+      "loss": 0.3929,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018027431397587992,
+      "loss": 0.8096,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001802371275480378,
+      "loss": 0.7771,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018019990994465234,
+      "loss": 0.752,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801626611801842,
+      "loss": 0.3756,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018012538126910608,
+      "loss": 0.3621,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018008807022590283,
+      "loss": 0.7861,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001800507280650713,
+      "loss": 0.7876,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018001335480112064,
+      "loss": 0.8513,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799759504485719,
+      "loss": 0.8062,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017993851502195822,
+      "loss": 0.8167,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017990104853582493,
+      "loss": 0.825,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017986355100472928,
+      "loss": 0.8845,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017982602244324074,
+      "loss": 0.7676,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797884628659407,
+      "loss": 0.3596,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017975087228742272,
+      "loss": 0.3606,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017971325072229226,
+      "loss": 0.7734,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017967559818516694,
+      "loss": 0.7834,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017963791469067638,
+      "loss": 0.7222,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001796002002534622,
+      "loss": 0.7725,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017956245488817812,
+      "loss": 0.801,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017952467860948973,
+      "loss": 0.791,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017948687143207483,
+      "loss": 0.8516,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017944903337062303,
+      "loss": 0.7886,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017941116443983613,
+      "loss": 0.3983,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017937326465442772,
+      "loss": 0.8054,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017933533402912354,
+      "loss": 0.8125,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792973725786612,
+      "loss": 0.7957,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017925938031779044,
+      "loss": 0.7649,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017922135726127275,
+      "loss": 0.8071,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791833034238818,
+      "loss": 0.7888,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791452188204031,
+      "loss": 0.8555,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017910710346563416,
+      "loss": 0.8252,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017906895737438436,
+      "loss": 0.7656,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017903078056147513,
+      "loss": 0.7743,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017899257304173976,
+      "loss": 0.8298,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017895433483002354,
+      "loss": 0.7957,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017891606594118363,
+      "loss": 0.8057,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017887776639008914,
+      "loss": 0.7712,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000178839436191621,
+      "loss": 0.8184,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017880107536067218,
+      "loss": 0.8572,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017876268391214754,
+      "loss": 0.7976,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017872426186096374,
+      "loss": 0.7954,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786858092220494,
+      "loss": 0.7783,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178647326010345,
+      "loss": 0.7576,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786088122408029,
+      "loss": 0.8208,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017857026792838737,
+      "loss": 0.7661,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017853169308807448,
+      "loss": 0.811,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017849308773485226,
+      "loss": 0.8298,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017845445188372047,
+      "loss": 0.804,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001784157855496908,
+      "loss": 0.8103,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017837708874778683,
+      "loss": 0.835,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783383614930438,
+      "loss": 0.7983,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178299603800509,
+      "loss": 0.8181,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782608156852414,
+      "loss": 0.8005,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017822199716231184,
+      "loss": 0.8369,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178183148246803,
+      "loss": 0.7615,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781442689538093,
+      "loss": 0.8196,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017810535929843702,
+      "loss": 0.801,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017806641929580419,
+      "loss": 0.8169,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017802744896104072,
+      "loss": 0.821,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017798844830928817,
+      "loss": 0.8105,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017794941735570005,
+      "loss": 0.8149,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017791035611544148,
+      "loss": 0.7874,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001778712646036894,
+      "loss": 0.7915,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017783214283563262,
+      "loss": 0.791,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017779299082647148,
+      "loss": 0.8091,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017775380859141835,
+      "loss": 0.8418,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001777145961456971,
+      "loss": 0.8157,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017767535350454345,
+      "loss": 0.7688,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776360806832049,
+      "loss": 0.7844,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001775967776969405,
+      "loss": 0.7837,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017755744456102122,
+      "loss": 0.8086,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017751808129072967,
+      "loss": 0.8257,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774786879013601,
+      "loss": 0.8591,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017743926440821857,
+      "loss": 0.8101,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017739981082662276,
+      "loss": 0.8005,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017736032717190206,
+      "loss": 0.7839,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773208134593976,
+      "loss": 0.7532,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017728126970446208,
+      "loss": 0.7815,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017724169592245995,
+      "loss": 0.8127,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017720209212876737,
+      "loss": 0.8179,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017716245833877201,
+      "loss": 0.8311,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017712279456787332,
+      "loss": 0.8186,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017708310083148243,
+      "loss": 0.8191,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017704337714502194,
+      "loss": 0.821,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001770036235239263,
+      "loss": 0.8025,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017696383998364138,
+      "loss": 0.7756,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001769240265396249,
+      "loss": 0.8528,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017688418320734598,
+      "loss": 0.8291,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017684431000228548,
+      "loss": 0.7593,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017680440693993585,
+      "loss": 0.7913,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001767644740358011,
+      "loss": 0.7731,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017672451130539696,
+      "loss": 0.8284,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017668451876425053,
+      "loss": 0.8435,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017664449642790068,
+      "loss": 0.801,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766044443118978,
+      "loss": 0.7537,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017656436243180385,
+      "loss": 0.8091,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765242508031923,
+      "loss": 0.8286,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017648410944164823,
+      "loss": 0.8435,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017644393836276832,
+      "loss": 0.8,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017640373758216077,
+      "loss": 0.7761,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017636350711544517,
+      "loss": 0.7991,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017632324697825288,
+      "loss": 0.8359,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017628295718622665,
+      "loss": 0.8108,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762426377550208,
+      "loss": 0.8345,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017620228870030108,
+      "loss": 0.8049,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017616191003774488,
+      "loss": 0.8279,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000176121501783041,
+      "loss": 0.8171,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017608106395188981,
+      "loss": 0.7979,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760405965600031,
+      "loss": 0.8105,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017600009962310417,
+      "loss": 0.8125,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017595957315692782,
+      "loss": 0.8215,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759190171772203,
+      "loss": 0.8049,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017587843169973935,
+      "loss": 0.7944,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017583781674025413,
+      "loss": 0.7798,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757971723145453,
+      "loss": 0.8313,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017575649843840497,
+      "loss": 0.8354,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017571579512763665,
+      "loss": 0.844,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001756750623980553,
+      "loss": 0.3899,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017563430026548734,
+      "loss": 0.7893,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017559350874577065,
+      "loss": 0.8088,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017555268785475437,
+      "loss": 0.7905,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001755118376082992,
+      "loss": 0.8071,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017547095802227723,
+      "loss": 0.7717,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017543004911257186,
+      "loss": 0.7368,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175389110895078,
+      "loss": 0.7876,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017534814338570186,
+      "loss": 0.375,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017530714660036112,
+      "loss": 0.8181,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017526612055498468,
+      "loss": 0.7871,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175225065265513,
+      "loss": 0.7983,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017518398074789775,
+      "loss": 0.8367,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017514286701810202,
+      "loss": 0.7419,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017510172409210031,
+      "loss": 0.8086,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017506055198587833,
+      "loss": 0.769,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017501935071543323,
+      "loss": 0.8667,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017497812029677344,
+      "loss": 0.8198,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017493686074591873,
+      "loss": 0.802,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017489557207890023,
+      "loss": 0.8276,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748542543117603,
+      "loss": 0.7795,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748129074605527,
+      "loss": 0.7764,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017477153154134243,
+      "loss": 0.8167,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001747301265702058,
+      "loss": 0.8193,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017468869256323036,
+      "loss": 0.7671,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017464722953651504,
+      "loss": 0.7681,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017460573750616994,
+      "loss": 0.7603,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017456421648831655,
+      "loss": 0.7866,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017452266649908756,
+      "loss": 0.7682,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744810875546268,
+      "loss": 0.7854,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744394796710896,
+      "loss": 0.8428,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001743978428646423,
+      "loss": 0.78,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017435617715146263,
+      "loss": 0.7925,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017431448254773944,
+      "loss": 0.7822,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001742727590696729,
+      "loss": 0.7983,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017423100673347435,
+      "loss": 0.813,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017418922555536637,
+      "loss": 0.8362,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017414741555158266,
+      "loss": 0.835,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741055767383683,
+      "loss": 0.7976,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017406370913197936,
+      "loss": 0.7942,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017402181274868322,
+      "loss": 0.7749,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739798876047584,
+      "loss": 0.7388,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017393793371649462,
+      "loss": 0.7842,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738959511001928,
+      "loss": 0.7903,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017385393977216484,
+      "loss": 0.8379,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017381189974873407,
+      "loss": 0.7939,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017376983104623475,
+      "loss": 0.7917,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001737277336810124,
+      "loss": 0.3773,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017368560766942362,
+      "loss": 0.3832,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001736434530278362,
+      "loss": 0.7803,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173601269772629,
+      "loss": 0.3835,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017355905792019194,
+      "loss": 0.8071,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735168174869262,
+      "loss": 0.8064,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017347454848924397,
+      "loss": 0.803,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017343225094356855,
+      "loss": 0.7876,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017338992486633437,
+      "loss": 0.8062,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017334757027398688,
+      "loss": 0.7693,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017330518718298264,
+      "loss": 0.8132,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017326277560978932,
+      "loss": 0.8308,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017322033557088558,
+      "loss": 0.8416,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017317786708276122,
+      "loss": 0.7776,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017313537016191706,
+      "loss": 0.825,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017309284482486495,
+      "loss": 0.3688,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017305029108812776,
+      "loss": 0.8345,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001730077089682395,
+      "loss": 0.7159,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017296509848174508,
+      "loss": 0.8025,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017292245964520056,
+      "loss": 0.7975,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017287979247517286,
+      "loss": 0.8276,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017283709698824002,
+      "loss": 0.7908,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727943732009911,
+      "loss": 0.8333,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727516211300261,
+      "loss": 0.7817,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017270884079195596,
+      "loss": 0.8633,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726660322034027,
+      "loss": 0.8391,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001726231953809993,
+      "loss": 0.7678,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017258033034138972,
+      "loss": 0.8059,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017253743710122875,
+      "loss": 0.8145,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017249451567718236,
+      "loss": 0.8208,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017245156608592727,
+      "loss": 0.7698,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017240858834415126,
+      "loss": 0.7715,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000172365582468553,
+      "loss": 0.3782,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017232254847584212,
+      "loss": 0.8,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017227948638273916,
+      "loss": 0.7224,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017223639620597556,
+      "loss": 0.8145,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001721932779622937,
+      "loss": 0.7646,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017215013166844688,
+      "loss": 0.8342,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017210695734119923,
+      "loss": 0.8403,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017206375499732586,
+      "loss": 0.7935,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017202052465361268,
+      "loss": 0.8538,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017197726632685653,
+      "loss": 0.8235,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001719339800338651,
+      "loss": 0.7915,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017189066579145705,
+      "loss": 0.8711,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017184732361646168,
+      "loss": 0.8184,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001718039535257194,
+      "loss": 0.8433,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717605555360812,
+      "loss": 0.8018,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017171712966440917,
+      "loss": 0.8472,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017167367592757602,
+      "loss": 0.8069,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017163019434246547,
+      "loss": 0.7803,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017158668492597186,
+      "loss": 0.8228,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017154314769500054,
+      "loss": 0.8237,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017149958266646754,
+      "loss": 0.7817,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017145598985729977,
+      "loss": 0.7678,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017141236928443484,
+      "loss": 0.8152,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001713687209648212,
+      "loss": 0.8301,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017132504491541818,
+      "loss": 0.7839,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017128134115319566,
+      "loss": 0.8228,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001712376096951345,
+      "loss": 0.8184,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017119385055822619,
+      "loss": 0.8069,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017115006375947303,
+      "loss": 0.7363,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017110624931588806,
+      "loss": 0.7617,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017106240724449508,
+      "loss": 0.7595,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017101853756232855,
+      "loss": 0.7996,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017097464028643374,
+      "loss": 0.3662,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017093071543386668,
+      "loss": 0.8147,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017088676302169393,
+      "loss": 0.7979,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017084278306699294,
+      "loss": 0.7947,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017079877558685177,
+      "loss": 0.8228,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001707547405983692,
+      "loss": 0.7788,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.7756,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017066658816482854,
+      "loss": 0.8408,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017062247075402137,
+      "loss": 0.8198,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001705783259033748,
+      "loss": 0.7789,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001705341536300409,
+      "loss": 0.7726,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017048995395118252,
+      "loss": 0.8638,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017044572688397315,
+      "loss": 0.3573,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017040147244559684,
+      "loss": 0.8428,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017035719065324837,
+      "loss": 0.8083,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017031288152413306,
+      "loss": 0.7996,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017026854507546692,
+      "loss": 0.8225,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017022418132447655,
+      "loss": 0.7971,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017017979028839916,
+      "loss": 0.7773,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017013537198448252,
+      "loss": 0.7881,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001700909264299851,
+      "loss": 0.821,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017004645364217583,
+      "loss": 0.7795,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017000195363833434,
+      "loss": 0.7944,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016995742643575074,
+      "loss": 0.8291,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016991287205172573,
+      "loss": 0.801,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016986829050357068,
+      "loss": 0.7988,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016982368180860728,
+      "loss": 0.845,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016977904598416803,
+      "loss": 0.7458,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016973438304759578,
+      "loss": 0.7461,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016968969301624402,
+      "loss": 0.822,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696449759074767,
+      "loss": 0.8025,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016960023173866835,
+      "loss": 0.8398,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016955546052720396,
+      "loss": 0.7593,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001695106622904791,
+      "loss": 0.8042,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016946583704589973,
+      "loss": 0.8156,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016942098481088243,
+      "loss": 0.8092,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016937610560285418,
+      "loss": 0.8083,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016933119943925243,
+      "loss": 0.8462,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016928626633752521,
+      "loss": 0.3707,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016924130631513088,
+      "loss": 0.835,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016919631938953832,
+      "loss": 0.8203,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016915130557822695,
+      "loss": 0.7849,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016910626489868649,
+      "loss": 0.8225,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016906119736841716,
+      "loss": 0.8296,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016901610300492961,
+      "loss": 0.7817,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168970981825745,
+      "loss": 0.793,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001689258338483947,
+      "loss": 0.7937,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016888065909042075,
+      "loss": 0.844,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001688354575693754,
+      "loss": 0.8162,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687902293028214,
+      "loss": 0.782,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016874497430833182,
+      "loss": 0.8037,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016869969260349018,
+      "loss": 0.8103,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016865438420589034,
+      "loss": 0.7654,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016860904913313656,
+      "loss": 0.8301,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016856368740284343,
+      "loss": 0.7986,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685182990326359,
+      "loss": 0.738,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016847288404014935,
+      "loss": 0.7913,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001684274424430294,
+      "loss": 0.8215,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016838197425893202,
+      "loss": 0.7715,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016833647950552355,
+      "loss": 0.7522,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016829095820048072,
+      "loss": 0.8125,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016824541036149037,
+      "loss": 0.8298,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016819983600624986,
+      "loss": 0.7522,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016815423515246673,
+      "loss": 0.7893,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016810860781785888,
+      "loss": 0.8054,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016806295402015448,
+      "loss": 0.8015,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016801727377709194,
+      "loss": 0.8093,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016797156710642,
+      "loss": 0.8323,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016792583402589766,
+      "loss": 0.7971,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001678800745532942,
+      "loss": 0.8308,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016783428870638902,
+      "loss": 0.7778,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016778847650297197,
+      "loss": 0.8118,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016774263796084303,
+      "loss": 0.7834,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016769677309781239,
+      "loss": 0.8096,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016765088193170053,
+      "loss": 0.8035,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016760496448033811,
+      "loss": 0.8103,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016755902076156604,
+      "loss": 0.3691,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001675130507932354,
+      "loss": 0.7893,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016746705459320745,
+      "loss": 0.8083,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016742103217935373,
+      "loss": 0.7957,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016737498356955585,
+      "loss": 0.7876,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001673289087817057,
+      "loss": 0.8181,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672828078337053,
+      "loss": 0.7996,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672366807434668,
+      "loss": 0.7502,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671905275289126,
+      "loss": 0.7441,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016714434820797513,
+      "loss": 0.7537,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016709814279859702,
+      "loss": 0.3777,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016705191131873107,
+      "loss": 0.3896,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016700565378634021,
+      "loss": 0.762,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669593702193974,
+      "loss": 0.8306,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016691306063588583,
+      "loss": 0.3893,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668667250537987,
+      "loss": 0.7737,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016682036349113937,
+      "loss": 0.8696,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016677397596592136,
+      "loss": 0.3994,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016672756249616808,
+      "loss": 0.7639,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666811230999132,
+      "loss": 0.8198,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666346577952004,
+      "loss": 0.8132,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016658816660008343,
+      "loss": 0.8267,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016654164953262612,
+      "loss": 0.7949,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664951066109023,
+      "loss": 0.7712,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001664485378529959,
+      "loss": 0.7605,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016640194327700086,
+      "loss": 0.802,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016635532290102114,
+      "loss": 0.7473,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016630867674317077,
+      "loss": 0.7612,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016626200482157378,
+      "loss": 0.7764,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016621530715436414,
+      "loss": 0.8054,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016616858375968595,
+      "loss": 0.7627,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016612183465569322,
+      "loss": 0.8079,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016607505986054996,
+      "loss": 0.8096,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016602825939243017,
+      "loss": 0.8169,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016598143326951786,
+      "loss": 0.8352,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016593458151000688,
+      "loss": 0.8286,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016588770413210126,
+      "loss": 0.8047,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016584080115401476,
+      "loss": 0.7751,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016579387259397127,
+      "loss": 0.7739,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016574691847020447,
+      "loss": 0.8267,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016569993880095806,
+      "loss": 0.8267,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016565293360448565,
+      "loss": 0.7773,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016560590289905073,
+      "loss": 0.7847,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001655588467029268,
+      "loss": 0.8425,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016551176503439713,
+      "loss": 0.7771,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016546465791175496,
+      "loss": 0.7817,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016541752535330345,
+      "loss": 0.7305,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016537036737735564,
+      "loss": 0.8359,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016532318400223428,
+      "loss": 0.7815,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016527597524627224,
+      "loss": 0.803,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016522874112781213,
+      "loss": 0.7395,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016518148166520637,
+      "loss": 0.8101,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016513419687681728,
+      "loss": 0.8069,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016508688678101705,
+      "loss": 0.7859,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016503955139618762,
+      "loss": 0.8413,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016499219074072086,
+      "loss": 0.7554,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016494480483301836,
+      "loss": 0.7896,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648973936914916,
+      "loss": 0.8203,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016484995733456178,
+      "loss": 0.8228,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016480249578066,
+      "loss": 0.7864,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016475500904822706,
+      "loss": 0.7493,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647074971557136,
+      "loss": 0.7872,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016465996012157995,
+      "loss": 0.8088,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001646123979642964,
+      "loss": 0.7546,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016456481070234276,
+      "loss": 0.813,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016451719835420877,
+      "loss": 0.8423,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016446956093839384,
+      "loss": 0.812,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016442189847340713,
+      "loss": 0.7839,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016437421097776752,
+      "loss": 0.7952,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001643264984700037,
+      "loss": 0.8062,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016427876096865394,
+      "loss": 0.7827,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016423099849226632,
+      "loss": 0.7893,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016418321105939862,
+      "loss": 0.8208,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016413539868861827,
+      "loss": 0.7764,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640875613985024,
+      "loss": 0.8191,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016403969920763788,
+      "loss": 0.8008,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001639918121346212,
+      "loss": 0.824,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016394390019805848,
+      "loss": 0.8579,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001638959634165656,
+      "loss": 0.7668,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016384800180876803,
+      "loss": 0.7686,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001638000153933009,
+      "loss": 0.8503,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016375200418880894,
+      "loss": 0.845,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637039682139466,
+      "loss": 0.7463,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016365590748737785,
+      "loss": 0.3835,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001636078220277764,
+      "loss": 0.8235,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016355971185382545,
+      "loss": 0.7729,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635115769842179,
+      "loss": 0.8301,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016346341743765614,
+      "loss": 0.8303,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016341523323285223,
+      "loss": 0.3727,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016336702438852777,
+      "loss": 0.8167,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000163318790923414,
+      "loss": 0.821,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016327053285625164,
+      "loss": 0.802,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016322225020579099,
+      "loss": 0.7729,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016317394299079194,
+      "loss": 0.813,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631256112300239,
+      "loss": 0.7725,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016307725494226584,
+      "loss": 0.8005,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630288741463062,
+      "loss": 0.8113,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000162980468860943,
+      "loss": 0.7759,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016293203910498376,
+      "loss": 0.8081,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628835848972455,
+      "loss": 0.814,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016283510625655472,
+      "loss": 0.7827,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016278660320174747,
+      "loss": 0.8318,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016273807575166926,
+      "loss": 0.3898,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016268952392517504,
+      "loss": 0.8025,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001626409477411293,
+      "loss": 0.8479,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016259234721840591,
+      "loss": 0.7988,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001625437223758883,
+      "loss": 0.803,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016249507323246926,
+      "loss": 0.7603,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016244639980705103,
+      "loss": 0.8369,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016239770211854538,
+      "loss": 0.7913,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016234898018587337,
+      "loss": 0.7935,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016230023402796556,
+      "loss": 0.7568,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016225146366376198,
+      "loss": 0.7844,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001622026691122119,
+      "loss": 0.7649,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621538503922741,
+      "loss": 0.8174,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621050075229168,
+      "loss": 0.8008,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001620561405231175,
+      "loss": 0.7561,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016200724941186305,
+      "loss": 0.3842,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016195833420814984,
+      "loss": 0.825,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016190939493098344,
+      "loss": 0.8213,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016186043159937882,
+      "loss": 0.7939,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016181144423236038,
+      "loss": 0.7812,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617624328489618,
+      "loss": 0.7883,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016171339746822604,
+      "loss": 0.8052,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016166433810920549,
+      "loss": 0.7461,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016161525479096178,
+      "loss": 0.8435,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001615661475325658,
+      "loss": 0.8118,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016151701635309796,
+      "loss": 0.7883,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016146786127164771,
+      "loss": 0.3826,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016141868230731395,
+      "loss": 0.7944,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016136947947920476,
+      "loss": 0.8313,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016132025280643752,
+      "loss": 0.7295,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016127100230813897,
+      "loss": 0.3602,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016122172800344493,
+      "loss": 0.7644,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016117242991150064,
+      "loss": 0.7898,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001611231080514605,
+      "loss": 0.3623,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001610737624424881,
+      "loss": 0.769,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016102439310375637,
+      "loss": 0.7432,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016097500005444738,
+      "loss": 0.8149,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016092558331375244,
+      "loss": 0.8333,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016087614290087208,
+      "loss": 0.7456,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016082667883501593,
+      "loss": 0.7717,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016077719113540302,
+      "loss": 0.8042,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001607276798212613,
+      "loss": 0.792,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016067814491182807,
+      "loss": 0.7688,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606285864263498,
+      "loss": 0.8359,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000160579004384082,
+      "loss": 0.8218,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016052939880428944,
+      "loss": 0.8145,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016047976970624603,
+      "loss": 0.8425,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016043011710923475,
+      "loss": 0.8208,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016038044103254775,
+      "loss": 0.811,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001603307414954863,
+      "loss": 0.7712,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016028101851736084,
+      "loss": 0.7622,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016023127211749083,
+      "loss": 0.7937,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016018150231520486,
+      "loss": 0.7876,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016013170912984058,
+      "loss": 0.7422,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016008189258074484,
+      "loss": 0.8433,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600320526872735,
+      "loss": 0.8059,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001599821894687914,
+      "loss": 0.7468,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015993230294467258,
+      "loss": 0.8179,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015988239313430004,
+      "loss": 0.752,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015983246005706593,
+      "loss": 0.8381,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015978250373237132,
+      "loss": 0.7939,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597325241796264,
+      "loss": 0.7661,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015968252141825037,
+      "loss": 0.7666,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015963249546767144,
+      "loss": 0.7441,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015958244634732674,
+      "loss": 0.7449,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595323740766626,
+      "loss": 0.7673,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015948227867513415,
+      "loss": 0.7485,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015943216016220562,
+      "loss": 0.8252,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015938201855735014,
+      "loss": 0.8027,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015933185388005,
+      "loss": 0.7852,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015928166614979613,
+      "loss": 0.8088,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015923145538608875,
+      "loss": 0.7974,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015918122160843678,
+      "loss": 0.7722,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015913096483635824,
+      "loss": 0.7806,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015908068508938,
+      "loss": 0.8103,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015903038238703794,
+      "loss": 0.8452,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015898005674887673,
+      "loss": 0.7212,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589297081944501,
+      "loss": 0.7871,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015887933674332046,
+      "loss": 0.8286,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015882894241505945,
+      "loss": 0.7573,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015877852522924732,
+      "loss": 0.8145,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015872808520547334,
+      "loss": 0.8008,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586776223633355,
+      "loss": 0.7971,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015862713672244093,
+      "loss": 0.7947,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015857662830240528,
+      "loss": 0.7322,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015852609712285339,
+      "loss": 0.8003,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015847554320341865,
+      "loss": 0.791,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015842496656374346,
+      "loss": 0.7898,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000158374367223479,
+      "loss": 0.7769,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015832374520228525,
+      "loss": 0.7954,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015827310051983107,
+      "loss": 0.8433,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015822243319579403,
+      "loss": 0.8523,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015817174324986058,
+      "loss": 0.7631,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581210307017259,
+      "loss": 0.7944,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015807029557109398,
+      "loss": 0.8042,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580195378776776,
+      "loss": 0.8022,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015796875764119826,
+      "loss": 0.8171,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015791795488138624,
+      "loss": 0.8186,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001578671296179806,
+      "loss": 0.7866,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001578162818707291,
+      "loss": 0.7847,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001577654116593883,
+      "loss": 0.7927,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001577145190037234,
+      "loss": 0.7981,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015766360392350837,
+      "loss": 0.8196,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015761266643852587,
+      "loss": 0.7974,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015756170656856737,
+      "loss": 0.832,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015751072433343286,
+      "loss": 0.771,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015745971975293115,
+      "loss": 0.7607,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574086928468797,
+      "loss": 0.8101,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001573576436351046,
+      "loss": 0.7998,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001573065721374407,
+      "loss": 0.8076,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572554783737314,
+      "loss": 0.7944,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015720436236382884,
+      "loss": 0.7656,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015715322412759375,
+      "loss": 0.8416,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015710206368489552,
+      "loss": 0.7566,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015705088105561217,
+      "loss": 0.7935,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015699967625963033,
+      "loss": 0.8137,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001569484493168452,
+      "loss": 0.7925,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001568972002471607,
+      "loss": 0.8025,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015684592907048926,
+      "loss": 0.7844,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015679463580675187,
+      "loss": 0.7971,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567433204758782,
+      "loss": 0.7698,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001566919830978064,
+      "loss": 0.8254,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015664062369248328,
+      "loss": 0.8496,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015658924227986414,
+      "loss": 0.7585,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015653783887991282,
+      "loss": 0.7874,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015648641351260177,
+      "loss": 0.8022,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001564349661979119,
+      "loss": 0.8208,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015638349695583275,
+      "loss": 0.8137,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001563320058063622,
+      "loss": 0.8076,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015628049276950688,
+      "loss": 0.7913,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001562289578652817,
+      "loss": 0.8345,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015617740111371022,
+      "loss": 0.7817,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015612582253482445,
+      "loss": 0.8149,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001560742221486648,
+      "loss": 0.8262,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015602259997528028,
+      "loss": 0.7732,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015597095603472828,
+      "loss": 0.7629,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001559192903470747,
+      "loss": 0.7717,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001558676029323938,
+      "loss": 0.7932,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001558158938107684,
+      "loss": 0.7795,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001557641630022897,
+      "loss": 0.7764,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015571241052705722,
+      "loss": 0.7676,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015566063640517916,
+      "loss": 0.8162,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015560884065677185,
+      "loss": 0.8196,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015555702330196023,
+      "loss": 0.8022,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001555051843608775,
+      "loss": 0.7554,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001554533238536653,
+      "loss": 0.7798,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015540144180047367,
+      "loss": 0.7506,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015534953822146096,
+      "loss": 0.7749,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015529761313679393,
+      "loss": 0.7898,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015524566656664775,
+      "loss": 0.8416,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001551936985312058,
+      "loss": 0.7673,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015514170905065995,
+      "loss": 0.7346,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015508969814521025,
+      "loss": 0.813,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015503766583506524,
+      "loss": 0.8096,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549856121404416,
+      "loss": 0.7917,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015493353708156442,
+      "loss": 0.844,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548814406786671,
+      "loss": 0.7661,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548293229519914,
+      "loss": 0.4109,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015477718392178716,
+      "loss": 0.7717,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015472502360831266,
+      "loss": 0.8127,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015467284203183435,
+      "loss": 0.7795,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001546206392126271,
+      "loss": 0.8384,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015456841517097385,
+      "loss": 0.8484,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545161699271659,
+      "loss": 0.7731,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015446390350150273,
+      "loss": 0.7827,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015441161591429208,
+      "loss": 0.8132,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015435930718584994,
+      "loss": 0.7478,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015430697733650044,
+      "loss": 0.8335,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015425462638657595,
+      "loss": 0.3817,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001542022543564171,
+      "loss": 0.7769,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015414986126637258,
+      "loss": 0.7842,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015409744713679941,
+      "loss": 0.3845,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015404501198806268,
+      "loss": 0.7598,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015399255584053567,
+      "loss": 0.8381,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015394007871459984,
+      "loss": 0.7471,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538875806306448,
+      "loss": 0.8384,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015383506160906825,
+      "loss": 0.7856,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015378252167027614,
+      "loss": 0.7585,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001537299608346824,
+      "loss": 0.7986,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001536773791227092,
+      "loss": 0.8074,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015362477655478677,
+      "loss": 0.7991,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015357215315135348,
+      "loss": 0.7764,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535195089328557,
+      "loss": 0.8499,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015346684391974794,
+      "loss": 0.7864,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015341415813249288,
+      "loss": 0.8206,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015336145159156115,
+      "loss": 0.8357,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001533087243174315,
+      "loss": 0.7329,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015325597633059074,
+      "loss": 0.7551,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015320320765153367,
+      "loss": 0.7903,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015315041830076318,
+      "loss": 0.7629,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015309760829879018,
+      "loss": 0.7563,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015304477766613366,
+      "loss": 0.8135,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001529919264233205,
+      "loss": 0.7812,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015293905459088568,
+      "loss": 0.417,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001528861621893722,
+      "loss": 0.8289,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015283324923933097,
+      "loss": 0.7544,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015278031576132094,
+      "loss": 0.7834,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000152727361775909,
+      "loss": 0.7827,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001526743873036701,
+      "loss": 0.7578,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015262139236518695,
+      "loss": 0.8201,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015256837698105047,
+      "loss": 0.7578,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015251534117185936,
+      "loss": 0.3706,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015246228495822023,
+      "loss": 0.7971,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015240920836074776,
+      "loss": 0.8042,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015235611140006444,
+      "loss": 0.787,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015230299409680072,
+      "loss": 0.7529,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522498564715949,
+      "loss": 0.8135,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015219669854509322,
+      "loss": 0.7947,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001521435203379498,
+      "loss": 0.7671,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520903218708267,
+      "loss": 0.7788,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001520371031643937,
+      "loss": 0.7983,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015198386423932855,
+      "loss": 0.3628,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519306051163169,
+      "loss": 0.7673,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015187732581605217,
+      "loss": 0.7863,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001518240263592356,
+      "loss": 0.8271,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015177070676657633,
+      "loss": 0.8184,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015171736705879126,
+      "loss": 0.8291,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001516640072566052,
+      "loss": 0.7544,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015161062738075067,
+      "loss": 0.75,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000151557227451968,
+      "loss": 0.7737,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015150380749100545,
+      "loss": 0.7998,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514503675186188,
+      "loss": 0.7837,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015139690755557186,
+      "loss": 0.7844,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015134342762263605,
+      "loss": 0.3549,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015128992774059063,
+      "loss": 0.7808,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015123640793022255,
+      "loss": 0.7979,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001511828682123266,
+      "loss": 0.79,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001511293086077052,
+      "loss": 0.8113,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015107572913716858,
+      "loss": 0.7678,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015102212982153458,
+      "loss": 0.8059,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015096851068162884,
+      "loss": 0.801,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001509148717382847,
+      "loss": 0.7947,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015086121301234316,
+      "loss": 0.8286,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015080753452465296,
+      "loss": 0.7554,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015075383629607042,
+      "loss": 0.8469,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015070011834745963,
+      "loss": 0.8071,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015064638069969228,
+      "loss": 0.8047,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505926233736478,
+      "loss": 0.7888,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505388463902131,
+      "loss": 0.8098,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015048504977028287,
+      "loss": 0.3713,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015043123353475943,
+      "loss": 0.8079,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015037739770455263,
+      "loss": 0.7722,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015032354230058004,
+      "loss": 0.739,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001502696673437667,
+      "loss": 0.7538,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015021577285504538,
+      "loss": 0.8303,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015016185885535638,
+      "loss": 0.7712,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015010792536564762,
+      "loss": 0.79,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015005397240687446,
+      "loss": 0.8135,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.8547,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014994600816599482,
+      "loss": 0.7783,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014989199692583705,
+      "loss": 0.7661,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001498379663005123,
+      "loss": 0.3796,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014978391631101384,
+      "loss": 0.8066,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001497298469783424,
+      "loss": 0.7456,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014967575832350614,
+      "loss": 0.7668,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014962165036752086,
+      "loss": 0.3565,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014956752313140977,
+      "loss": 0.8547,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001495133766362037,
+      "loss": 0.7996,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014945921090294076,
+      "loss": 0.8286,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494050259526667,
+      "loss": 0.7922,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001493508218064347,
+      "loss": 0.8691,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014929659848530538,
+      "loss": 0.8362,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014924235601034672,
+      "loss": 0.8711,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014918809440263436,
+      "loss": 0.7954,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014913381368325115,
+      "loss": 0.4073,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014907951387328752,
+      "loss": 0.7886,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014902519499384123,
+      "loss": 0.7666,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001489708570660175,
+      "loss": 0.772,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014891650011092896,
+      "loss": 0.7834,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014886212414969553,
+      "loss": 0.7749,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488077292034446,
+      "loss": 0.76,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014875331529331097,
+      "loss": 0.3737,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014869888244043673,
+      "loss": 0.783,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014864443066597139,
+      "loss": 0.7817,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014858995999107174,
+      "loss": 0.8364,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148535470436902,
+      "loss": 0.7527,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014848096202463372,
+      "loss": 0.8079,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014842643477544563,
+      "loss": 0.7954,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000148371888710524,
+      "loss": 0.7786,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483173238510622,
+      "loss": 0.7913,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482627402182611,
+      "loss": 0.8474,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482081378333287,
+      "loss": 0.3561,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001481535167174804,
+      "loss": 0.8103,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014809887689193877,
+      "loss": 0.7449,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014804421837793377,
+      "loss": 0.7644,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014798954119670253,
+      "loss": 0.7637,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014793484536948948,
+      "loss": 0.8235,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014788013091754628,
+      "loss": 0.7742,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014782539786213183,
+      "loss": 0.78,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014777064622451222,
+      "loss": 0.7394,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014771587602596084,
+      "loss": 0.7849,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014766108728775823,
+      "loss": 0.7971,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014760628003119213,
+      "loss": 0.3481,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014755145427755754,
+      "loss": 0.793,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014749661004815655,
+      "loss": 0.7715,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014744174736429849,
+      "loss": 0.7913,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014738686624729986,
+      "loss": 0.7881,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014733196671848435,
+      "loss": 0.8188,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472770487991827,
+      "loss": 0.7954,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014722211251073293,
+      "loss": 0.775,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014716715787448006,
+      "loss": 0.3613,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014711218491177637,
+      "loss": 0.8081,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014705719364398117,
+      "loss": 0.7991,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014700218409246087,
+      "loss": 0.7991,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014694715627858908,
+      "loss": 0.3945,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014689211022374645,
+      "loss": 0.7886,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014683704594932071,
+      "loss": 0.7622,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014678196347670666,
+      "loss": 0.7969,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467268628273062,
+      "loss": 0.7747,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466717440225283,
+      "loss": 0.7874,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466166070837889,
+      "loss": 0.8584,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014656145203251114,
+      "loss": 0.8533,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014650627889012507,
+      "loss": 0.7866,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014645108767806777,
+      "loss": 0.7581,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014639587841778343,
+      "loss": 0.7917,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014634065113072316,
+      "loss": 0.8113,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014628540583834512,
+      "loss": 0.7881,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001462301425621145,
+      "loss": 0.7957,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014617486132350343,
+      "loss": 0.8386,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014611956214399097,
+      "loss": 0.8169,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014606424504506324,
+      "loss": 0.8337,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014600891004821332,
+      "loss": 0.8406,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014595355717494116,
+      "loss": 0.7568,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014589818644675378,
+      "loss": 0.8068,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000145842797885165,
+      "loss": 0.7471,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014578739151169568,
+      "loss": 0.8184,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014573196734787356,
+      "loss": 0.7998,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014567652541523323,
+      "loss": 0.8591,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001456210657353163,
+      "loss": 0.7852,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014556558832967123,
+      "loss": 0.7676,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014551009321985329,
+      "loss": 0.7822,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014545458042742476,
+      "loss": 0.7732,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014539904997395468,
+      "loss": 0.7791,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014534350188101904,
+      "loss": 0.7849,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014528793617020063,
+      "loss": 0.8149,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014523235286308908,
+      "loss": 0.7698,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014517675198128085,
+      "loss": 0.7865,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451211335463793,
+      "loss": 0.7976,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014506549757999454,
+      "loss": 0.7988,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001450098441037435,
+      "loss": 0.7572,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014495417313924996,
+      "loss": 0.4177,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014489848470814447,
+      "loss": 0.7581,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014484277883206433,
+      "loss": 0.7842,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014478705553265362,
+      "loss": 0.8242,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014473131483156327,
+      "loss": 0.7811,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014467555675045087,
+      "loss": 0.7988,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014461978131098088,
+      "loss": 0.741,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014456398853482437,
+      "loss": 0.7474,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014450817844365921,
+      "loss": 0.7721,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001444523510591701,
+      "loss": 0.7427,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014439650640304822,
+      "loss": 0.7834,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014434064449699168,
+      "loss": 0.7747,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014428476536270515,
+      "loss": 0.8032,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014422886902190014,
+      "loss": 0.8093,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001441729554962947,
+      "loss": 0.8013,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014411702480761362,
+      "loss": 0.8086,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014406107697758838,
+      "loss": 0.8342,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001440051120279571,
+      "loss": 0.8223,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439491299804645,
+      "loss": 0.7485,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014389313085686205,
+      "loss": 0.823,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014383711467890774,
+      "loss": 0.8091,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001437810814683663,
+      "loss": 0.7983,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014372503124700898,
+      "loss": 0.7644,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001436689640366137,
+      "loss": 0.8506,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014361287985896495,
+      "loss": 0.766,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014355677873585383,
+      "loss": 0.3712,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014350066068907798,
+      "loss": 0.4241,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014344452574044172,
+      "loss": 0.7629,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014338837391175582,
+      "loss": 0.8044,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014333220522483767,
+      "loss": 0.7688,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014327601970151122,
+      "loss": 0.7502,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001432198173636069,
+      "loss": 0.3796,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014316359823296173,
+      "loss": 0.8008,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014310736233141925,
+      "loss": 0.7505,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014305110968082952,
+      "loss": 0.7839,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014299484030304906,
+      "loss": 0.8313,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014293855421994094,
+      "loss": 0.7917,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014288225145337466,
+      "loss": 0.8069,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014282593202522627,
+      "loss": 0.8201,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001427695959573783,
+      "loss": 0.7844,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014271324327171968,
+      "loss": 0.7725,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014265687399014584,
+      "loss": 0.752,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014260048813455865,
+      "loss": 0.7751,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014254408572686642,
+      "loss": 0.7673,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014248766678898387,
+      "loss": 0.8059,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014243123134283217,
+      "loss": 0.759,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014237477941033888,
+      "loss": 0.7539,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014231831101343801,
+      "loss": 0.8118,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014226182617406996,
+      "loss": 0.8064,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014220532491418143,
+      "loss": 0.7576,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014214880725572563,
+      "loss": 0.8005,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014209227322066204,
+      "loss": 0.8552,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014203572283095657,
+      "loss": 0.8513,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014197915610858144,
+      "loss": 0.7649,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014192257307551518,
+      "loss": 0.7783,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001418659737537428,
+      "loss": 0.3997,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014180935816525552,
+      "loss": 0.8167,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417527263320509,
+      "loss": 0.7793,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014169607827613283,
+      "loss": 0.7952,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014163941401951145,
+      "loss": 0.8142,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415827335842033,
+      "loss": 0.7759,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001415260369922311,
+      "loss": 0.7815,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014146932426562392,
+      "loss": 0.7864,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014141259542641704,
+      "loss": 0.7395,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014135585049665207,
+      "loss": 0.8303,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001412990894983768,
+      "loss": 0.7219,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014124231245364534,
+      "loss": 0.8079,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014118551938451795,
+      "loss": 0.7507,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014112871031306119,
+      "loss": 0.8096,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014107188526134773,
+      "loss": 0.8188,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014101504425145662,
+      "loss": 0.7891,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014095818730547297,
+      "loss": 0.7422,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014090131444548812,
+      "loss": 0.7849,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014084442569359964,
+      "loss": 0.7654,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014078752107191117,
+      "loss": 0.7637,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014073060060253265,
+      "loss": 0.7917,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014067366430758004,
+      "loss": 0.8213,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014061671220917557,
+      "loss": 0.7903,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014055974432944753,
+      "loss": 0.7546,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014050276069053035,
+      "loss": 0.8296,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014044576131456467,
+      "loss": 0.8215,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014038874622369714,
+      "loss": 0.8208,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014033171544008052,
+      "loss": 0.7942,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014027466898587374,
+      "loss": 0.3879,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014021760688324176,
+      "loss": 0.8013,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014016052915435565,
+      "loss": 0.7979,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014010343582139252,
+      "loss": 0.7507,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001400463269065356,
+      "loss": 0.8003,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013998920243197407,
+      "loss": 0.3845,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013993206241990334,
+      "loss": 0.8186,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013987490689252463,
+      "loss": 0.3722,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013981773587204535,
+      "loss": 0.3181,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013976054938067884,
+      "loss": 0.781,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001397033474406445,
+      "loss": 0.7747,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013964613007416777,
+      "loss": 0.8293,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013958889730347998,
+      "loss": 0.7664,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013953164915081852,
+      "loss": 0.7676,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394743856384267,
+      "loss": 0.8208,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013941710678855396,
+      "loss": 0.761,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001393598126234554,
+      "loss": 0.8176,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013930250316539238,
+      "loss": 0.7593,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013924517843663198,
+      "loss": 0.7881,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013918783845944736,
+      "loss": 0.7598,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001391304832561175,
+      "loss": 0.7727,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013907311284892736,
+      "loss": 0.7769,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013901572726016781,
+      "loss": 0.7559,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013895832651213561,
+      "loss": 0.7773,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001389009106271333,
+      "loss": 0.8079,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013884347962746948,
+      "loss": 0.3846,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013878603353545854,
+      "loss": 0.8042,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013872857237342068,
+      "loss": 0.7498,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013867109616368207,
+      "loss": 0.7836,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013861360492857462,
+      "loss": 0.7526,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001385560986904362,
+      "loss": 0.7908,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013849857747161035,
+      "loss": 0.7327,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013844104129444654,
+      "loss": 0.7556,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013838349018130007,
+      "loss": 0.3859,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013832592415453191,
+      "loss": 0.8418,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.8037,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013821074744960394,
+      "loss": 0.7959,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013815313681619516,
+      "loss": 0.8118,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013809551135866683,
+      "loss": 0.7532,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013803787109940892,
+      "loss": 0.3837,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013798021606081714,
+      "loss": 0.759,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013792254626529286,
+      "loss": 0.7632,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013786486173524332,
+      "loss": 0.8359,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001378071624930814,
+      "loss": 0.7732,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013774944856122568,
+      "loss": 0.371,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013769171996210052,
+      "loss": 0.8386,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013763397671813596,
+      "loss": 0.7563,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013757621885176767,
+      "loss": 0.8049,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001375184463854371,
+      "loss": 0.7649,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013746065934159123,
+      "loss": 0.7825,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013740285774268283,
+      "loss": 0.7529,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013734504161117027,
+      "loss": 0.7668,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013728721096951761,
+      "loss": 0.7705,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013722936584019453,
+      "loss": 0.7627,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013717150624567627,
+      "loss": 0.8071,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013711363220844379,
+      "loss": 0.6968,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013705574375098365,
+      "loss": 0.7683,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001369978408957879,
+      "loss": 0.8311,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013693992366535434,
+      "loss": 0.793,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013688199208218622,
+      "loss": 0.7622,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013682404616879247,
+      "loss": 0.7715,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013676608594768755,
+      "loss": 0.7646,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013670811144139144,
+      "loss": 0.7285,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013665012267242974,
+      "loss": 0.8542,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013659211966333357,
+      "loss": 0.7629,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013653410243663952,
+      "loss": 0.7667,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013647607101488978,
+      "loss": 0.7603,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013641802542063206,
+      "loss": 0.7834,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001363599656764195,
+      "loss": 0.373,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013630189180481082,
+      "loss": 0.7781,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013624380382837016,
+      "loss": 0.7793,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013618570176966724,
+      "loss": 0.7673,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013612758565127711,
+      "loss": 0.8853,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001360694554957804,
+      "loss": 0.8406,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001360113113257631,
+      "loss": 0.7964,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013595315316381677,
+      "loss": 0.7432,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013589498103253828,
+      "loss": 0.7175,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013583679495453,
+      "loss": 0.8457,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013577859495239976,
+      "loss": 0.824,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013572038104876067,
+      "loss": 0.825,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001356621532662313,
+      "loss": 0.7546,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013560391162743569,
+      "loss": 0.8081,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013554565615500312,
+      "loss": 0.7905,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001354873868715684,
+      "loss": 0.7019,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013542910379977157,
+      "loss": 0.7834,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013537080696225814,
+      "loss": 0.7671,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013531249638167888,
+      "loss": 0.7585,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013525417208068996,
+      "loss": 0.8005,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013519583408195282,
+      "loss": 0.7407,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351374824081343,
+      "loss": 0.3821,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013507911708190645,
+      "loss": 0.7629,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013502073812594675,
+      "loss": 0.741,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001349623455629379,
+      "loss": 0.3861,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013490393941556787,
+      "loss": 0.7769,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013484551970652995,
+      "loss": 0.3911,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013478708645852272,
+      "loss": 0.8014,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013472863969425,
+      "loss": 0.7632,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013467017943642073,
+      "loss": 0.7681,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001346117057077493,
+      "loss": 0.7764,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013455321853095525,
+      "loss": 0.8083,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013449471792876334,
+      "loss": 0.8101,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001344362039239035,
+      "loss": 0.7668,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013437767653911098,
+      "loss": 0.8003,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001343191357971261,
+      "loss": 0.8281,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013426058172069452,
+      "loss": 0.7244,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013420201433256689,
+      "loss": 0.7988,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013414343365549918,
+      "loss": 0.7925,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001340848397122525,
+      "loss": 0.7363,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013402623252559308,
+      "loss": 0.3954,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013396761211829228,
+      "loss": 0.7004,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013390897851312667,
+      "loss": 0.7944,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013385033173287788,
+      "loss": 0.783,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013379167180033272,
+      "loss": 0.7791,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013373299873828303,
+      "loss": 0.7944,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013367431256952579,
+      "loss": 0.7637,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001336156133168631,
+      "loss": 0.7957,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013355690100310213,
+      "loss": 0.731,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013349817565105508,
+      "loss": 0.8303,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001334394372835393,
+      "loss": 0.825,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333806859233771,
+      "loss": 0.7886,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013332192159339594,
+      "loss": 0.833,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013326314431642822,
+      "loss": 0.8054,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013320435411531145,
+      "loss": 0.8071,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013314555101288808,
+      "loss": 0.824,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013308673503200567,
+      "loss": 0.7903,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013302790619551674,
+      "loss": 0.7993,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013296906452627872,
+      "loss": 0.8027,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013291021004715423,
+      "loss": 0.8088,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013285134278101063,
+      "loss": 0.7627,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013279246275072046,
+      "loss": 0.7749,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013273356997916104,
+      "loss": 0.7166,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013267466448921476,
+      "loss": 0.7621,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013261574630376888,
+      "loss": 0.7942,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013255681544571568,
+      "loss": 0.7871,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013249787193795226,
+      "loss": 0.7546,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013243891580338072,
+      "loss": 0.767,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132379947064908,
+      "loss": 0.7505,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013232096574544602,
+      "loss": 0.7419,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013226197186791148,
+      "loss": 0.832,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132202965455226,
+      "loss": 0.8284,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013214394653031616,
+      "loss": 0.7715,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001320849151161133,
+      "loss": 0.7749,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001320258712355536,
+      "loss": 0.3842,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013196681491157818,
+      "loss": 0.7632,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013190774616713287,
+      "loss": 0.7766,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013184866502516845,
+      "loss": 0.7971,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013178957150864046,
+      "loss": 0.7751,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013173046564050924,
+      "loss": 0.7834,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001316713474437399,
+      "loss": 0.804,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013161221694130246,
+      "loss": 0.3749,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013155307415617154,
+      "loss": 0.7517,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013149391911132673,
+      "loss": 0.7271,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013143475182975222,
+      "loss": 0.8022,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013137557233443707,
+      "loss": 0.791,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013131638064837494,
+      "loss": 0.7847,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013125717679456447,
+      "loss": 0.738,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013119796079600875,
+      "loss": 0.8352,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013113873267571577,
+      "loss": 0.8168,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001310794924566982,
+      "loss": 0.7317,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013102024016197337,
+      "loss": 0.8313,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001309609758145633,
+      "loss": 0.7649,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.7681,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001308424110537991,
+      "loss": 0.7546,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001307831106865124,
+      "loss": 0.8032,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013072379835867538,
+      "loss": 0.7366,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013066447409333345,
+      "loss": 0.7878,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013060513791353655,
+      "loss": 0.7891,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013054578984233935,
+      "loss": 0.7988,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001304864299028011,
+      "loss": 0.769,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013042705811798565,
+      "loss": 0.7815,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013036767451096148,
+      "loss": 0.7512,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013030827910480165,
+      "loss": 0.7932,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013024887192258385,
+      "loss": 0.7539,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001301894529873902,
+      "loss": 0.8184,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001301300223223076,
+      "loss": 0.7671,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013007057995042732,
+      "loss": 0.8423,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001300111258948453,
+      "loss": 0.3854,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012995166017866193,
+      "loss": 0.3659,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001298921828249822,
+      "loss": 0.8079,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001298326938569156,
+      "loss": 0.7893,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012977319329757615,
+      "loss": 0.7695,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001297136811700823,
+      "loss": 0.7485,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012965415749755709,
+      "loss": 0.792,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000129594622303128,
+      "loss": 0.7573,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012953507560992694,
+      "loss": 0.8389,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012947551744109043,
+      "loss": 0.8291,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012941594781975928,
+      "loss": 0.7646,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012935636676907885,
+      "loss": 0.8152,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001292967743121989,
+      "loss": 0.4011,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012923717047227368,
+      "loss": 0.8079,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012917755527246179,
+      "loss": 0.7449,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012911792873592626,
+      "loss": 0.3544,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001290582908858346,
+      "loss": 0.801,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012899864174535864,
+      "loss": 0.7612,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012893898133767455,
+      "loss": 0.7822,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000128879309685963,
+      "loss": 0.7515,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012881962681340894,
+      "loss": 0.7734,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012875993274320173,
+      "loss": 0.8228,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012870022749853508,
+      "loss": 0.781,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012864051110260695,
+      "loss": 0.7129,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001285807835786198,
+      "loss": 0.7168,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012852104494978024,
+      "loss": 0.7703,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012846129523929928,
+      "loss": 0.7734,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012840153447039228,
+      "loss": 0.7866,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001283417626662788,
+      "loss": 0.7612,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012828197985018276,
+      "loss": 0.7126,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012822218604533232,
+      "loss": 0.751,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001281623812749599,
+      "loss": 0.7554,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012810256556230223,
+      "loss": 0.7957,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012804273893060028,
+      "loss": 0.8132,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012798290140309923,
+      "loss": 0.7725,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012792305300304848,
+      "loss": 0.8174,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001278631937537017,
+      "loss": 0.8237,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012780332367831677,
+      "loss": 0.7715,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012774344280015577,
+      "loss": 0.7937,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012768355114248494,
+      "loss": 0.7998,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001276236487285748,
+      "loss": 0.8184,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001275637355816999,
+      "loss": 0.762,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001275038117251392,
+      "loss": 0.741,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001274438771821755,
+      "loss": 0.7224,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012738393197609602,
+      "loss": 0.7871,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012732397613019202,
+      "loss": 0.7468,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001272640096677589,
+      "loss": 0.811,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012720403261209614,
+      "loss": 0.7253,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012714404498650743,
+      "loss": 0.7422,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012708404681430053,
+      "loss": 0.7993,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012702403811878726,
+      "loss": 0.8025,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012696401892328357,
+      "loss": 0.77,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012690398925110943,
+      "loss": 0.7988,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012684394912558896,
+      "loss": 0.7581,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012678389857005034,
+      "loss": 0.7422,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012672383760782568,
+      "loss": 0.7444,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012666376626225133,
+      "loss": 0.7788,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012660368455666752,
+      "loss": 0.7617,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012654359251441854,
+      "loss": 0.7705,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012648349015885273,
+      "loss": 0.8188,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012642337751332238,
+      "loss": 0.7756,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012636325460118388,
+      "loss": 0.7554,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012630312144579748,
+      "loss": 0.7408,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001262429780705275,
+      "loss": 0.741,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012618282449874222,
+      "loss": 0.8074,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012612266075381386,
+      "loss": 0.8093,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012606248685911853,
+      "loss": 0.7891,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012600230283803647,
+      "loss": 0.7964,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012594210871395164,
+      "loss": 0.7664,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012588190451025207,
+      "loss": 0.7562,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012582169025032966,
+      "loss": 0.7883,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012576146595758017,
+      "loss": 0.7571,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012570123165540336,
+      "loss": 0.7764,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012564098736720283,
+      "loss": 0.781,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012558073311638604,
+      "loss": 0.7771,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012552046892636427,
+      "loss": 0.8025,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012546019482055276,
+      "loss": 0.7737,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012539991082237062,
+      "loss": 0.7581,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001253396169552407,
+      "loss": 0.7676,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012527931324258975,
+      "loss": 0.6949,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012521899970784834,
+      "loss": 0.781,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012515867637445086,
+      "loss": 0.7495,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012509834326583546,
+      "loss": 0.7654,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012503800040544416,
+      "loss": 0.7896,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001249776478167227,
+      "loss": 0.7668,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012491728552312066,
+      "loss": 0.7185,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012485691354809133,
+      "loss": 0.7634,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012479653191509182,
+      "loss": 0.7449,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012473614064758296,
+      "loss": 0.8091,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012467573976902935,
+      "loss": 0.3511,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012461532930289933,
+      "loss": 0.7949,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012455490927266484,
+      "loss": 0.7446,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012449447970180172,
+      "loss": 0.7969,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244340406137894,
+      "loss": 0.8276,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012437359203211108,
+      "loss": 0.7747,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012431313398025354,
+      "loss": 0.7563,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001242526664817074,
+      "loss": 0.7627,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012419218955996676,
+      "loss": 0.7629,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001241317032385296,
+      "loss": 0.7783,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012407120754089732,
+      "loss": 0.7546,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012401070249057512,
+      "loss": 0.7705,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001239501881110718,
+      "loss": 0.3631,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001238896644258998,
+      "loss": 0.7358,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012382913145857506,
+      "loss": 0.7483,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012376858923261733,
+      "loss": 0.405,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012370803777154977,
+      "loss": 0.7815,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012364747709889924,
+      "loss": 0.7878,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001235869072381961,
+      "loss": 0.783,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012352632821297437,
+      "loss": 0.7847,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012346574004677154,
+      "loss": 0.7847,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012340514276312873,
+      "loss": 0.7769,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012334453638559057,
+      "loss": 0.7681,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012328392093770516,
+      "loss": 0.8228,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012322329644302426,
+      "loss": 0.7804,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012316266292510306,
+      "loss": 0.7527,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001231020204075002,
+      "loss": 0.3799,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012304136891377795,
+      "loss": 0.7668,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012298070846750197,
+      "loss": 0.7673,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012292003909224143,
+      "loss": 0.7922,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012285936081156897,
+      "loss": 0.8022,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012279867364906071,
+      "loss": 0.7502,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012273797762829615,
+      "loss": 0.7827,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012267727277285833,
+      "loss": 0.813,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012261655910633364,
+      "loss": 0.3816,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012255583665231196,
+      "loss": 0.8174,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001224951054343865,
+      "loss": 0.8652,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000122434365476154,
+      "loss": 0.7605,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012237361680121444,
+      "loss": 0.7969,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012231285943317132,
+      "loss": 0.781,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012225209339563145,
+      "loss": 0.855,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012219131871220505,
+      "loss": 0.7957,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012213053540650564,
+      "loss": 0.8257,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012206974350215015,
+      "loss": 0.8103,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012200894302275879,
+      "loss": 0.7949,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012194813399195518,
+      "loss": 0.7998,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012188731643336614,
+      "loss": 0.8032,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012182649037062202,
+      "loss": 0.8281,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012176565582735625,
+      "loss": 0.7278,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012170481282720563,
+      "loss": 0.7881,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012164396139381029,
+      "loss": 0.7183,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012158310155081361,
+      "loss": 0.7861,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012152223332186222,
+      "loss": 0.7869,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012146135673060599,
+      "loss": 0.7568,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012140047180069813,
+      "loss": 0.7896,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012133957855579501,
+      "loss": 0.7629,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012127867701955622,
+      "loss": 0.7524,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012121776721564464,
+      "loss": 0.8093,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012115684916772629,
+      "loss": 0.78,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012109592289947045,
+      "loss": 0.801,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012103498843454959,
+      "loss": 0.7183,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012097404579663929,
+      "loss": 0.7737,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012091309500941841,
+      "loss": 0.8169,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012085213609656891,
+      "loss": 0.7864,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012079116908177593,
+      "loss": 0.762,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012073019398872778,
+      "loss": 0.7632,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012066921084111583,
+      "loss": 0.8101,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012060821966263465,
+      "loss": 0.7681,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012054722047698192,
+      "loss": 0.7786,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012048621330785841,
+      "loss": 0.7798,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012042519817896804,
+      "loss": 0.7899,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012036417511401774,
+      "loss": 0.7283,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012030314413671762,
+      "loss": 0.7415,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012024210527078079,
+      "loss": 0.8472,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012018105853992346,
+      "loss": 0.731,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012012000396786485,
+      "loss": 0.7839,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012005894157832729,
+      "loss": 0.7549,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011999787139503614,
+      "loss": 0.7559,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011993679344171973,
+      "loss": 0.7478,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011987570774210948,
+      "loss": 0.3452,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011981461431993977,
+      "loss": 0.7349,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011975351319894801,
+      "loss": 0.7556,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011969240440287457,
+      "loss": 0.8103,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011963128795546285,
+      "loss": 0.7896,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011957016388045917,
+      "loss": 0.7593,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011950903220161285,
+      "loss": 0.7981,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011944789294267614,
+      "loss": 0.7529,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011938674612740423,
+      "loss": 0.7397,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011932559177955533,
+      "loss": 0.824,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011926442992289046,
+      "loss": 0.7573,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011920326058117364,
+      "loss": 0.8127,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011914208377817172,
+      "loss": 0.7299,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011908089953765449,
+      "loss": 0.8408,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011901970788339469,
+      "loss": 0.7922,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011895850883916786,
+      "loss": 0.8198,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011889730242875243,
+      "loss": 0.7268,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001188360886759297,
+      "loss": 0.7521,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011877486760448382,
+      "loss": 0.7258,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011871363923820183,
+      "loss": 0.7219,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011865240360087348,
+      "loss": 0.74,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011859116071629149,
+      "loss": 0.8057,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011852991060825127,
+      "loss": 0.7898,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011846865330055115,
+      "loss": 0.7947,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011840738881699219,
+      "loss": 0.759,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011834611718137824,
+      "loss": 0.8188,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011828483841751596,
+      "loss": 0.7166,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011822355254921478,
+      "loss": 0.8335,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001181622596002868,
+      "loss": 0.3756,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000118100959594547,
+      "loss": 0.3813,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011803965255581304,
+      "loss": 0.8018,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011797833850790527,
+      "loss": 0.7769,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001179170174746469,
+      "loss": 0.3329,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011785568947986367,
+      "loss": 0.7805,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011779435454738419,
+      "loss": 0.7651,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011773301270103968,
+      "loss": 0.7876,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011767166396466403,
+      "loss": 0.7672,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011761030836209383,
+      "loss": 0.762,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011754894591716841,
+      "loss": 0.7556,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011748757665372964,
+      "loss": 0.7605,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001174262005956221,
+      "loss": 0.719,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 0.8193,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011730342819079227,
+      "loss": 0.781,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011724203189177228,
+      "loss": 0.8123,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011718062889348815,
+      "loss": 0.7649,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011711921921979754,
+      "loss": 0.8384,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011705780289456068,
+      "loss": 0.7239,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011699637994164052,
+      "loss": 0.7058,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011693495038490245,
+      "loss": 0.7834,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011687351424821449,
+      "loss": 0.743,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001168120715554472,
+      "loss": 0.8098,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011675062233047364,
+      "loss": 0.7812,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011668916659716951,
+      "loss": 0.7878,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011662770437941293,
+      "loss": 0.7886,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011656623570108461,
+      "loss": 0.8123,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011650476058606777,
+      "loss": 0.801,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011644327905824808,
+      "loss": 0.7761,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011638179114151377,
+      "loss": 0.7617,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001163202968597555,
+      "loss": 0.811,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001162587962368664,
+      "loss": 0.7954,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011619728929674209,
+      "loss": 0.7781,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011613577606328068,
+      "loss": 0.7147,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011607425656038263,
+      "loss": 0.7937,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011601273081195087,
+      "loss": 0.7952,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011595119884189086,
+      "loss": 0.7358,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011588966067411034,
+      "loss": 0.8403,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001158281163325195,
+      "loss": 0.749,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011576656584103096,
+      "loss": 0.7578,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011570500922355972,
+      "loss": 0.7788,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001156434465040231,
+      "loss": 0.8093,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011558187770634088,
+      "loss": 0.7661,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011552030285443515,
+      "loss": 0.7937,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011545872197223038,
+      "loss": 0.7957,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011539713508365335,
+      "loss": 0.7202,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011533554221263321,
+      "loss": 0.376,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001152739433831014,
+      "loss": 0.7205,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011521233861899167,
+      "loss": 0.835,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011515072794424013,
+      "loss": 0.7927,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011508911138278515,
+      "loss": 0.7793,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011502748895856737,
+      "loss": 0.8003,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011496586069552972,
+      "loss": 0.7993,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011490422661761744,
+      "loss": 0.3572,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011484258674877799,
+      "loss": 0.7932,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011478094111296109,
+      "loss": 0.7979,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011471928973411865,
+      "loss": 0.8105,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011465763263620491,
+      "loss": 0.7455,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011459596984317622,
+      "loss": 0.7725,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011453430137899129,
+      "loss": 0.7437,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011447262726761087,
+      "loss": 0.7832,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011441094753299801,
+      "loss": 0.8079,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011434926219911793,
+      "loss": 0.7998,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011428757128993802,
+      "loss": 0.8064,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011422587482942777,
+      "loss": 0.7532,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011416417284155891,
+      "loss": 0.7922,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011410246535030533,
+      "loss": 0.7925,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011404075237964296,
+      "loss": 0.7947,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011397903395354996,
+      "loss": 0.8154,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011391731009600654,
+      "loss": 0.7444,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001138555808309951,
+      "loss": 0.7839,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011379384618250002,
+      "loss": 0.7729,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011373210617450793,
+      "loss": 0.833,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011367036083100735,
+      "loss": 0.7788,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011360861017598905,
+      "loss": 0.7981,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011354685423344578,
+      "loss": 0.7749,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011348509302737232,
+      "loss": 0.7849,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011342332658176555,
+      "loss": 0.7583,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011336155492062439,
+      "loss": 0.7688,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011329977806794976,
+      "loss": 0.7666,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011323799604774457,
+      "loss": 0.7578,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001131762088840138,
+      "loss": 0.7876,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011311441660076434,
+      "loss": 0.792,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011305261922200519,
+      "loss": 0.7549,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011299081677174722,
+      "loss": 0.7495,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011292900927400333,
+      "loss": 0.7842,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011286719675278838,
+      "loss": 0.7622,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011280537923211918,
+      "loss": 0.7639,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011274355673601444,
+      "loss": 0.7754,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011268172928849486,
+      "loss": 0.8108,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000112619896913583,
+      "loss": 0.7462,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001125580596353034,
+      "loss": 0.759,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011249621747768247,
+      "loss": 0.7566,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011243437046474853,
+      "loss": 0.8186,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011237251862053177,
+      "loss": 0.7783,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001123106619690643,
+      "loss": 0.741,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011224880053438005,
+      "loss": 0.7273,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011218693434051475,
+      "loss": 0.7505,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011212506341150615,
+      "loss": 0.7573,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001120631877713937,
+      "loss": 0.7195,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011200130744421878,
+      "loss": 0.7485,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011193942245402443,
+      "loss": 0.8032,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011187753282485569,
+      "loss": 0.7031,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001118156385807593,
+      "loss": 0.7974,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011175373974578378,
+      "loss": 0.7681,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011169183634397949,
+      "loss": 0.7307,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011162992839939852,
+      "loss": 0.7637,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011156801593609478,
+      "loss": 0.783,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011150609897812387,
+      "loss": 0.7815,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001114441775495432,
+      "loss": 0.7983,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011138225167441186,
+      "loss": 0.7224,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001113203213767907,
+      "loss": 0.4054,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011125838668074225,
+      "loss": 0.7849,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011119644761033078,
+      "loss": 0.7463,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011113450418962233,
+      "loss": 0.3532,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011107255644268449,
+      "loss": 0.7952,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011101060439358659,
+      "loss": 0.755,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011094864806639971,
+      "loss": 0.7629,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011088668748519647,
+      "loss": 0.739,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011082472267405122,
+      "loss": 0.7795,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001107627536570399,
+      "loss": 0.8086,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011070078045824013,
+      "loss": 0.7864,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011063880310173113,
+      "loss": 0.7568,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011057682161159379,
+      "loss": 0.8408,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001105148360119105,
+      "loss": 0.3731,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011045284632676536,
+      "loss": 0.7595,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011039085258024396,
+      "loss": 0.7407,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011032885479643351,
+      "loss": 0.7662,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011026685299942285,
+      "loss": 0.7305,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011020484721330227,
+      "loss": 0.7544,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011014283746216365,
+      "loss": 0.7893,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011008082377010045,
+      "loss": 0.7764,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011001880616120763,
+      "loss": 0.8213,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010995678465958168,
+      "loss": 0.7944,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010989475928932056,
+      "loss": 0.7827,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010983273007452379,
+      "loss": 0.7766,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010977069703929242,
+      "loss": 0.8193,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010970866020772883,
+      "loss": 0.7173,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010964661960393703,
+      "loss": 0.782,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010958457525202241,
+      "loss": 0.7656,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010952252717609188,
+      "loss": 0.3945,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010946047540025372,
+      "loss": 0.7725,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010939841994861772,
+      "loss": 0.7686,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010933636084529506,
+      "loss": 0.7769,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010927429811439837,
+      "loss": 0.7551,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010921223178004164,
+      "loss": 0.7632,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010915016186634026,
+      "loss": 0.751,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010908808839741106,
+      "loss": 0.7673,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010902601139737225,
+      "loss": 0.7463,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010896393089034336,
+      "loss": 0.79,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010890184690044534,
+      "loss": 0.7822,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010883975945180043,
+      "loss": 0.7512,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010877766856853226,
+      "loss": 0.7695,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010871557427476583,
+      "loss": 0.7583,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010865347659462737,
+      "loss": 0.8174,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010859137555224447,
+      "loss": 0.7712,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010852927117174607,
+      "loss": 0.7778,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010846716347726233,
+      "loss": 0.8027,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010840505249292476,
+      "loss": 0.8032,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010834293824286612,
+      "loss": 0.3667,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010828082075122043,
+      "loss": 0.7781,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010821870004212305,
+      "loss": 0.3732,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001081565761397104,
+      "loss": 0.731,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010809444906812033,
+      "loss": 0.8132,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010803231885149189,
+      "loss": 0.8162,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010797018551396527,
+      "loss": 0.8301,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010790804907968193,
+      "loss": 0.7666,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001078459095727845,
+      "loss": 0.7404,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010778376701741689,
+      "loss": 0.7549,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010772162143772407,
+      "loss": 0.7573,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010765947285785227,
+      "loss": 0.7744,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010759732130194887,
+      "loss": 0.7656,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010753516679416235,
+      "loss": 0.7622,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010747300935864243,
+      "loss": 0.7874,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010741084901953995,
+      "loss": 0.7825,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010734868580100677,
+      "loss": 0.8035,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010728651972719603,
+      "loss": 0.7612,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010722435082226185,
+      "loss": 0.7288,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010716217911035951,
+      "loss": 0.7861,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010710000461564535,
+      "loss": 0.7981,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010703782736227687,
+      "loss": 0.7773,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010697564737441252,
+      "loss": 0.7791,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010691346467621192,
+      "loss": 0.7773,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010685127929183567,
+      "loss": 0.3641,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010678909124544547,
+      "loss": 0.7537,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010672690056120399,
+      "loss": 0.7449,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010666470726327496,
+      "loss": 0.7837,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010660251137582315,
+      "loss": 0.7708,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010654031292301432,
+      "loss": 0.7058,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010647811192901518,
+      "loss": 0.7798,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010641590841799351,
+      "loss": 0.7874,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010635370241411802,
+      "loss": 0.7412,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010629149394155838,
+      "loss": 0.7633,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010622928302448523,
+      "loss": 0.8137,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010616706968707015,
+      "loss": 0.729,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010610485395348571,
+      "loss": 0.7969,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010604263584790532,
+      "loss": 0.7688,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010598041539450343,
+      "loss": 0.7607,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010591819261745528,
+      "loss": 0.7791,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010585596754093712,
+      "loss": 0.6833,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010579374018912602,
+      "loss": 0.3752,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010573151058619994,
+      "loss": 0.7974,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010566927875633776,
+      "loss": 0.8093,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010560704472371919,
+      "loss": 0.8201,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001055448085125248,
+      "loss": 0.7834,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010548257014693601,
+      "loss": 0.7844,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010542032965113509,
+      "loss": 0.7313,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001053580870493051,
+      "loss": 0.7505,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010529584236562995,
+      "loss": 0.8418,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001052335956242944,
+      "loss": 0.7856,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010517134684948388,
+      "loss": 0.7473,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010510909606538477,
+      "loss": 0.7576,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010504684329618409,
+      "loss": 0.801,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010498458856606972,
+      "loss": 0.771,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010492233189923029,
+      "loss": 0.8323,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010486007331985515,
+      "loss": 0.7842,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010479781285213439,
+      "loss": 0.7656,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010473555052025892,
+      "loss": 0.7688,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010467328634842024,
+      "loss": 0.749,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010461102036081067,
+      "loss": 0.7495,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001045487525816232,
+      "loss": 0.7363,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010448648303505151,
+      "loss": 0.7458,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010442421174528995,
+      "loss": 0.7341,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010436193873653361,
+      "loss": 0.3708,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010429966403297818,
+      "loss": 0.8164,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010423738765882005,
+      "loss": 0.8171,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010417510963825625,
+      "loss": 0.7783,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010411282999548442,
+      "loss": 0.7366,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010405054875470286,
+      "loss": 0.7783,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001039882659401105,
+      "loss": 0.7944,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010392598157590688,
+      "loss": 0.8047,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010386369568629208,
+      "loss": 0.7893,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001038014082954669,
+      "loss": 0.8149,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001037391194276326,
+      "loss": 0.7317,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010367682910699103,
+      "loss": 0.3599,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010361453735774467,
+      "loss": 0.7346,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010355224420409652,
+      "loss": 0.7976,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010348994967025012,
+      "loss": 0.813,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010342765378040953,
+      "loss": 0.7505,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010336535655877942,
+      "loss": 0.7729,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010330305802956483,
+      "loss": 0.7803,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010324075821697146,
+      "loss": 0.7607,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010317845714520542,
+      "loss": 0.4044,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010311615483847332,
+      "loss": 0.845,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010305385132098229,
+      "loss": 0.8079,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010299154661693987,
+      "loss": 0.7874,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001029292407505541,
+      "loss": 0.772,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010286693374603347,
+      "loss": 0.7678,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001028046256275869,
+      "loss": 0.7883,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010274231641942378,
+      "loss": 0.8198,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010268000614575385,
+      "loss": 0.77,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010261769483078733,
+      "loss": 0.7605,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010255538249873482,
+      "loss": 0.8105,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001024930691738073,
+      "loss": 0.8027,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010243075488021616,
+      "loss": 0.7426,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010236843964217318,
+      "loss": 0.7349,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010230612348389048,
+      "loss": 0.8044,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010224380642958052,
+      "loss": 0.749,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010218148850345613,
+      "loss": 0.7771,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010211916972973048,
+      "loss": 0.7705,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010205685013261712,
+      "loss": 0.8091,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010199452973632981,
+      "loss": 0.7495,
+      "step": 2644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001019322085650827,
+      "loss": 0.717,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010186988664309023,
+      "loss": 0.7427,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001018075639945671,
+      "loss": 0.7831,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010174524064372837,
+      "loss": 0.7325,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010168291661478923,
+      "loss": 0.7593,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001016205919319653,
+      "loss": 0.7268,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010155826661947231,
+      "loss": 0.6992,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010149594070152638,
+      "loss": 0.804,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010143361420234369,
+      "loss": 0.7694,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010137128714614081,
+      "loss": 0.844,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010130895955713445,
+      "loss": 0.74,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010124663145954152,
+      "loss": 0.3923,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010118430287757914,
+      "loss": 0.782,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001011219738354646,
+      "loss": 0.7615,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010105964435741544,
+      "loss": 0.7461,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010099731446764926,
+      "loss": 0.7581,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010093498419038394,
+      "loss": 0.7709,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001008726535498374,
+      "loss": 0.7583,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010081032257022775,
+      "loss": 0.7795,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001007479912757733,
+      "loss": 0.7922,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010068565969069235,
+      "loss": 0.359,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010062332783920336,
+      "loss": 0.7739,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010056099574552497,
+      "loss": 0.7471,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010049866343387581,
+      "loss": 0.8125,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010043633092847466,
+      "loss": 0.7146,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010037399825354035,
+      "loss": 0.7197,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010031166543329178,
+      "loss": 0.8289,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010024933249194792,
+      "loss": 0.7983,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010018699945372775,
+      "loss": 0.772,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010012466634285034,
+      "loss": 0.8284,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010006233318353472,
+      "loss": 0.7571,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001,
+      "loss": 0.7646,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99376668164653e-05,
+      "loss": 0.7114,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987533365714968e-05,
+      "loss": 0.8066,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981300054627226e-05,
+      "loss": 0.7571,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975066750805209e-05,
+      "loss": 0.7415,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968833456670824e-05,
+      "loss": 0.7773,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962600174645968e-05,
+      "loss": 0.7852,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956366907152535e-05,
+      "loss": 0.7683,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950133656612421e-05,
+      "loss": 0.7312,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943900425447505e-05,
+      "loss": 0.781,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937667216079665e-05,
+      "loss": 0.7844,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931434030930768e-05,
+      "loss": 0.8379,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925200872422671e-05,
+      "loss": 0.3845,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918967742977226e-05,
+      "loss": 0.7668,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912734645016263e-05,
+      "loss": 0.8196,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90650158096161e-05,
+      "loss": 0.7834,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900268553235076e-05,
+      "loss": 0.7559,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89403556425846e-05,
+      "loss": 0.7699,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887802616453542e-05,
+      "loss": 0.7297,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881569712242089e-05,
+      "loss": 0.7781,
+      "step": 2695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875336854045851e-05,
+      "loss": 0.7639,
+      "step": 2696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869104044286558e-05,
+      "loss": 0.7585,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862871285385921e-05,
+      "loss": 0.7878,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856638579765632e-05,
+      "loss": 0.7241,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850405929847366e-05,
+      "loss": 0.77,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84417333805277e-05,
+      "loss": 0.7266,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837940806803473e-05,
+      "loss": 0.7944,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831708338521078e-05,
+      "loss": 0.7639,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825475935627165e-05,
+      "loss": 0.7515,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819243600543291e-05,
+      "loss": 0.7168,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813011335690981e-05,
+      "loss": 0.7622,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806779143491733e-05,
+      "loss": 0.7629,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800547026367022e-05,
+      "loss": 0.7451,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794314986738292e-05,
+      "loss": 0.7961,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788083027026953e-05,
+      "loss": 0.7896,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78185114965439e-05,
+      "loss": 0.3796,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775619357041952e-05,
+      "loss": 0.7627,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769387651610955e-05,
+      "loss": 0.3862,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763156035782686e-05,
+      "loss": 0.7937,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756924511978387e-05,
+      "loss": 0.7646,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750693082619273e-05,
+      "loss": 0.7747,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744461750126522e-05,
+      "loss": 0.7532,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73823051692127e-05,
+      "loss": 0.8118,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731999385424618e-05,
+      "loss": 0.7764,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725768358057625e-05,
+      "loss": 0.3726,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719537437241312e-05,
+      "loss": 0.3734,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713306625396658e-05,
+      "loss": 0.7898,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707075924944595e-05,
+      "loss": 0.7651,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700845338306018e-05,
+      "loss": 0.7932,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694614867901775e-05,
+      "loss": 0.7272,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688384516152672e-05,
+      "loss": 0.761,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682154285479462e-05,
+      "loss": 0.7177,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675924178302856e-05,
+      "loss": 0.7905,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669694197043519e-05,
+      "loss": 0.7888,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663464344122064e-05,
+      "loss": 0.7456,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657234621959051e-05,
+      "loss": 0.8223,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651005032974994e-05,
+      "loss": 0.7495,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644775579590353e-05,
+      "loss": 0.717,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638546264225538e-05,
+      "loss": 0.7507,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632317089300902e-05,
+      "loss": 0.7815,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626088057236745e-05,
+      "loss": 0.782,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619859170453314e-05,
+      "loss": 0.7554,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613630431370794e-05,
+      "loss": 0.7732,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607401842409317e-05,
+      "loss": 0.7488,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601173405988953e-05,
+      "loss": 0.7949,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594945124529719e-05,
+      "loss": 0.7588,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588717000451563e-05,
+      "loss": 0.7362,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58248903617438e-05,
+      "loss": 0.7866,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576261234117998e-05,
+      "loss": 0.782,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570033596702184e-05,
+      "loss": 0.8228,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563806126346642e-05,
+      "loss": 0.7859,
+      "step": 2746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557578825471009e-05,
+      "loss": 0.7595,
+      "step": 2747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551351696494854e-05,
+      "loss": 0.761,
+      "step": 2748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545124741837685e-05,
+      "loss": 0.7529,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538897963918935e-05,
+      "loss": 0.7634,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53267136515798e-05,
+      "loss": 0.8044,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526444947974111e-05,
+      "loss": 0.7722,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520218714786563e-05,
+      "loss": 0.76,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513992668014489e-05,
+      "loss": 0.7688,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50776681007697e-05,
+      "loss": 0.7407,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501541143393028e-05,
+      "loss": 0.7096,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495315670381591e-05,
+      "loss": 0.7452,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489090393461524e-05,
+      "loss": 0.3796,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48286531505161e-05,
+      "loss": 0.7402,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476640437570562e-05,
+      "loss": 0.7617,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470415763437004e-05,
+      "loss": 0.7769,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46419129506949e-05,
+      "loss": 0.7205,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457967034886492e-05,
+      "loss": 0.7527,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451742985306398e-05,
+      "loss": 0.7515,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44551914874752e-05,
+      "loss": 0.7629,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439295527628081e-05,
+      "loss": 0.7463,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433072124366223e-05,
+      "loss": 0.741,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426848941380008e-05,
+      "loss": 0.686,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4206259810874e-05,
+      "loss": 0.7434,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414403245906289e-05,
+      "loss": 0.7947,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408180738254471e-05,
+      "loss": 0.751,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401958460549658e-05,
+      "loss": 0.7322,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395736415209468e-05,
+      "loss": 0.7651,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38951460465143e-05,
+      "loss": 0.7546,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383293031292985e-05,
+      "loss": 0.7529,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37707169755148e-05,
+      "loss": 0.7529,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370850605844163e-05,
+      "loss": 0.7114,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3646297585882e-05,
+      "loss": 0.7661,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358409158200648e-05,
+      "loss": 0.7554,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352188807098481e-05,
+      "loss": 0.7405,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345968707698569e-05,
+      "loss": 0.7678,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339748862417686e-05,
+      "loss": 0.7749,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333529273672504e-05,
+      "loss": 0.7783,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327309943879604e-05,
+      "loss": 0.7734,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321090875455455e-05,
+      "loss": 0.7346,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314872070816434e-05,
+      "loss": 0.7513,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308653532378809e-05,
+      "loss": 0.7373,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302435262558747e-05,
+      "loss": 0.7161,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296217263772314e-05,
+      "loss": 0.7568,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289999538435463e-05,
+      "loss": 0.79,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.283782088964049e-05,
+      "loss": 0.782,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277564917773816e-05,
+      "loss": 0.7358,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2713480272804e-05,
+      "loss": 0.7522,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265131419899324e-05,
+      "loss": 0.7654,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258915098046009e-05,
+      "loss": 0.7751,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252699064135758e-05,
+      "loss": 0.7134,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246483320583767e-05,
+      "loss": 0.8079,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240267869805117e-05,
+      "loss": 0.7864,
+      "step": 2798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234052714214776e-05,
+      "loss": 0.7986,
+      "step": 2799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227837856227594e-05,
+      "loss": 0.7432,
+      "step": 2800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221623298258315e-05,
+      "loss": 0.7793,
+      "step": 2801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.215409042721552e-05,
+      "loss": 0.7419,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209195092031811e-05,
+      "loss": 0.7998,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202981448603477e-05,
+      "loss": 0.8162,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196768114850813e-05,
+      "loss": 0.7869,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190555093187967e-05,
+      "loss": 0.7961,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184342386028962e-05,
+      "loss": 0.7295,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178129995787699e-05,
+      "loss": 0.7961,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171917924877959e-05,
+      "loss": 0.7144,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16570617571339e-05,
+      "loss": 0.7322,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159494750707526e-05,
+      "loss": 0.728,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153283652273768e-05,
+      "loss": 0.8213,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147072882825395e-05,
+      "loss": 0.7876,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140862444775554e-05,
+      "loss": 0.7639,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134652340537265e-05,
+      "loss": 0.7878,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128442572523417e-05,
+      "loss": 0.7344,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122233143146776e-05,
+      "loss": 0.3734,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.116024054819959e-05,
+      "loss": 0.7334,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10981530995547e-05,
+      "loss": 0.7617,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103606910965666e-05,
+      "loss": 0.771,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097398860262776e-05,
+      "loss": 0.7722,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091191160258895e-05,
+      "loss": 0.7505,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084983813365978e-05,
+      "loss": 0.759,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078776821995839e-05,
+      "loss": 0.7585,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072570188560164e-05,
+      "loss": 0.8054,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066363915470495e-05,
+      "loss": 0.7847,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06015800513823e-05,
+      "loss": 0.7756,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05395245997463e-05,
+      "loss": 0.7715,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047747282390815e-05,
+      "loss": 0.7844,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04154247479776e-05,
+      "loss": 0.7544,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0353380396063e-05,
+      "loss": 0.7722,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029133979227119e-05,
+      "loss": 0.7744,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022930296070761e-05,
+      "loss": 0.3745,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016726992547622e-05,
+      "loss": 0.7361,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010524071067946e-05,
+      "loss": 0.7026,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004321534041835e-05,
+      "loss": 0.7988,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99811938387924e-05,
+      "loss": 0.7747,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991917622989956e-05,
+      "loss": 0.7371,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985716253783636e-05,
+      "loss": 0.8281,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979515278669775e-05,
+      "loss": 0.741,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973314700057717e-05,
+      "loss": 0.7556,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967114520356651e-05,
+      "loss": 0.7676,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960914741975608e-05,
+      "loss": 0.7883,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-05,
+      "loss": 0.7927,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948516398808952e-05,
+      "loss": 0.739,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942317838840623e-05,
+      "loss": 0.7678,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936119689826888e-05,
+      "loss": 0.7546,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92992195417599e-05,
+      "loss": 0.738,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923724634296012e-05,
+      "loss": 0.7788,
+      "step": 2849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917527732594881e-05,
+      "loss": 0.7393,
+      "step": 2850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911331251480357e-05,
+      "loss": 0.7812,
+      "step": 2851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905135193360033e-05,
+      "loss": 0.7737,
+      "step": 2852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898939560641342e-05,
+      "loss": 0.7832,
+      "step": 2853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892744355731555e-05,
+      "loss": 0.7688,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88654958103777e-05,
+      "loss": 0.7644,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880355238966923e-05,
+      "loss": 0.7908,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874161331925778e-05,
+      "loss": 0.8062,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867967862320934e-05,
+      "loss": 0.7578,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861774832558818e-05,
+      "loss": 0.7444,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855582245045683e-05,
+      "loss": 0.7627,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.849390102187614e-05,
+      "loss": 0.7351,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843198406390524e-05,
+      "loss": 0.7744,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837007160060149e-05,
+      "loss": 0.7839,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830816365602053e-05,
+      "loss": 0.7458,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824626025421626e-05,
+      "loss": 0.8096,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818436141924072e-05,
+      "loss": 0.7461,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812246717514433e-05,
+      "loss": 0.7397,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806057754597558e-05,
+      "loss": 0.7043,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799869255578126e-05,
+      "loss": 0.7585,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79368122286063e-05,
+      "loss": 0.8157,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787493658849386e-05,
+      "loss": 0.771,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781306565948528e-05,
+      "loss": 0.3456,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775119946562e-05,
+      "loss": 0.7859,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768933803093573e-05,
+      "loss": 0.7588,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.762748137946826e-05,
+      "loss": 0.7611,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756562953525152e-05,
+      "loss": 0.7397,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750378252231757e-05,
+      "loss": 0.7537,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744194036469664e-05,
+      "loss": 0.7522,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738010308641704e-05,
+      "loss": 0.8135,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73182707115052e-05,
+      "loss": 0.7886,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725644326398558e-05,
+      "loss": 0.7305,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719462076788084e-05,
+      "loss": 0.7375,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713280324721163e-05,
+      "loss": 0.8127,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70709907259967e-05,
+      "loss": 0.7883,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700918322825282e-05,
+      "loss": 0.7031,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694738077799488e-05,
+      "loss": 0.7424,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68855833992357e-05,
+      "loss": 0.7935,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682379111598627e-05,
+      "loss": 0.7686,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676200395225547e-05,
+      "loss": 0.8003,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670022193205027e-05,
+      "loss": 0.7834,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663844507937562e-05,
+      "loss": 0.7715,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657667341823448e-05,
+      "loss": 0.7056,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651490697262773e-05,
+      "loss": 0.7769,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645314576655428e-05,
+      "loss": 0.7563,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6391389824011e-05,
+      "loss": 0.7444,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632963916899268e-05,
+      "loss": 0.7546,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626789382549212e-05,
+      "loss": 0.3162,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620615381749999e-05,
+      "loss": 0.7542,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614441916900493e-05,
+      "loss": 0.7783,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608268990399349e-05,
+      "loss": 0.7954,
+      "step": 2900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602096604645009e-05,
+      "loss": 0.7869,
+      "step": 2901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595924762035709e-05,
+      "loss": 0.7585,
+      "step": 2902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589753464969472e-05,
+      "loss": 0.7228,
+      "step": 2903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583582715844111e-05,
+      "loss": 0.7983,
+      "step": 2904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577412517057227e-05,
+      "loss": 0.8264,
+      "step": 2905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571242871006202e-05,
+      "loss": 0.761,
+      "step": 2906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565073780088208e-05,
+      "loss": 0.7739,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558905246700201e-05,
+      "loss": 0.7622,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552737273238918e-05,
+      "loss": 0.7512,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546569862100876e-05,
+      "loss": 0.7488,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540403015682381e-05,
+      "loss": 0.739,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534236736379516e-05,
+      "loss": 0.7678,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528071026588137e-05,
+      "loss": 0.7224,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521905888703893e-05,
+      "loss": 0.7461,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515741325122202e-05,
+      "loss": 0.7712,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509577338238255e-05,
+      "loss": 0.7452,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503413930447028e-05,
+      "loss": 0.3915,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497251104143264e-05,
+      "loss": 0.7319,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491088861721486e-05,
+      "loss": 0.7095,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484927205575985e-05,
+      "loss": 0.7642,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478766138100834e-05,
+      "loss": 0.7832,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472605661689862e-05,
+      "loss": 0.7368,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466445778736679e-05,
+      "loss": 0.7512,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460286491634663e-05,
+      "loss": 0.366,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45412780277696e-05,
+      "loss": 0.735,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447969714556484e-05,
+      "loss": 0.7468,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441812229365912e-05,
+      "loss": 0.7341,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435655349597689e-05,
+      "loss": 0.7646,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429499077644029e-05,
+      "loss": 0.7803,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423343415896905e-05,
+      "loss": 0.7859,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417188366748052e-05,
+      "loss": 0.7107,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411033932588967e-05,
+      "loss": 0.7528,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404880115810914e-05,
+      "loss": 0.719,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398726918804911e-05,
+      "loss": 0.7402,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.392574343961738e-05,
+      "loss": 0.7937,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.386422393671933e-05,
+      "loss": 0.7673,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380271070325789e-05,
+      "loss": 0.7854,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374120376313361e-05,
+      "loss": 0.772,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367970314024451e-05,
+      "loss": 0.7803,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361820885848624e-05,
+      "loss": 0.7944,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35567209417519e-05,
+      "loss": 0.7651,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349523941393224e-05,
+      "loss": 0.7769,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.343376429891539e-05,
+      "loss": 0.7585,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337229562058707e-05,
+      "loss": 0.7683,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33108334028305e-05,
+      "loss": 0.8247,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324937766952638e-05,
+      "loss": 0.7703,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318792844455282e-05,
+      "loss": 0.7825,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31264857517855e-05,
+      "loss": 0.7394,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306504961509754e-05,
+      "loss": 0.7817,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300362005835948e-05,
+      "loss": 0.7788,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294219710543932e-05,
+      "loss": 0.7734,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288078078020249e-05,
+      "loss": 0.7903,
+      "step": 2952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281937110651186e-05,
+      "loss": 0.7996,
+      "step": 2953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275796810822773e-05,
+      "loss": 0.748,
+      "step": 2954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269657180920774e-05,
+      "loss": 0.7981,
+      "step": 2955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263518223330697e-05,
+      "loss": 0.7246,
+      "step": 2956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257379940437791e-05,
+      "loss": 0.7539,
+      "step": 2957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251242334627038e-05,
+      "loss": 0.4015,
+      "step": 2958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245105408283161e-05,
+      "loss": 0.7393,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238969163790618e-05,
+      "loss": 0.7563,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2328336035336e-05,
+      "loss": 0.387,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226698729896035e-05,
+      "loss": 0.7517,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220564545261584e-05,
+      "loss": 0.8015,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214431052013634e-05,
+      "loss": 0.3463,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208298252535314e-05,
+      "loss": 0.781,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202166149209474e-05,
+      "loss": 0.7529,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1960347444187e-05,
+      "loss": 0.759,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189904040545301e-05,
+      "loss": 0.3818,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183774039971321e-05,
+      "loss": 0.75,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177644745078526e-05,
+      "loss": 0.7664,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.171516158248406e-05,
+      "loss": 0.7781,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165388281862178e-05,
+      "loss": 0.7803,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159261118300782e-05,
+      "loss": 0.772,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153134669944886e-05,
+      "loss": 0.7419,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147008939174875e-05,
+      "loss": 0.7893,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140883928370855e-05,
+      "loss": 0.7563,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134759639912654e-05,
+      "loss": 0.7466,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12863607617982e-05,
+      "loss": 0.7734,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122513239551619e-05,
+      "loss": 0.7217,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116391132407033e-05,
+      "loss": 0.6776,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11026975712476e-05,
+      "loss": 0.7959,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104149116083216e-05,
+      "loss": 0.424,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098029211660533e-05,
+      "loss": 0.7314,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091910046234552e-05,
+      "loss": 0.7798,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085791622182832e-05,
+      "loss": 0.7729,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07967394188264e-05,
+      "loss": 0.7415,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073557007710957e-05,
+      "loss": 0.7329,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067440822044469e-05,
+      "loss": 0.7832,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061325387259578e-05,
+      "loss": 0.7161,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05521070573239e-05,
+      "loss": 0.7285,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049096779838719e-05,
+      "loss": 0.7908,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042983611954086e-05,
+      "loss": 0.7749,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036871204453717e-05,
+      "loss": 0.7463,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030759559712544e-05,
+      "loss": 0.7278,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0246486801052e-05,
+      "loss": 0.3881,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018538568006027e-05,
+      "loss": 0.8428,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012429225789056e-05,
+      "loss": 0.7983,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00632065582803e-05,
+      "loss": 0.7932,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00021286049639e-05,
+      "loss": 0.7385,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994105842167273e-05,
+      "loss": 0.7883,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987999603213519e-05,
+      "loss": 0.7568,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981894146007658e-05,
+      "loss": 0.7087,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975789472921922e-05,
+      "loss": 0.7693,
+      "step": 3003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96968558632824e-05,
+      "loss": 0.729,
+      "step": 3004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963582488598227e-05,
+      "loss": 0.7722,
+      "step": 3005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957480182103198e-05,
+      "loss": 0.7651,
+      "step": 3006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95137866921416e-05,
+      "loss": 0.7661,
+      "step": 3007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94527795230181e-05,
+      "loss": 0.7629,
+      "step": 3008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939178033736538e-05,
+      "loss": 0.7422,
+      "step": 3009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93307891588842e-05,
+      "loss": 0.3484,
+      "step": 3010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926980601127225e-05,
+      "loss": 0.7415,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920883091822408e-05,
+      "loss": 0.7585,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914786390343113e-05,
+      "loss": 0.7598,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908690499058161e-05,
+      "loss": 0.7886,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902595420336073e-05,
+      "loss": 0.7888,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896501156545045e-05,
+      "loss": 0.8,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890407710052956e-05,
+      "loss": 0.7888,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884315083227373e-05,
+      "loss": 0.7781,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878223278435538e-05,
+      "loss": 0.3851,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872132298044382e-05,
+      "loss": 0.3648,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866042144420502e-05,
+      "loss": 0.7522,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859952819930188e-05,
+      "loss": 0.7773,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853864326939402e-05,
+      "loss": 0.7686,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84777666781378e-05,
+      "loss": 0.792,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841689844918642e-05,
+      "loss": 0.7466,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835603860618972e-05,
+      "loss": 0.7468,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829518717279438e-05,
+      "loss": 0.7944,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823434417264378e-05,
+      "loss": 0.7759,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817350962937802e-05,
+      "loss": 0.7644,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.811268356663387e-05,
+      "loss": 0.7595,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805186600804489e-05,
+      "loss": 0.7656,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799105697724126e-05,
+      "loss": 0.7437,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79302564978499e-05,
+      "loss": 0.7271,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78694645934944e-05,
+      "loss": 0.7874,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780868128779499e-05,
+      "loss": 0.7402,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436858e-05,
+      "loss": 0.7205,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768714056682872e-05,
+      "loss": 0.7883,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76263831987856e-05,
+      "loss": 0.7292,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756563452384606e-05,
+      "loss": 0.696,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750489456561352e-05,
+      "loss": 0.8208,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.744416334768808e-05,
+      "loss": 0.7361,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738344089366638e-05,
+      "loss": 0.7842,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73227272271417e-05,
+      "loss": 0.7905,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726202237170387e-05,
+      "loss": 0.7959,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720132635093934e-05,
+      "loss": 0.7864,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714063918843106e-05,
+      "loss": 0.769,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707996090775861e-05,
+      "loss": 0.7993,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701929153249807e-05,
+      "loss": 0.8042,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695863108622209e-05,
+      "loss": 0.7258,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689797959249984e-05,
+      "loss": 0.6858,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683733707489699e-05,
+      "loss": 0.804,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677670355697577e-05,
+      "loss": 0.803,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671607906229488e-05,
+      "loss": 0.7737,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66554636144095e-05,
+      "loss": 0.6863,
+      "step": 3054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659485723687131e-05,
+      "loss": 0.7169,
+      "step": 3055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653425995322851e-05,
+      "loss": 0.7173,
+      "step": 3056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647367178702568e-05,
+      "loss": 0.7505,
+      "step": 3057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641309276180394e-05,
+      "loss": 0.7412,
+      "step": 3058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635252290110081e-05,
+      "loss": 0.7212,
+      "step": 3059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629196222845026e-05,
+      "loss": 0.726,
+      "step": 3060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623141076738271e-05,
+      "loss": 0.7078,
+      "step": 3061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617086854142498e-05,
+      "loss": 0.7927,
+      "step": 3062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611033557410026e-05,
+      "loss": 0.3889,
+      "step": 3063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604981188892823e-05,
+      "loss": 0.7996,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598929750942491e-05,
+      "loss": 0.7427,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592879245910273e-05,
+      "loss": 0.7966,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586829676147045e-05,
+      "loss": 0.7961,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580781044003324e-05,
+      "loss": 0.7756,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574733351829263e-05,
+      "loss": 0.7734,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568686601974644e-05,
+      "loss": 0.7883,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562640796788892e-05,
+      "loss": 0.7051,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556595938621058e-05,
+      "loss": 0.7566,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550552029819827e-05,
+      "loss": 0.7517,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544509072733514e-05,
+      "loss": 0.7832,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53846706971007e-05,
+      "loss": 0.7345,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532426023097063e-05,
+      "loss": 0.7556,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526385935241703e-05,
+      "loss": 0.7393,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.520346808490817e-05,
+      "loss": 0.8123,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514308645190867e-05,
+      "loss": 0.7606,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508271447687936e-05,
+      "loss": 0.7932,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502235218327731e-05,
+      "loss": 0.8101,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.496199959455584e-05,
+      "loss": 0.7656,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490165673416455e-05,
+      "loss": 0.7744,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484132362554915e-05,
+      "loss": 0.7666,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478100029215165e-05,
+      "loss": 0.7688,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.472068675741025e-05,
+      "loss": 0.7891,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46603830447593e-05,
+      "loss": 0.7927,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460008917762939e-05,
+      "loss": 0.7539,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453980517944723e-05,
+      "loss": 0.7549,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447953107363574e-05,
+      "loss": 0.7673,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4419266883614e-05,
+      "loss": 0.7668,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435901263279716e-05,
+      "loss": 0.7834,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429876834459662e-05,
+      "loss": 0.739,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42385340424198e-05,
+      "loss": 0.3915,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417830974967034e-05,
+      "loss": 0.7622,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411809548974792e-05,
+      "loss": 0.7312,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405789128604836e-05,
+      "loss": 0.7893,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399769716196352e-05,
+      "loss": 0.7124,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393751314088146e-05,
+      "loss": 0.8,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387733924618617e-05,
+      "loss": 0.761,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38171755012578e-05,
+      "loss": 0.7302,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37570219294725e-05,
+      "loss": 0.7759,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.369687855420253e-05,
+      "loss": 0.7927,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363674539881613e-05,
+      "loss": 0.3834,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35766224866776e-05,
+      "loss": 0.3542,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351650984114728e-05,
+      "loss": 0.7358,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345640748558146e-05,
+      "loss": 0.7454,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339631544333249e-05,
+      "loss": 0.7664,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333623373774867e-05,
+      "loss": 0.8066,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.327616239217431e-05,
+      "loss": 0.7817,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32161014299497e-05,
+      "loss": 0.8079,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315605087441106e-05,
+      "loss": 0.7217,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309601074889061e-05,
+      "loss": 0.7134,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303598107671647e-05,
+      "loss": 0.7617,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297596188121277e-05,
+      "loss": 0.78,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291595318569951e-05,
+      "loss": 0.7495,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285595501349258e-05,
+      "loss": 0.7612,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279596738790388e-05,
+      "loss": 0.7876,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273599033224114e-05,
+      "loss": 0.7578,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2676023869808e-05,
+      "loss": 0.7244,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2616068023904e-05,
+      "loss": 0.7834,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255612281782452e-05,
+      "loss": 0.67,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249618827486084e-05,
+      "loss": 0.7559,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243626441830009e-05,
+      "loss": 0.7363,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237635127142524e-05,
+      "loss": 0.8069,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231644885751507e-05,
+      "loss": 0.7803,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225655719984426e-05,
+      "loss": 0.7793,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219667632168325e-05,
+      "loss": 0.7229,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213680624629832e-05,
+      "loss": 0.748,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207694699695154e-05,
+      "loss": 0.7549,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20170985969008e-05,
+      "loss": 0.7131,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195726106939974e-05,
+      "loss": 0.8096,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189743443769778e-05,
+      "loss": 0.7734,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183761872504011e-05,
+      "loss": 0.688,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177781395466772e-05,
+      "loss": 0.7495,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171802014981726e-05,
+      "loss": 0.7319,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165823733372121e-05,
+      "loss": 0.7667,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159846552960774e-05,
+      "loss": 0.7776,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153870476070073e-05,
+      "loss": 0.7908,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14789550502198e-05,
+      "loss": 0.7424,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.141921642138025e-05,
+      "loss": 0.7698,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135948889739306e-05,
+      "loss": 0.7024,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.129977250146496e-05,
+      "loss": 0.7703,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124006725679828e-05,
+      "loss": 0.7961,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118037318659108e-05,
+      "loss": 0.7266,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112069031403704e-05,
+      "loss": 0.7827,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106101866232546e-05,
+      "loss": 0.7603,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100135825464139e-05,
+      "loss": 0.7839,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094170911416543e-05,
+      "loss": 0.7322,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088207126407373e-05,
+      "loss": 0.8264,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082244472753822e-05,
+      "loss": 0.8174,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076282952772633e-05,
+      "loss": 0.7444,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07032256878011e-05,
+      "loss": 0.8132,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064363323092117e-05,
+      "loss": 0.7322,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058405218024073e-05,
+      "loss": 0.7761,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.052448255890957e-05,
+      "loss": 0.7468,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046492439007307e-05,
+      "loss": 0.7678,
+      "step": 3157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040537769687203e-05,
+      "loss": 0.7288,
+      "step": 3158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.034584250244291e-05,
+      "loss": 0.8127,
+      "step": 3159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02863188299177e-05,
+      "loss": 0.802,
+      "step": 3160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022680670242387e-05,
+      "loss": 0.7996,
+      "step": 3161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01673061430844e-05,
+      "loss": 0.7639,
+      "step": 3162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01078171750178e-05,
+      "loss": 0.7412,
+      "step": 3163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004833982133808e-05,
+      "loss": 0.7688,
+      "step": 3164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.998887410515472e-05,
+      "loss": 0.7014,
+      "step": 3165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992942004957271e-05,
+      "loss": 0.7759,
+      "step": 3166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986997767769243e-05,
+      "loss": 0.8032,
+      "step": 3167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.981054701260981e-05,
+      "loss": 0.7402,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975112807741619e-05,
+      "loss": 0.7703,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969172089519835e-05,
+      "loss": 0.7603,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963232548903853e-05,
+      "loss": 0.7971,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957294188201438e-05,
+      "loss": 0.7388,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951357009719892e-05,
+      "loss": 0.7583,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94542101576607e-05,
+      "loss": 0.7737,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939486208646349e-05,
+      "loss": 0.7295,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.933552590666659e-05,
+      "loss": 0.7627,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927620164132463e-05,
+      "loss": 0.7747,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921688931348762e-05,
+      "loss": 0.3682,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915758894620092e-05,
+      "loss": 0.8091,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.7915,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903902418543671e-05,
+      "loss": 0.7571,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897975983802666e-05,
+      "loss": 0.7151,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892050754330182e-05,
+      "loss": 0.7991,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886126732428424e-05,
+      "loss": 0.7703,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.880203920399127e-05,
+      "loss": 0.761,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874282320543557e-05,
+      "loss": 0.7671,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868361935162506e-05,
+      "loss": 0.7329,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862442766556297e-05,
+      "loss": 0.6885,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85652481702478e-05,
+      "loss": 0.7068,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850608088867332e-05,
+      "loss": 0.7495,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844692584382848e-05,
+      "loss": 0.7322,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838778305869759e-05,
+      "loss": 0.7598,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832865255626013e-05,
+      "loss": 0.7651,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826953435949081e-05,
+      "loss": 0.7537,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821042849135957e-05,
+      "loss": 0.7493,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815133497483157e-05,
+      "loss": 0.7988,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809225383286714e-05,
+      "loss": 0.7944,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803318508842187e-05,
+      "loss": 0.7683,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797412876444645e-05,
+      "loss": 0.7292,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791508488388675e-05,
+      "loss": 0.7825,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785605346968386e-05,
+      "loss": 0.8276,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779703454477401e-05,
+      "loss": 0.7498,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773802813208856e-05,
+      "loss": 0.7634,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767903425455401e-05,
+      "loss": 0.7869,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762005293509201e-05,
+      "loss": 0.7283,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756108419661931e-05,
+      "loss": 0.7461,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750212806204777e-05,
+      "loss": 0.7446,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744318455428436e-05,
+      "loss": 0.7344,
+      "step": 3208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738425369623114e-05,
+      "loss": 0.7933,
+      "step": 3209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732533551078528e-05,
+      "loss": 0.7847,
+      "step": 3210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7266430020839e-05,
+      "loss": 0.7664,
+      "step": 3211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720753724927958e-05,
+      "loss": 0.3457,
+      "step": 3212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714865721898937e-05,
+      "loss": 0.7595,
+      "step": 3213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70897899528458e-05,
+      "loss": 0.7981,
+      "step": 3214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70309354737213e-05,
+      "loss": 0.7251,
+      "step": 3215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697209380448333e-05,
+      "loss": 0.7393,
+      "step": 3216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691326496799436e-05,
+      "loss": 0.8083,
+      "step": 3217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685444898711196e-05,
+      "loss": 0.7424,
+      "step": 3218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67956458846886e-05,
+      "loss": 0.7729,
+      "step": 3219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673685568357182e-05,
+      "loss": 0.7812,
+      "step": 3220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66780784066041e-05,
+      "loss": 0.7239,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661931407662292e-05,
+      "loss": 0.6989,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656056271646076e-05,
+      "loss": 0.77,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650182434894496e-05,
+      "loss": 0.7705,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644309899689793e-05,
+      "loss": 0.7463,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638438668313695e-05,
+      "loss": 0.7634,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632568743047427e-05,
+      "loss": 0.7764,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626700126171702e-05,
+      "loss": 0.7505,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62083281996673e-05,
+      "loss": 0.7651,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614966826712212e-05,
+      "loss": 0.7812,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609102148687333e-05,
+      "loss": 0.7253,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603238788170772e-05,
+      "loss": 0.7605,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597376747440693e-05,
+      "loss": 0.7744,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59151602877475e-05,
+      "loss": 0.7981,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585656634450081e-05,
+      "loss": 0.7476,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579798566743314e-05,
+      "loss": 0.3365,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573941827930551e-05,
+      "loss": 0.76,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568086420287389e-05,
+      "loss": 0.7876,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562232346088902e-05,
+      "loss": 0.3702,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556379607609648e-05,
+      "loss": 0.7349,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550528207123668e-05,
+      "loss": 0.752,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544678146904474e-05,
+      "loss": 0.7444,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538829429225069e-05,
+      "loss": 0.7394,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532982056357928e-05,
+      "loss": 0.3813,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527136030575004e-05,
+      "loss": 0.8245,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521291354147727e-05,
+      "loss": 0.7817,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515448029347004e-05,
+      "loss": 0.7737,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509606058443213e-05,
+      "loss": 0.7986,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503765443706211e-05,
+      "loss": 0.7366,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497926187405326e-05,
+      "loss": 0.7651,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492088291809354e-05,
+      "loss": 0.7588,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486251759186572e-05,
+      "loss": 0.7512,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48041659180472e-05,
+      "loss": 0.7241,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474582791931006e-05,
+      "loss": 0.7808,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.468750361832112e-05,
+      "loss": 0.873,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462919303774186e-05,
+      "loss": 0.7192,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457089620022841e-05,
+      "loss": 0.7927,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451261312843159e-05,
+      "loss": 0.759,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445434384499687e-05,
+      "loss": 0.7461,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439608837256432e-05,
+      "loss": 0.7314,
+      "step": 3260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43378467337687e-05,
+      "loss": 0.7456,
+      "step": 3261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427961895123935e-05,
+      "loss": 0.7349,
+      "step": 3262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422140504760024e-05,
+      "loss": 0.7354,
+      "step": 3263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416320504546997e-05,
+      "loss": 0.7698,
+      "step": 3264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410501896746171e-05,
+      "loss": 0.7535,
+      "step": 3265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404684683618325e-05,
+      "loss": 0.7133,
+      "step": 3266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398868867423691e-05,
+      "loss": 0.7495,
+      "step": 3267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393054450421963e-05,
+      "loss": 0.8032,
+      "step": 3268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387241434872294e-05,
+      "loss": 0.7483,
+      "step": 3269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38142982303328e-05,
+      "loss": 0.8115,
+      "step": 3270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375619617162985e-05,
+      "loss": 0.7058,
+      "step": 3271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36981081951892e-05,
+      "loss": 0.7678,
+      "step": 3272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364003432358051e-05,
+      "loss": 0.7717,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358197457936796e-05,
+      "loss": 0.748,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352392898511021e-05,
+      "loss": 0.7314,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34658975633605e-05,
+      "loss": 0.8123,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340788033666645e-05,
+      "loss": 0.3435,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334987732757029e-05,
+      "loss": 0.7368,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32918885586086e-05,
+      "loss": 0.7297,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323391405231249e-05,
+      "loss": 0.7583,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317595383120756e-05,
+      "loss": 0.749,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311800791781381e-05,
+      "loss": 0.7507,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30600763346457e-05,
+      "loss": 0.401,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300215910421212e-05,
+      "loss": 0.7921,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294425624901638e-05,
+      "loss": 0.7532,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288636779155621e-05,
+      "loss": 0.7585,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282849375432374e-05,
+      "loss": 0.7881,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277063415980549e-05,
+      "loss": 0.7617,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27127890304824e-05,
+      "loss": 0.7749,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265495838882974e-05,
+      "loss": 0.3671,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259714225731718e-05,
+      "loss": 0.3643,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25393406584088e-05,
+      "loss": 0.8154,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248155361456292e-05,
+      "loss": 0.7942,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242378114823234e-05,
+      "loss": 0.7463,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236602328186405e-05,
+      "loss": 0.7173,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230828003789949e-05,
+      "loss": 0.7781,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225055143877434e-05,
+      "loss": 0.7622,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.219283750691864e-05,
+      "loss": 0.7881,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213513826475671e-05,
+      "loss": 0.759,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207745373470716e-05,
+      "loss": 0.7551,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20197839391829e-05,
+      "loss": 0.7444,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196212890059112e-05,
+      "loss": 0.7673,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19044886413332e-05,
+      "loss": 0.8058,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184686318380489e-05,
+      "loss": 0.7634,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17892525503961e-05,
+      "loss": 0.7678,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 0.7844,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167407584546811e-05,
+      "loss": 0.7524,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161650981869998e-05,
+      "loss": 0.7197,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155895870555348e-05,
+      "loss": 0.7253,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150142252838966e-05,
+      "loss": 0.7864,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144390130956384e-05,
+      "loss": 0.7358,
+      "step": 3311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138639507142539e-05,
+      "loss": 0.731,
+      "step": 3312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.132890383631796e-05,
+      "loss": 0.689,
+      "step": 3313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127142762657933e-05,
+      "loss": 0.7715,
+      "step": 3314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121396646454149e-05,
+      "loss": 0.7346,
+      "step": 3315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115652037253053e-05,
+      "loss": 0.7388,
+      "step": 3316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109908937286672e-05,
+      "loss": 0.8154,
+      "step": 3317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104167348786442e-05,
+      "loss": 0.7422,
+      "step": 3318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.09842727398322e-05,
+      "loss": 0.7314,
+      "step": 3319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092688715107264e-05,
+      "loss": 0.7433,
+      "step": 3320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0869516743882516e-05,
+      "loss": 0.6832,
+      "step": 3321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081216154055267e-05,
+      "loss": 0.7317,
+      "step": 3322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075482156336804e-05,
+      "loss": 0.7678,
+      "step": 3323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069749683460765e-05,
+      "loss": 0.8076,
+      "step": 3324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06401873765446e-05,
+      "loss": 0.7437,
+      "step": 3325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058289321144608e-05,
+      "loss": 0.7561,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052561436157329e-05,
+      "loss": 0.7227,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046835084918151e-05,
+      "loss": 0.7734,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041110269652005e-05,
+      "loss": 0.7671,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.035386992583225e-05,
+      "loss": 0.7637,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02966525593555e-05,
+      "loss": 0.7811,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023945061932119e-05,
+      "loss": 0.7781,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018226412795469e-05,
+      "loss": 0.7612,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012509310747538e-05,
+      "loss": 0.7766,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00679375800967e-05,
+      "loss": 0.4128,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001079756802592e-05,
+      "loss": 0.7869,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995367309346442e-05,
+      "loss": 0.7384,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9896564178607495e-05,
+      "loss": 0.7615,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983947084564438e-05,
+      "loss": 0.7622,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978239311675826e-05,
+      "loss": 0.7593,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9725331014126294e-05,
+      "loss": 0.7878,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9668284559919504e-05,
+      "loss": 0.781,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96112537763029e-05,
+      "loss": 0.7639,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955423868543537e-05,
+      "loss": 0.8159,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9497239309469664e-05,
+      "loss": 0.7493,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944025567055251e-05,
+      "loss": 0.7668,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938328779082447e-05,
+      "loss": 0.7244,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9326335692419995e-05,
+      "loss": 0.7603,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9269399397467394e-05,
+      "loss": 0.7285,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921247892808885e-05,
+      "loss": 0.7483,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9155574306400395e-05,
+      "loss": 0.3272,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909868555451191e-05,
+      "loss": 0.6801,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9041812694527064e-05,
+      "loss": 0.7205,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8984955748543414e-05,
+      "loss": 0.7622,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89281147386523e-05,
+      "loss": 0.7383,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887128968693887e-05,
+      "loss": 0.7625,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881448061548208e-05,
+      "loss": 0.7588,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875768754635469e-05,
+      "loss": 0.791,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8700910501623204e-05,
+      "loss": 0.7837,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864414950334796e-05,
+      "loss": 0.7332,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858740457358298e-05,
+      "loss": 0.7648,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853067573437612e-05,
+      "loss": 0.7476,
+      "step": 3362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847396300776893e-05,
+      "loss": 0.7803,
+      "step": 3363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8417266415796745e-05,
+      "loss": 0.7549,
+      "step": 3364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836058598048859e-05,
+      "loss": 0.7563,
+      "step": 3365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830392172386723e-05,
+      "loss": 0.792,
+      "step": 3366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824727366794914e-05,
+      "loss": 0.78,
+      "step": 3367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.819064183474451e-05,
+      "loss": 0.7354,
+      "step": 3368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8134026246257225e-05,
+      "loss": 0.792,
+      "step": 3369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807742692448483e-05,
+      "loss": 0.7332,
+      "step": 3370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802084389141862e-05,
+      "loss": 0.7251,
+      "step": 3371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796427716904347e-05,
+      "loss": 0.7939,
+      "step": 3372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790772677933799e-05,
+      "loss": 0.745,
+      "step": 3373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78511927442744e-05,
+      "loss": 0.781,
+      "step": 3374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7794675085818596e-05,
+      "loss": 0.7253,
+      "step": 3375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773817382593008e-05,
+      "loss": 0.7651,
+      "step": 3376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7681688986562e-05,
+      "loss": 0.7815,
+      "step": 3377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762522058966113e-05,
+      "loss": 0.7389,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756876865716787e-05,
+      "loss": 0.7305,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751233321101617e-05,
+      "loss": 0.7329,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7455914273133646e-05,
+      "loss": 0.7917,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73995118654414e-05,
+      "loss": 0.7876,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7343126009854206e-05,
+      "loss": 0.7422,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7286756728280366e-05,
+      "loss": 0.7656,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72304040426217e-05,
+      "loss": 0.7178,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717406797477372e-05,
+      "loss": 0.8032,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711774854662535e-05,
+      "loss": 0.7141,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7061445780059074e-05,
+      "loss": 0.7859,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7005159696950936e-05,
+      "loss": 0.708,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694889031917047e-05,
+      "loss": 0.76,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6892637668580716e-05,
+      "loss": 0.7334,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683640176703824e-05,
+      "loss": 0.7722,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6780182636393085e-05,
+      "loss": 0.7424,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672398029848881e-05,
+      "loss": 0.8064,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666779477516234e-05,
+      "loss": 0.7051,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6611626088244194e-05,
+      "loss": 0.7117,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655547425955829e-05,
+      "loss": 0.74,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649933931092202e-05,
+      "loss": 0.7366,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6443221264146194e-05,
+      "loss": 0.7539,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638712014103507e-05,
+      "loss": 0.6882,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6331035963386315e-05,
+      "loss": 0.7634,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627496875299102e-05,
+      "loss": 0.7961,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6218918531633704e-05,
+      "loss": 0.7107,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616288532109225e-05,
+      "loss": 0.7275,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610686914313796e-05,
+      "loss": 0.7705,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6050870019535494e-05,
+      "loss": 0.7335,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599488797204291e-05,
+      "loss": 0.802,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5938923022411615e-05,
+      "loss": 0.7876,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588297519238637e-05,
+      "loss": 0.7664,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5827044503705326e-05,
+      "loss": 0.696,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577113097809989e-05,
+      "loss": 0.7322,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571523463729487e-05,
+      "loss": 0.7832,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565935550300836e-05,
+      "loss": 0.7255,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560349359695181e-05,
+      "loss": 0.7488,
+      "step": 3414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5547648940829935e-05,
+      "loss": 0.7888,
+      "step": 3415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549182155634076e-05,
+      "loss": 0.7993,
+      "step": 3416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5436011465175634e-05,
+      "loss": 0.7761,
+      "step": 3417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5380218689019125e-05,
+      "loss": 0.7195,
+      "step": 3418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5324443249549106e-05,
+      "loss": 0.7412,
+      "step": 3419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526868516843673e-05,
+      "loss": 0.4083,
+      "step": 3420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521294446734637e-05,
+      "loss": 0.7454,
+      "step": 3421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515722116793569e-05,
+      "loss": 0.793,
+      "step": 3422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510151529185553e-05,
+      "loss": 0.7168,
+      "step": 3423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5045826860750014e-05,
+      "loss": 0.3776,
+      "step": 3424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499015589625649e-05,
+      "loss": 0.7781,
+      "step": 3425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493450242000546e-05,
+      "loss": 0.7383,
+      "step": 3426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487886645362075e-05,
+      "loss": 0.7324,
+      "step": 3427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482324801871919e-05,
+      "loss": 0.7251,
+      "step": 3428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476764713691098e-05,
+      "loss": 0.7341,
+      "step": 3429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471206382979941e-05,
+      "loss": 0.7627,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465649811898098e-05,
+      "loss": 0.7211,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4600950026045326e-05,
+      "loss": 0.7463,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4545419572575265e-05,
+      "loss": 0.7197,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448990678014672e-05,
+      "loss": 0.3978,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44344116703288e-05,
+      "loss": 0.4205,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43789342646837e-05,
+      "loss": 0.7644,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432347458476678e-05,
+      "loss": 0.7712,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426803265212647e-05,
+      "loss": 0.7495,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4212608488304315e-05,
+      "loss": 0.7642,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415720211483499e-05,
+      "loss": 0.78,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410181355324622e-05,
+      "loss": 0.7485,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404644282505882e-05,
+      "loss": 0.7478,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399108995178672e-05,
+      "loss": 0.7588,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393575495493679e-05,
+      "loss": 0.7317,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388043785600907e-05,
+      "loss": 0.7466,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382513867649663e-05,
+      "loss": 0.7639,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376985743788552e-05,
+      "loss": 0.7739,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3714594161654896e-05,
+      "loss": 0.8464,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3659348869276874e-05,
+      "loss": 0.7207,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36041215822166e-05,
+      "loss": 0.7686,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3548912321932244e-05,
+      "loss": 0.7986,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349372110987496e-05,
+      "loss": 0.8333,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343854796748886e-05,
+      "loss": 0.7253,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338339291621108e-05,
+      "loss": 0.7334,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.332825597747171e-05,
+      "loss": 0.731,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32731371726938e-05,
+      "loss": 0.7424,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321803652329333e-05,
+      "loss": 0.7278,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316295405067929e-05,
+      "loss": 0.7493,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3107889776253536e-05,
+      "loss": 0.7563,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305284372141095e-05,
+      "loss": 0.3634,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2997815907539164e-05,
+      "loss": 0.7415,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294280635601889e-05,
+      "loss": 0.7712,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288781508822367e-05,
+      "loss": 0.7417,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2832842125519964e-05,
+      "loss": 0.7415,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277788748926711e-05,
+      "loss": 0.7964,
+      "step": 3465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272295120081732e-05,
+      "loss": 0.75,
+      "step": 3466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266803328151567e-05,
+      "loss": 0.7634,
+      "step": 3467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261313375270014e-05,
+      "loss": 0.7021,
+      "step": 3468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255825263570153e-05,
+      "loss": 0.7263,
+      "step": 3469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2503389951843475e-05,
+      "loss": 0.7588,
+      "step": 3470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2448545722442486e-05,
+      "loss": 0.7651,
+      "step": 3471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2393719968807864e-05,
+      "loss": 0.72,
+      "step": 3472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233891271224178e-05,
+      "loss": 0.7524,
+      "step": 3473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2284123974039154e-05,
+      "loss": 0.7307,
+      "step": 3474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2229353775487764e-05,
+      "loss": 0.3967,
+      "step": 3475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217460213786821e-05,
+      "loss": 0.7441,
+      "step": 3476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211986908245374e-05,
+      "loss": 0.7146,
+      "step": 3477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2065154630510535e-05,
+      "loss": 0.6985,
+      "step": 3478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201045880329749e-05,
+      "loss": 0.7236,
+      "step": 3479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195578162206627e-05,
+      "loss": 0.7352,
+      "step": 3480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190112310806126e-05,
+      "loss": 0.7678,
+      "step": 3481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184648328251964e-05,
+      "loss": 0.7759,
+      "step": 3482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179186216667132e-05,
+      "loss": 0.8069,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1737259781738936e-05,
+      "loss": 0.7715,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16826761489378e-05,
+      "loss": 0.3677,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162811128947602e-05,
+      "loss": 0.7797,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1573565224554365e-05,
+      "loss": 0.7583,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15190379753663e-05,
+      "loss": 0.7971,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146452956309798e-05,
+      "loss": 0.7319,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.141004000892824e-05,
+      "loss": 0.7793,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135556933402862e-05,
+      "loss": 0.7516,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130111755956327e-05,
+      "loss": 0.7358,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1246684706689076e-05,
+      "loss": 0.3598,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119227079655544e-05,
+      "loss": 0.3738,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113787585030454e-05,
+      "loss": 0.7183,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108349988907111e-05,
+      "loss": 0.7639,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102914293398252e-05,
+      "loss": 0.7478,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097480500615879e-05,
+      "loss": 0.7454,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092048612671252e-05,
+      "loss": 0.7537,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086618631674888e-05,
+      "loss": 0.7544,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0811905597365684e-05,
+      "loss": 0.7705,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07576439896533e-05,
+      "loss": 0.688,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070340151469468e-05,
+      "loss": 0.7552,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0649178193565314e-05,
+      "loss": 0.7078,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059497404733331e-05,
+      "loss": 0.6863,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054078909705926e-05,
+      "loss": 0.7188,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048662336379635e-05,
+      "loss": 0.7552,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043247686859024e-05,
+      "loss": 0.7762,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0378349632479216e-05,
+      "loss": 0.7078,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032424167649394e-05,
+      "loss": 0.7397,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0270153021657674e-05,
+      "loss": 0.6869,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02160836889862e-05,
+      "loss": 0.7029,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016203369948773e-05,
+      "loss": 0.7781,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0108003074163e-05,
+      "loss": 0.3604,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005399183400521e-05,
+      "loss": 0.7534,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.7712,
+      "step": 3516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9946027593125565e-05,
+      "loss": 0.771,
+      "step": 3517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989207463435244e-05,
+      "loss": 0.7371,
+      "step": 3518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9838141144643634e-05,
+      "loss": 0.7178,
+      "step": 3519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978422714495464e-05,
+      "loss": 0.7979,
+      "step": 3520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973033265623332e-05,
+      "loss": 0.7288,
+      "step": 3521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967645769942e-05,
+      "loss": 0.7896,
+      "step": 3522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962260229544738e-05,
+      "loss": 0.6953,
+      "step": 3523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956876646524059e-05,
+      "loss": 0.7615,
+      "step": 3524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951495022971713e-05,
+      "loss": 0.7419,
+      "step": 3525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946115360978696e-05,
+      "loss": 0.6863,
+      "step": 3526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940737662635227e-05,
+      "loss": 0.7175,
+      "step": 3527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935361930030774e-05,
+      "loss": 0.6772,
+      "step": 3528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92998816525404e-05,
+      "loss": 0.7576,
+      "step": 3529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924616370392961e-05,
+      "loss": 0.6906,
+      "step": 3530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919246547534708e-05,
+      "loss": 0.7427,
+      "step": 3531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913878698765686e-05,
+      "loss": 0.3599,
+      "step": 3532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908512826171533e-05,
+      "loss": 0.7617,
+      "step": 3533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9031489318371194e-05,
+      "loss": 0.7544,
+      "step": 3534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8977870178465467e-05,
+      "loss": 0.7468,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892427086283147e-05,
+      "loss": 0.783,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887069139229481e-05,
+      "loss": 0.812,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881713178767341e-05,
+      "loss": 0.7703,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8763592069777456e-05,
+      "loss": 0.3628,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8710072259409386e-05,
+      "loss": 0.3838,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8656572377363974e-05,
+      "loss": 0.7507,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860309244442821e-05,
+      "loss": 0.7056,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854963248138121e-05,
+      "loss": 0.7615,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8496192508994576e-05,
+      "loss": 0.7783,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844277254803198e-05,
+      "loss": 0.3672,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8389372619249326e-05,
+      "loss": 0.7197,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83359927433948e-05,
+      "loss": 0.7773,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8282632941208725e-05,
+      "loss": 0.771,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822929323342369e-05,
+      "loss": 0.7588,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8175973640764406e-05,
+      "loss": 0.6843,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8122674183947836e-05,
+      "loss": 0.8052,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806939488368308e-05,
+      "loss": 0.7378,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801613576067143e-05,
+      "loss": 0.7097,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796289683560631e-05,
+      "loss": 0.7969,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7909678129173344e-05,
+      "loss": 0.7517,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78564796620502e-05,
+      "loss": 0.7346,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78033014549068e-05,
+      "loss": 0.7659,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7750143528405126e-05,
+      "loss": 0.7595,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.76970059031993e-05,
+      "loss": 0.3494,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764388859993556e-05,
+      "loss": 0.71,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.759079163925223e-05,
+      "loss": 0.7468,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753771504177976e-05,
+      "loss": 0.7131,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748465882814066e-05,
+      "loss": 0.6873,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743162301894952e-05,
+      "loss": 0.7854,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7378607634813043e-05,
+      "loss": 0.7018,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732561269632992e-05,
+      "loss": 0.7693,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727263822409099e-05,
+      "loss": 0.6926,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7219684238679066e-05,
+      "loss": 0.7361,
+      "step": 3568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716675076066902e-05,
+      "loss": 0.759,
+      "step": 3569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7113837810627794e-05,
+      "loss": 0.7366,
+      "step": 3570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706094540911429e-05,
+      "loss": 0.7153,
+      "step": 3571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700807357667952e-05,
+      "loss": 0.7576,
+      "step": 3572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6955222333866365e-05,
+      "loss": 0.7598,
+      "step": 3573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6902391701209824e-05,
+      "loss": 0.7063,
+      "step": 3574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684958169923683e-05,
+      "loss": 0.7808,
+      "step": 3575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6796792348466356e-05,
+      "loss": 0.7493,
+      "step": 3576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674402366940928e-05,
+      "loss": 0.7964,
+      "step": 3577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66912756825685e-05,
+      "loss": 0.7649,
+      "step": 3578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6638548408438856e-05,
+      "loss": 0.7173,
+      "step": 3579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658584186750713e-05,
+      "loss": 0.7266,
+      "step": 3580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653315608025207e-05,
+      "loss": 0.7034,
+      "step": 3581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648049106714434e-05,
+      "loss": 0.7695,
+      "step": 3582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642784684864656e-05,
+      "loss": 0.7634,
+      "step": 3583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6375223445213235e-05,
+      "loss": 0.7087,
+      "step": 3584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63226208772908e-05,
+      "loss": 0.7385,
+      "step": 3585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6270039165317605e-05,
+      "loss": 0.7649,
+      "step": 3586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621747832972389e-05,
+      "loss": 0.7355,
+      "step": 3587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616493839093179e-05,
+      "loss": 0.8154,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611241936935525e-05,
+      "loss": 0.7639,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6059921285400196e-05,
+      "loss": 0.7249,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600744415946438e-05,
+      "loss": 0.7444,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595498801193736e-05,
+      "loss": 0.7712,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590255286320062e-05,
+      "loss": 0.7189,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5850138733627434e-05,
+      "loss": 0.7212,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579774564358293e-05,
+      "loss": 0.8057,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574537361342407e-05,
+      "loss": 0.7629,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569302266349959e-05,
+      "loss": 0.7419,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564069281415009e-05,
+      "loss": 0.78,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5588384085707925e-05,
+      "loss": 0.7864,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5536096498497295e-05,
+      "loss": 0.7278,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548383007283412e-05,
+      "loss": 0.6628,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543158482902616e-05,
+      "loss": 0.7598,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537936078737292e-05,
+      "loss": 0.698,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532715796816564e-05,
+      "loss": 0.7712,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52749763916874e-05,
+      "loss": 0.749,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522281607821288e-05,
+      "loss": 0.7468,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5170677048008636e-05,
+      "loss": 0.7793,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5118559321332885e-05,
+      "loss": 0.6841,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5066462918435604e-05,
+      "loss": 0.7539,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501438785955845e-05,
+      "loss": 0.7712,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4962334164934806e-05,
+      "loss": 0.7793,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491030185478976e-05,
+      "loss": 0.7294,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485829094934006e-05,
+      "loss": 0.7603,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480630146879419e-05,
+      "loss": 0.3923,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4754333433352244e-05,
+      "loss": 0.6888,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470238686320606e-05,
+      "loss": 0.7343,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465046177853905e-05,
+      "loss": 0.7898,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459855819952635e-05,
+      "loss": 0.7418,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454667614633471e-05,
+      "loss": 0.7388,
+      "step": 3619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449481563912251e-05,
+      "loss": 0.7524,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444297669803981e-05,
+      "loss": 0.731,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439115934322816e-05,
+      "loss": 0.7655,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4339363594820874e-05,
+      "loss": 0.7463,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428758947294278e-05,
+      "loss": 0.7461,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423583699771036e-05,
+      "loss": 0.7333,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4184106189231625e-05,
+      "loss": 0.7358,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4132397067606216e-05,
+      "loss": 0.7827,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4080709652925336e-05,
+      "loss": 0.6899,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402904396527172e-05,
+      "loss": 0.7314,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3977400024719725e-05,
+      "loss": 0.7167,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3925777851335206e-05,
+      "loss": 0.7554,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387417746517557e-05,
+      "loss": 0.741,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382259888628978e-05,
+      "loss": 0.7417,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3771042134718306e-05,
+      "loss": 0.759,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371950723049314e-05,
+      "loss": 0.7463,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3667994193637796e-05,
+      "loss": 0.7429,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361650304416728e-05,
+      "loss": 0.7932,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3565033802088116e-05,
+      "loss": 0.7419,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351358648739826e-05,
+      "loss": 0.6416,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34621611200872e-05,
+      "loss": 0.8418,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341075772013589e-05,
+      "loss": 0.7576,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335937630751674e-05,
+      "loss": 0.7729,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330801690219361e-05,
+      "loss": 0.7446,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3256679524121834e-05,
+      "loss": 0.749,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3205364193248154e-05,
+      "loss": 0.7539,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315407092951078e-05,
+      "loss": 0.7896,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310279975283932e-05,
+      "loss": 0.7659,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305155068315481e-05,
+      "loss": 0.7708,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30003237403697e-05,
+      "loss": 0.3807,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2949118944387846e-05,
+      "loss": 0.7454,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289793631510448e-05,
+      "loss": 0.7307,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2846775872406256e-05,
+      "loss": 0.7227,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279563763617116e-05,
+      "loss": 0.7603,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2744521626268597e-05,
+      "loss": 0.7317,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269342786255933e-05,
+      "loss": 0.6752,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264235636489542e-05,
+      "loss": 0.7329,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259130715312033e-05,
+      "loss": 0.7527,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2540280247068874e-05,
+      "loss": 0.3665,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248927566656716e-05,
+      "loss": 0.3594,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2438293431432665e-05,
+      "loss": 0.7881,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238733356147414e-05,
+      "loss": 0.7214,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233639607649168e-05,
+      "loss": 0.3413,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228548099627665e-05,
+      "loss": 0.6836,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2234588340611746e-05,
+      "loss": 0.6946,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218371812927092e-05,
+      "loss": 0.8022,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213287038201943e-05,
+      "loss": 0.7297,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208204511861379e-05,
+      "loss": 0.739,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203124235880178e-05,
+      "loss": 0.7684,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1980462122322425e-05,
+      "loss": 0.7668,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1929704428906026e-05,
+      "loss": 0.8049,
+      "step": 3670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187896929827414e-05,
+      "loss": 0.7559,
+      "step": 3671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182825675013945e-05,
+      "loss": 0.7493,
+      "step": 3672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177756680420599e-05,
+      "loss": 0.792,
+      "step": 3673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172689948016896e-05,
+      "loss": 0.8206,
+      "step": 3674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1676254797714766e-05,
+      "loss": 0.7131,
+      "step": 3675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1625632776521037e-05,
+      "loss": 0.7456,
+      "step": 3676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1575033436256586e-05,
+      "loss": 0.7629,
+      "step": 3677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152445679658139e-05,
+      "loss": 0.7018,
+      "step": 3678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1473902877146664e-05,
+      "loss": 0.7693,
+      "step": 3679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142337169759473e-05,
+      "loss": 0.7246,
+      "step": 3680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137286327755913e-05,
+      "loss": 0.7759,
+      "step": 3681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132237763666451e-05,
+      "loss": 0.7173,
+      "step": 3682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127191479452671e-05,
+      "loss": 0.7067,
+      "step": 3683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-05,
+      "loss": 0.7153,
+      "step": 3684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117105758494056e-05,
+      "loss": 0.7698,
+      "step": 3685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112066325667954e-05,
+      "loss": 0.7472,
+      "step": 3686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107029180554996e-05,
+      "loss": 0.7322,
+      "step": 3687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1019943251123315e-05,
+      "loss": 0.7391,
+      "step": 3688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096961761296211e-05,
+      "loss": 0.7502,
+      "step": 3689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091931491062002e-05,
+      "loss": 0.6832,
+      "step": 3690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.086903516364179e-05,
+      "loss": 0.7366,
+      "step": 3691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081877839156325e-05,
+      "loss": 0.6964,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0768544613911294e-05,
+      "loss": 0.8147,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0718333850203896e-05,
+      "loss": 0.739,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0668146119950054e-05,
+      "loss": 0.7478,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0617981442649855e-05,
+      "loss": 0.738,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056783983779442e-05,
+      "loss": 0.8289,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0517721324865884e-05,
+      "loss": 0.7192,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0467625923337436e-05,
+      "loss": 0.7178,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041755365267324e-05,
+      "loss": 0.7319,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03675045323286e-05,
+      "loss": 0.7622,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031747858174964e-05,
+      "loss": 0.8035,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02674758203736e-05,
+      "loss": 0.7516,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0217496267628694e-05,
+      "loss": 0.7778,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01675399429341e-05,
+      "loss": 0.8408,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0117606865699975e-05,
+      "loss": 0.7166,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0067697055327455e-05,
+      "loss": 0.7507,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001781053120863e-05,
+      "loss": 0.7666,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9967947312726525e-05,
+      "loss": 0.7522,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9918107419255145e-05,
+      "loss": 0.3346,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9868290870159405e-05,
+      "loss": 0.783,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981849768479517e-05,
+      "loss": 0.7905,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9768727882509186e-05,
+      "loss": 0.7441,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971898148263915e-05,
+      "loss": 0.7488,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9669258504513676e-05,
+      "loss": 0.7065,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.961955896745224e-05,
+      "loss": 0.7458,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9569882890765285e-05,
+      "loss": 0.7327,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9520230293754005e-05,
+      "loss": 0.7585,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947060119571058e-05,
+      "loss": 0.6946,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942099561591802e-05,
+      "loss": 0.7441,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937141357365023e-05,
+      "loss": 0.7512,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932185508817193e-05,
+      "loss": 0.7495,
+      "step": 3722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927232017873872e-05,
+      "loss": 0.7732,
+      "step": 3723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9222808864597004e-05,
+      "loss": 0.3423,
+      "step": 3724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917332116498406e-05,
+      "loss": 0.7019,
+      "step": 3725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9123857099127936e-05,
+      "loss": 0.7896,
+      "step": 3726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907441668624755e-05,
+      "loss": 0.7126,
+      "step": 3727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902499994555261e-05,
+      "loss": 0.8066,
+      "step": 3728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897560689624362e-05,
+      "loss": 0.7466,
+      "step": 3729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892623755751189e-05,
+      "loss": 0.7197,
+      "step": 3730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887689194853951e-05,
+      "loss": 0.7489,
+      "step": 3731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8827570088499356e-05,
+      "loss": 0.7393,
+      "step": 3732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877827199655506e-05,
+      "loss": 0.7246,
+      "step": 3733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872899769186107e-05,
+      "loss": 0.7551,
+      "step": 3734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867974719356249e-05,
+      "loss": 0.7466,
+      "step": 3735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863052052079528e-05,
+      "loss": 0.7429,
+      "step": 3736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858131769268607e-05,
+      "loss": 0.3976,
+      "step": 3737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853213872835228e-05,
+      "loss": 0.7683,
+      "step": 3738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848298364690204e-05,
+      "loss": 0.7217,
+      "step": 3739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843385246743417e-05,
+      "loss": 0.7354,
+      "step": 3740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838474520903825e-05,
+      "loss": 0.7383,
+      "step": 3741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8335661890794525e-05,
+      "loss": 0.7476,
+      "step": 3742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828660253177397e-05,
+      "loss": 0.7188,
+      "step": 3743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8237567151038224e-05,
+      "loss": 0.7258,
+      "step": 3744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818855576763962e-05,
+      "loss": 0.745,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813956840062118e-05,
+      "loss": 0.7522,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8090605069016595e-05,
+      "loss": 0.7634,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804166579185018e-05,
+      "loss": 0.7368,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799275058813694e-05,
+      "loss": 0.7527,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794385947688255e-05,
+      "loss": 0.7358,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7894992477083224e-05,
+      "loss": 0.7441,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78461496077259e-05,
+      "loss": 0.7305,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779733088778813e-05,
+      "loss": 0.7739,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774853633623806e-05,
+      "loss": 0.6997,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769976597203444e-05,
+      "loss": 0.7747,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126654e-05,
+      "loss": 0.7629,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7602297881454654e-05,
+      "loss": 0.7422,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.755360019294898e-05,
+      "loss": 0.79,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7504926767530765e-05,
+      "loss": 0.7673,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7456277624111724e-05,
+      "loss": 0.7771,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7407652781594095e-05,
+      "loss": 0.7212,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735905225887072e-05,
+      "loss": 0.707,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731047607482496e-05,
+      "loss": 0.7883,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7261924248330736e-05,
+      "loss": 0.7236,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.721339679825252e-05,
+      "loss": 0.7362,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.716489374344527e-05,
+      "loss": 0.7556,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711641510275454e-05,
+      "loss": 0.7578,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7067960895016275e-05,
+      "loss": 0.7896,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701953113905703e-05,
+      "loss": 0.7756,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6971125853693836e-05,
+      "loss": 0.7449,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692274505773419e-05,
+      "loss": 0.7729,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.687438876997612e-05,
+      "loss": 0.7361,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.682605700920808e-05,
+      "loss": 0.3616,
+      "step": 3773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677774979420904e-05,
+      "loss": 0.717,
+      "step": 3774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67294671437484e-05,
+      "loss": 0.7551,
+      "step": 3775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668120907658603e-05,
+      "loss": 0.7744,
+      "step": 3776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.663297561147224e-05,
+      "loss": 0.3673,
+      "step": 3777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6584766767147796e-05,
+      "loss": 0.7295,
+      "step": 3778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.653658256234389e-05,
+      "loss": 0.7007,
+      "step": 3779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6488423015782125e-05,
+      "loss": 0.7312,
+      "step": 3780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.644028814617454e-05,
+      "loss": 0.79,
+      "step": 3781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6392177972223594e-05,
+      "loss": 0.7498,
+      "step": 3782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.634409251262217e-05,
+      "loss": 0.7104,
+      "step": 3783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.629603178605345e-05,
+      "loss": 0.7598,
+      "step": 3784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62479958111911e-05,
+      "loss": 0.7744,
+      "step": 3785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6199984606699155e-05,
+      "loss": 0.7817,
+      "step": 3786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6151998191232e-05,
+      "loss": 0.7913,
+      "step": 3787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.610403658343443e-05,
+      "loss": 0.7164,
+      "step": 3788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6056099801941534e-05,
+      "loss": 0.7051,
+      "step": 3789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.600818786537883e-05,
+      "loss": 0.7803,
+      "step": 3790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596030079236212e-05,
+      "loss": 0.7744,
+      "step": 3791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591243860149759e-05,
+      "loss": 0.7524,
+      "step": 3792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.586460131138174e-05,
+      "loss": 0.7373,
+      "step": 3793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581678894060139e-05,
+      "loss": 0.7234,
+      "step": 3794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.576900150773368e-05,
+      "loss": 0.8188,
+      "step": 3795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5721239031346066e-05,
+      "loss": 0.7668,
+      "step": 3796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5673501529996314e-05,
+      "loss": 0.8042,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.562578902223247e-05,
+      "loss": 0.759,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5578101526592875e-05,
+      "loss": 0.7493,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5530439061606205e-05,
+      "loss": 0.7461,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.548280164579126e-05,
+      "loss": 0.751,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543518929765727e-05,
+      "loss": 0.7568,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538760203570364e-05,
+      "loss": 0.7559,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534003987842005e-05,
+      "loss": 0.7646,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5292502844286446e-05,
+      "loss": 0.7471,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5244990951772974e-05,
+      "loss": 0.7539,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.519750421934003e-05,
+      "loss": 0.7332,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5150042665438233e-05,
+      "loss": 0.7646,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510260630850842e-05,
+      "loss": 0.3608,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5055195166981645e-05,
+      "loss": 0.6953,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.500780925927915e-05,
+      "loss": 0.7612,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.496044860381238e-05,
+      "loss": 0.7788,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4913113218982965e-05,
+      "loss": 0.3562,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486580312318273e-05,
+      "loss": 0.7485,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481851833479365e-05,
+      "loss": 0.75,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.477125887218792e-05,
+      "loss": 0.6929,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472402475372778e-05,
+      "loss": 0.7444,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4676815997765754e-05,
+      "loss": 0.7703,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.462963262264444e-05,
+      "loss": 0.6929,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458247464669657e-05,
+      "loss": 0.7551,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453534208824507e-05,
+      "loss": 0.7722,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4488234965602916e-05,
+      "loss": 0.7495,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444115329707325e-05,
+      "loss": 0.7944,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439409710094929e-05,
+      "loss": 0.7744,
+      "step": 3824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.434706639551438e-05,
+      "loss": 0.6934,
+      "step": 3825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430006119904197e-05,
+      "loss": 0.76,
+      "step": 3826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.425308152979555e-05,
+      "loss": 0.687,
+      "step": 3827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4206127406028745e-05,
+      "loss": 0.6901,
+      "step": 3828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415919884598523e-05,
+      "loss": 0.7151,
+      "step": 3829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411229586789876e-05,
+      "loss": 0.7241,
+      "step": 3830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.406541848999312e-05,
+      "loss": 0.7703,
+      "step": 3831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.401856673048217e-05,
+      "loss": 0.7595,
+      "step": 3832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397174060756987e-05,
+      "loss": 0.7828,
+      "step": 3833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.392494013945009e-05,
+      "loss": 0.7727,
+      "step": 3834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3878165344306826e-05,
+      "loss": 0.7798,
+      "step": 3835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.383141624031408e-05,
+      "loss": 0.7346,
+      "step": 3836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.378469284563589e-05,
+      "loss": 0.7695,
+      "step": 3837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373799517842627e-05,
+      "loss": 0.7488,
+      "step": 3838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369132325682926e-05,
+      "loss": 0.3973,
+      "step": 3839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3644677098978894e-05,
+      "loss": 0.3617,
+      "step": 3840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359805672299918e-05,
+      "loss": 0.7737,
+      "step": 3841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355146214700413e-05,
+      "loss": 0.7494,
+      "step": 3842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.350489338909772e-05,
+      "loss": 0.7407,
+      "step": 3843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.345835046737391e-05,
+      "loss": 0.7427,
+      "step": 3844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.341183339991658e-05,
+      "loss": 0.7549,
+      "step": 3845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336534220479961e-05,
+      "loss": 0.7485,
+      "step": 3846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.331887690008681e-05,
+      "loss": 0.7776,
+      "step": 3847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3272437503831946e-05,
+      "loss": 0.7769,
+      "step": 3848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322602403407869e-05,
+      "loss": 0.7122,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317963650886064e-05,
+      "loss": 0.6968,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3133274946201334e-05,
+      "loss": 0.7485,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-05,
+      "loss": 0.7432,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304062978060263e-05,
+      "loss": 0.7532,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.299434621365982e-05,
+      "loss": 0.6954,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.294808868126894e-05,
+      "loss": 0.3628,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290185720140301e-05,
+      "loss": 0.7095,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285565179202491e-05,
+      "loss": 0.7163,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280947247108741e-05,
+      "loss": 0.78,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2763319256533175e-05,
+      "loss": 0.7468,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.271719216629468e-05,
+      "loss": 0.7356,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2671091218294284e-05,
+      "loss": 0.7393,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2625016430444164e-05,
+      "loss": 0.3682,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257896782064631e-05,
+      "loss": 0.7754,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253294540679257e-05,
+      "loss": 0.7141,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.248694920676464e-05,
+      "loss": 0.7852,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244097923843398e-05,
+      "loss": 0.35,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2395035519661896e-05,
+      "loss": 0.7812,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234911806829948e-05,
+      "loss": 0.7485,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2303226902187623e-05,
+      "loss": 0.7341,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225736203915698e-05,
+      "loss": 0.7749,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2211523497028015e-05,
+      "loss": 0.7471,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216571129361097e-05,
+      "loss": 0.7617,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211992544670582e-05,
+      "loss": 0.7542,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.207416597410232e-05,
+      "loss": 0.7429,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202843289357998e-05,
+      "loss": 0.7756,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.198272622290804e-05,
+      "loss": 0.7108,
+      "step": 3876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.193704597984552e-05,
+      "loss": 0.7402,
+      "step": 3877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.18913921821411e-05,
+      "loss": 0.751,
+      "step": 3878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.184576484753328e-05,
+      "loss": 0.7593,
+      "step": 3879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1800163993750166e-05,
+      "loss": 0.7583,
+      "step": 3880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1754589638509644e-05,
+      "loss": 0.814,
+      "step": 3881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1709041799519315e-05,
+      "loss": 0.7292,
+      "step": 3882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166352049447644e-05,
+      "loss": 0.7355,
+      "step": 3883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161802574106799e-05,
+      "loss": 0.757,
+      "step": 3884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157255755697062e-05,
+      "loss": 0.7261,
+      "step": 3885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152711595985065e-05,
+      "loss": 0.7524,
+      "step": 3886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148170096736408e-05,
+      "loss": 0.7568,
+      "step": 3887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143631259715658e-05,
+      "loss": 0.7117,
+      "step": 3888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1390950866863454e-05,
+      "loss": 0.782,
+      "step": 3889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134561579410966e-05,
+      "loss": 0.7007,
+      "step": 3890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.130030739650983e-05,
+      "loss": 0.7271,
+      "step": 3891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1255025691668185e-05,
+      "loss": 0.7388,
+      "step": 3892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.120977069717861e-05,
+      "loss": 0.7825,
+      "step": 3893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.116454243062459e-05,
+      "loss": 0.7822,
+      "step": 3894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1119340909579264e-05,
+      "loss": 0.7556,
+      "step": 3895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10741661516053e-05,
+      "loss": 0.7549,
+      "step": 3896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.102901817425504e-05,
+      "loss": 0.7524,
+      "step": 3897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0983896995070395e-05,
+      "loss": 0.7542,
+      "step": 3898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093880263158286e-05,
+      "loss": 0.7483,
+      "step": 3899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089373510131354e-05,
+      "loss": 0.7444,
+      "step": 3900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0848694421773075e-05,
+      "loss": 0.7639,
+      "step": 3901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080368061046167e-05,
+      "loss": 0.7417,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075869368486914e-05,
+      "loss": 0.7349,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.071373366247482e-05,
+      "loss": 0.339,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066880056074758e-05,
+      "loss": 0.7852,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062389439714584e-05,
+      "loss": 0.7585,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0579015189117576e-05,
+      "loss": 0.7661,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053416295410026e-05,
+      "loss": 0.7217,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0489337709520917e-05,
+      "loss": 0.7371,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.044453947279603e-05,
+      "loss": 0.7341,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0399768261331662e-05,
+      "loss": 0.3367,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0355024092523332e-05,
+      "loss": 0.7727,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031030698375602e-05,
+      "loss": 0.7664,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0265616952404253e-05,
+      "loss": 0.7759,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0220954015832003e-05,
+      "loss": 0.7153,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0176318191392726e-05,
+      "loss": 0.7351,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013170949642935e-05,
+      "loss": 0.702,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008712794827426e-05,
+      "loss": 0.39,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0042573564249265e-05,
+      "loss": 0.7361,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9998046361665666e-05,
+      "loss": 0.8149,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995354635782417e-05,
+      "loss": 0.7505,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9909073570014912e-05,
+      "loss": 0.713,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9864628015517483e-05,
+      "loss": 0.7402,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9820209711600854e-05,
+      "loss": 0.7822,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977581867552346e-05,
+      "loss": 0.7612,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9731454924533086e-05,
+      "loss": 0.7075,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9687118475866947e-05,
+      "loss": 0.7594,
+      "step": 3927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9642809346751678e-05,
+      "loss": 0.7581,
+      "step": 3928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9598527554403188e-05,
+      "loss": 0.7231,
+      "step": 3929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9554273116026887e-05,
+      "loss": 0.7742,
+      "step": 3930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9510046048817497e-05,
+      "loss": 0.7426,
+      "step": 3931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9465846369959127e-05,
+      "loss": 0.728,
+      "step": 3932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942167409662524e-05,
+      "loss": 0.728,
+      "step": 3933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.937752924597864e-05,
+      "loss": 0.7607,
+      "step": 3934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9333411835171477e-05,
+      "loss": 0.7299,
+      "step": 3935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.7771,
+      "step": 3936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.92452594016308e-05,
+      "loss": 0.7705,
+      "step": 3937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9201224413148255e-05,
+      "loss": 0.7852,
+      "step": 3938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9157216933007093e-05,
+      "loss": 0.7463,
+      "step": 3939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9113236978306103e-05,
+      "loss": 0.7185,
+      "step": 3940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.906928456613336e-05,
+      "loss": 0.3752,
+      "step": 3941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9025359713566246e-05,
+      "loss": 0.7234,
+      "step": 3942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8981462437671458e-05,
+      "loss": 0.731,
+      "step": 3943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893759275550494e-05,
+      "loss": 0.7734,
+      "step": 3944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8893750684111975e-05,
+      "loss": 0.6895,
+      "step": 3945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8849936240527008e-05,
+      "loss": 0.7603,
+      "step": 3946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8806149441773845e-05,
+      "loss": 0.7744,
+      "step": 3947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876239030486554e-05,
+      "loss": 0.7078,
+      "step": 3948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871865884680436e-05,
+      "loss": 0.7073,
+      "step": 3949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8674955084581857e-05,
+      "loss": 0.7148,
+      "step": 3950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8631279035178793e-05,
+      "loss": 0.7573,
+      "step": 3951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8587630715565183e-05,
+      "loss": 0.729,
+      "step": 3952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.854401014270026e-05,
+      "loss": 0.7163,
+      "step": 3953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850041733353247e-05,
+      "loss": 0.7454,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8456852304999482e-05,
+      "loss": 0.7053,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8413315074028158e-05,
+      "loss": 0.7002,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8369805657534575e-05,
+      "loss": 0.7493,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8326324072423993e-05,
+      "loss": 0.7834,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828287033559086e-05,
+      "loss": 0.782,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8239444463918807e-05,
+      "loss": 0.7043,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8196046474280667e-05,
+      "loss": 0.7734,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8152676383538345e-05,
+      "loss": 0.7368,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8109334208542993e-05,
+      "loss": 0.7893,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8066019966134904e-05,
+      "loss": 0.7476,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802273367314351e-05,
+      "loss": 0.7111,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797947534638736e-05,
+      "loss": 0.739,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7936245002674176e-05,
+      "loss": 0.6899,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7893042658800793e-05,
+      "loss": 0.7448,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7849868331553143e-05,
+      "loss": 0.6687,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.78067220377063e-05,
+      "loss": 0.7329,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776360379402445e-05,
+      "loss": 0.3672,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7720513617260856e-05,
+      "loss": 0.71,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7677451524157882e-05,
+      "loss": 0.7158,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763441753144701e-05,
+      "loss": 0.7468,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7591411655848753e-05,
+      "loss": 0.7507,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7548433914072734e-05,
+      "loss": 0.7449,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7505484322817655e-05,
+      "loss": 0.7341,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746256289877126e-05,
+      "loss": 0.8088,
+      "step": 3978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7419669658610325e-05,
+      "loss": 0.7417,
+      "step": 3979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7376804619000707e-05,
+      "loss": 0.7944,
+      "step": 3980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7333967796597315e-05,
+      "loss": 0.7734,
+      "step": 3981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729115920804408e-05,
+      "loss": 0.74,
+      "step": 3982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7248378869973945e-05,
+      "loss": 0.7646,
+      "step": 3983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.720562679900892e-05,
+      "loss": 0.3643,
+      "step": 3984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7162903011759987e-05,
+      "loss": 0.7769,
+      "step": 3985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7120207524827168e-05,
+      "loss": 0.7621,
+      "step": 3986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707754035479948e-05,
+      "loss": 0.717,
+      "step": 3987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7034901518254917e-05,
+      "loss": 0.772,
+      "step": 3988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6992291031760518e-05,
+      "loss": 0.7454,
+      "step": 3989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.694970891187225e-05,
+      "loss": 0.78,
+      "step": 3990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.690715517513508e-05,
+      "loss": 0.8003,
+      "step": 3991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6864629838082956e-05,
+      "loss": 0.7224,
+      "step": 3992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6822132917238785e-05,
+      "loss": 0.7412,
+      "step": 3993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677966442911445e-05,
+      "loss": 0.7393,
+      "step": 3994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6737224390210726e-05,
+      "loss": 0.7507,
+      "step": 3995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669481281701739e-05,
+      "loss": 0.7351,
+      "step": 3996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665242972601316e-05,
+      "loss": 0.7588,
+      "step": 3997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6610075133665668e-05,
+      "loss": 0.709,
+      "step": 3998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6567749056431467e-05,
+      "loss": 0.7468,
+      "step": 3999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652545151075606e-05,
+      "loss": 0.7405,
+      "step": 4000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6483182513073835e-05,
+      "loss": 0.7898,
+      "step": 4001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6440942079808096e-05,
+      "loss": 0.6797,
+      "step": 4002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639873022737105e-05,
+      "loss": 0.751,
+      "step": 4003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.635654697216382e-05,
+      "loss": 0.6927,
+      "step": 4004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6314392330576378e-05,
+      "loss": 0.7212,
+      "step": 4005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6272266318987603e-05,
+      "loss": 0.7461,
+      "step": 4006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6230168953765256e-05,
+      "loss": 0.7001,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6188100251265945e-05,
+      "loss": 0.7529,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6146060227835157e-05,
+      "loss": 0.7197,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6104048899807244e-05,
+      "loss": 0.6836,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6062066283505403e-05,
+      "loss": 0.7395,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6020112395241624e-05,
+      "loss": 0.7227,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5978187251316823e-05,
+      "loss": 0.8171,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5936290868020686e-05,
+      "loss": 0.7654,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5894423261631717e-05,
+      "loss": 0.7822,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5852584448417328e-05,
+      "loss": 0.7891,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5810774444633644e-05,
+      "loss": 0.718,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5768993266525653e-05,
+      "loss": 0.7324,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57272409303271e-05,
+      "loss": 0.7681,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5685517452260567e-05,
+      "loss": 0.7717,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.564382284853738e-05,
+      "loss": 0.7625,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5602157135357697e-05,
+      "loss": 0.7229,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5560520328910396e-05,
+      "loss": 0.7551,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55189124453732e-05,
+      "loss": 0.7117,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547733350091247e-05,
+      "loss": 0.7357,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5435783511683443e-05,
+      "loss": 0.7642,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5394262493830055e-05,
+      "loss": 0.7737,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5352770463484987e-05,
+      "loss": 0.7219,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531130743676966e-05,
+      "loss": 0.7395,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5269873429794234e-05,
+      "loss": 0.7253,
+      "step": 4030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5228468458657584e-05,
+      "loss": 0.7203,
+      "step": 4031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5187092539447298e-05,
+      "loss": 0.7441,
+      "step": 4032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5145745688239696e-05,
+      "loss": 0.7285,
+      "step": 4033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5104427921099782e-05,
+      "loss": 0.7788,
+      "step": 4034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506313925408127e-05,
+      "loss": 0.7172,
+      "step": 4035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502187970322657e-05,
+      "loss": 0.7603,
+      "step": 4036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.498064928456678e-05,
+      "loss": 0.7178,
+      "step": 4037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493944801412167e-05,
+      "loss": 0.7532,
+      "step": 4038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489827590789968e-05,
+      "loss": 0.7166,
+      "step": 4039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485713298189798e-05,
+      "loss": 0.7294,
+      "step": 4040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4816019252102273e-05,
+      "loss": 0.7542,
+      "step": 4041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477493473448703e-05,
+      "loss": 0.7161,
+      "step": 4042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4733879445015328e-05,
+      "loss": 0.7544,
+      "step": 4043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4692853399638917e-05,
+      "loss": 0.7466,
+      "step": 4044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4651856614298143e-05,
+      "loss": 0.791,
+      "step": 4045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461088910492202e-05,
+      "loss": 0.7866,
+      "step": 4046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4569950887428152e-05,
+      "loss": 0.7043,
+      "step": 4047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45290419777228e-05,
+      "loss": 0.7842,
+      "step": 4048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4488162391700808e-05,
+      "loss": 0.748,
+      "step": 4049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4447312145245647e-05,
+      "loss": 0.7263,
+      "step": 4050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440649125422937e-05,
+      "loss": 0.7546,
+      "step": 4051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4365699734512637e-05,
+      "loss": 0.7158,
+      "step": 4052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4324937601944685e-05,
+      "loss": 0.7683,
+      "step": 4053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4284204872363357e-05,
+      "loss": 0.7043,
+      "step": 4054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4243501561595026e-05,
+      "loss": 0.3435,
+      "step": 4055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420282768545469e-05,
+      "loss": 0.7673,
+      "step": 4056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4162183259745896e-05,
+      "loss": 0.7488,
+      "step": 4057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412156830026069e-05,
+      "loss": 0.7523,
+      "step": 4058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408098282277972e-05,
+      "loss": 0.783,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4040426843072206e-05,
+      "loss": 0.7347,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3999900376895845e-05,
+      "loss": 0.3322,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3959403439996907e-05,
+      "loss": 0.7568,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3918936048110195e-05,
+      "loss": 0.7463,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3878498216958988e-05,
+      "loss": 0.7429,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.383808996225513e-05,
+      "loss": 0.7307,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3797711299698923e-05,
+      "loss": 0.7871,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375736224497922e-05,
+      "loss": 0.7681,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.371704281377335e-05,
+      "loss": 0.7712,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.367675302174711e-05,
+      "loss": 0.7603,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3636492884554817e-05,
+      "loss": 0.7661,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3596262417839255e-05,
+      "loss": 0.7332,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355606163723165e-05,
+      "loss": 0.7615,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351589055835177e-05,
+      "loss": 0.718,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347574919680773e-05,
+      "loss": 0.7603,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.343563756819619e-05,
+      "loss": 0.7375,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339555568810221e-05,
+      "loss": 0.7554,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3355503572099326e-05,
+      "loss": 0.7544,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331548123574948e-05,
+      "loss": 0.7208,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3275488694603075e-05,
+      "loss": 0.7629,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323552596419889e-05,
+      "loss": 0.7859,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.319559306006417e-05,
+      "loss": 0.7324,
+      "step": 4081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3155689997714548e-05,
+      "loss": 0.7607,
+      "step": 4082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3115816792654056e-05,
+      "loss": 0.7896,
+      "step": 4083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3075973460375133e-05,
+      "loss": 0.7103,
+      "step": 4084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.303616001635861e-05,
+      "loss": 0.769,
+      "step": 4085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2996376476073723e-05,
+      "loss": 0.3526,
+      "step": 4086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295662285497805e-05,
+      "loss": 0.6993,
+      "step": 4087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291689916851758e-05,
+      "loss": 0.7334,
+      "step": 4088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2877205432126657e-05,
+      "loss": 0.7061,
+      "step": 4089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2837541661228025e-05,
+      "loss": 0.7808,
+      "step": 4090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279790787123267e-05,
+      "loss": 0.7144,
+      "step": 4091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.275830407754006e-05,
+      "loss": 0.7041,
+      "step": 4092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271873029553795e-05,
+      "loss": 0.761,
+      "step": 4093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2679186540602437e-05,
+      "loss": 0.7279,
+      "step": 4094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263967282809796e-05,
+      "loss": 0.7197,
+      "step": 4095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260018917337726e-05,
+      "loss": 0.8003,
+      "step": 4096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2560735591781447e-05,
+      "loss": 0.821,
+      "step": 4097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2521312098639914e-05,
+      "loss": 0.7542,
+      "step": 4098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2481918709270345e-05,
+      "loss": 0.7086,
+      "step": 4099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2442555438978773e-05,
+      "loss": 0.6702,
+      "step": 4100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.240322230305951e-05,
+      "loss": 0.769,
+      "step": 4101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2363919316795133e-05,
+      "loss": 0.749,
+      "step": 4102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2324646495456546e-05,
+      "loss": 0.735,
+      "step": 4103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2285403854302912e-05,
+      "loss": 0.7583,
+      "step": 4104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2246191408581662e-05,
+      "loss": 0.7566,
+      "step": 4105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2207009173528527e-05,
+      "loss": 0.7722,
+      "step": 4106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2167857164367422e-05,
+      "loss": 0.7012,
+      "step": 4107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212873539631061e-05,
+      "loss": 0.384,
+      "step": 4108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2089643884558554e-05,
+      "loss": 0.7869,
+      "step": 4109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2050582644299978e-05,
+      "loss": 0.724,
+      "step": 4110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201155169071184e-05,
+      "loss": 0.7461,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197255103895931e-05,
+      "loss": 0.3702,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193358070419583e-05,
+      "loss": 0.7373,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189464070156302e-05,
+      "loss": 0.8015,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185573104619073e-05,
+      "loss": 0.3589,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181685175319702e-05,
+      "loss": 0.7605,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.177800283768817e-05,
+      "loss": 0.7307,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173918431475861e-05,
+      "loss": 0.7429,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1700396199491013e-05,
+      "loss": 0.7244,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1661638506956205e-05,
+      "loss": 0.7651,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1622911252213197e-05,
+      "loss": 0.7239,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1584214450309194e-05,
+      "loss": 0.8103,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154554811627956e-05,
+      "loss": 0.7485,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1506912265147772e-05,
+      "loss": 0.3375,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146830691192553e-05,
+      "loss": 0.7217,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.142973207161265e-05,
+      "loss": 0.707,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139118775919712e-05,
+      "loss": 0.7141,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1352673989655026e-05,
+      "loss": 0.7576,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1314190777950627e-05,
+      "loss": 0.7551,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.127573813903627e-05,
+      "loss": 0.7153,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1237316087852466e-05,
+      "loss": 0.7876,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119892463932781e-05,
+      "loss": 0.3654,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116056380837902e-05,
+      "loss": 0.7332,
+      "step": 4133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11222336099109e-05,
+      "loss": 0.74,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.108393405881638e-05,
+      "loss": 0.7551,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1045665169976468e-05,
+      "loss": 0.7278,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1007426958260245e-05,
+      "loss": 0.7295,
+      "step": 4137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096921943852489e-05,
+      "loss": 0.3707,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0931042625615694e-05,
+      "loss": 0.7424,
+      "step": 4139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0892896534365904e-05,
+      "loss": 0.6921,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0854781179596938e-05,
+      "loss": 0.7627,
+      "step": 4141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0816696576118223e-05,
+      "loss": 0.7239,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0778642738727272e-05,
+      "loss": 0.7322,
+      "step": 4143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0740619682209607e-05,
+      "loss": 0.7256,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0702627421338817e-05,
+      "loss": 0.744,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0664665970876496e-05,
+      "loss": 0.7371,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06267353455723e-05,
+      "loss": 0.7639,
+      "step": 4147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05888355601639e-05,
+      "loss": 0.3798,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0550966629376956e-05,
+      "loss": 0.7029,
+      "step": 4149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0513128567925178e-05,
+      "loss": 0.6635,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047532139051026e-05,
+      "loss": 0.7274,
+      "step": 4151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043754511182191e-05,
+      "loss": 0.7499,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0399799746537805e-05,
+      "loss": 0.7081,
+      "step": 4153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036208530932363e-05,
+      "loss": 0.7524,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0324401814833072e-05,
+      "loss": 0.7754,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0286749277707782e-05,
+      "loss": 0.7197,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0249127712577322e-05,
+      "loss": 0.7725,
+      "step": 4157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021153713405931e-05,
+      "loss": 0.3381,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0173977556759284e-05,
+      "loss": 0.7522,
+      "step": 4159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013644899527074e-05,
+      "loss": 0.7109,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009895146417512e-05,
+      "loss": 0.7246,
+      "step": 4161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0061484978041813e-05,
+      "loss": 0.7395,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0024049551428146e-05,
+      "loss": 0.7141,
+      "step": 4163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9986645198879385e-05,
+      "loss": 0.7451,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9949271934928705e-05,
+      "loss": 0.748,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9911929774097215e-05,
+      "loss": 0.729,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9874618730893946e-05,
+      "loss": 0.7317,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9837338819815808e-05,
+      "loss": 0.7137,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9800090055347653e-05,
+      "loss": 0.7424,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9762872451962212e-05,
+      "loss": 0.7869,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9725686024120095e-05,
+      "loss": 0.7317,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9688530786269855e-05,
+      "loss": 0.7502,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965140675284779e-05,
+      "loss": 0.679,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9614313938278272e-05,
+      "loss": 0.7363,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9577252356973398e-05,
+      "loss": 0.7358,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9540222023333166e-05,
+      "loss": 0.7479,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9503222951745447e-05,
+      "loss": 0.7395,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9466255156585967e-05,
+      "loss": 0.7639,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9429318652218276e-05,
+      "loss": 0.7373,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9392413452993786e-05,
+      "loss": 0.7478,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9355539573251734e-05,
+      "loss": 0.7141,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9318697027319198e-05,
+      "loss": 0.7554,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9281885829511083e-05,
+      "loss": 0.7385,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9245105994130085e-05,
+      "loss": 0.7444,
+      "step": 4184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9208357535466792e-05,
+      "loss": 0.7578,
+      "step": 4185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917164046779948e-05,
+      "loss": 0.78,
+      "step": 4186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9134954805394322e-05,
+      "loss": 0.3458,
+      "step": 4187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.7722,
+      "step": 4188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9061677753374042e-05,
+      "loss": 0.751,
+      "step": 4189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9025086392230163e-05,
+      "loss": 0.7673,
+      "step": 4190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.898852649329095e-05,
+      "loss": 0.7458,
+      "step": 4191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.895199807076148e-05,
+      "loss": 0.337,
+      "step": 4192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.891550113883458e-05,
+      "loss": 0.7234,
+      "step": 4193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8879035711690895e-05,
+      "loss": 0.7607,
+      "step": 4194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.884260180349877e-05,
+      "loss": 0.7277,
+      "step": 4195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.880619942841435e-05,
+      "loss": 0.8176,
+      "step": 4196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8769828600581496e-05,
+      "loss": 0.7134,
+      "step": 4197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8733489334131827e-05,
+      "loss": 0.7202,
+      "step": 4198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8697181643184703e-05,
+      "loss": 0.7886,
+      "step": 4199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8660905541847207e-05,
+      "loss": 0.7399,
+      "step": 4200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8624661044214152e-05,
+      "loss": 0.7524,
+      "step": 4201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.858844816436809e-05,
+      "loss": 0.7163,
+      "step": 4202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8552266916379225e-05,
+      "loss": 0.7296,
+      "step": 4203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8516117314305524e-05,
+      "loss": 0.7263,
+      "step": 4204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8479999372192657e-05,
+      "loss": 0.7422,
+      "step": 4205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8443913104073983e-05,
+      "loss": 0.7439,
+      "step": 4206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.840785852397057e-05,
+      "loss": 0.7302,
+      "step": 4207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8371835645891135e-05,
+      "loss": 0.762,
+      "step": 4208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.833584448383211e-05,
+      "loss": 0.6863,
+      "step": 4209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82998850517776e-05,
+      "loss": 0.756,
+      "step": 4210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.826395736369937e-05,
+      "loss": 0.7345,
+      "step": 4211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8228061433556866e-05,
+      "loss": 0.7466,
+      "step": 4212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8192197275297173e-05,
+      "loss": 0.7654,
+      "step": 4213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.815636490285506e-05,
+      "loss": 0.7708,
+      "step": 4214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8120564330152923e-05,
+      "loss": 0.7314,
+      "step": 4215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808479557110081e-05,
+      "loss": 0.7585,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8049058639596416e-05,
+      "loss": 0.7957,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801335354952507e-05,
+      "loss": 0.7434,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7977680314759694e-05,
+      "loss": 0.7593,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7942038949160854e-05,
+      "loss": 0.7349,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7906429466576767e-05,
+      "loss": 0.7207,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787085188084323e-05,
+      "loss": 0.731,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783530620578364e-05,
+      "loss": 0.804,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7799792455209018e-05,
+      "loss": 0.7234,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776431064291796e-05,
+      "loss": 0.7087,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7728860782696664e-05,
+      "loss": 0.3586,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7693442888318923e-05,
+      "loss": 0.7263,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.765805697354608e-05,
+      "loss": 0.7102,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7622703052127076e-05,
+      "loss": 0.7583,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7587381137798432e-05,
+      "loss": 0.7246,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7552091244284197e-05,
+      "loss": 0.7434,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7516833385296018e-05,
+      "loss": 0.3414,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7481607574533044e-05,
+      "loss": 0.7012,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744641382568204e-05,
+      "loss": 0.7397,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741125215241728e-05,
+      "loss": 0.7239,
+      "step": 4235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7376122568400532e-05,
+      "loss": 0.7622,
+      "step": 4236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.734102508728115e-05,
+      "loss": 0.6604,
+      "step": 4237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730595972269602e-05,
+      "loss": 0.7731,
+      "step": 4238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7270926488269502e-05,
+      "loss": 0.759,
+      "step": 4239
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723592539761353e-05,
+      "loss": 0.7202,
+      "step": 4240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.720095646432751e-05,
+      "loss": 0.825,
+      "step": 4241
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716601970199836e-05,
+      "loss": 0.7642,
+      "step": 4242
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7131115124200492e-05,
+      "loss": 0.7233,
+      "step": 4243
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7096242744495837e-05,
+      "loss": 0.7446,
+      "step": 4244
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061402576433793e-05,
+      "loss": 0.3823,
+      "step": 4245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702659463355125e-05,
+      "loss": 0.7334,
+      "step": 4246
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699181892937258e-05,
+      "loss": 0.7229,
+      "step": 4247
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.695707547740962e-05,
+      "loss": 0.7539,
+      "step": 4248
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6922364291161698e-05,
+      "loss": 0.7388,
+      "step": 4249
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.688768538411556e-05,
+      "loss": 0.7332,
+      "step": 4250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6853038769745467e-05,
+      "loss": 0.7463,
+      "step": 4251
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.681842446151313e-05,
+      "loss": 0.7883,
+      "step": 4252
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6783842472867607e-05,
+      "loss": 0.7256,
+      "step": 4253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674929281724551e-05,
+      "loss": 0.7424,
+      "step": 4254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6714775508070866e-05,
+      "loss": 0.7124,
+      "step": 4255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668029055875512e-05,
+      "loss": 0.8093,
+      "step": 4256
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6645837982697122e-05,
+      "loss": 0.7952,
+      "step": 4257
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661141779328319e-05,
+      "loss": 0.7651,
+      "step": 4258
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6577030003887017e-05,
+      "loss": 0.751,
+      "step": 4259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6542674627869737e-05,
+      "loss": 0.7625,
+      "step": 4260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6508351678579882e-05,
+      "loss": 0.3332,
+      "step": 4261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.647406116935336e-05,
+      "loss": 0.7283,
+      "step": 4262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6439803113513507e-05,
+      "loss": 0.7375,
+      "step": 4263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640557752437103e-05,
+      "loss": 0.6709,
+      "step": 4264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6371384415224045e-05,
+      "loss": 0.3535,
+      "step": 4265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6337223799358026e-05,
+      "loss": 0.7363,
+      "step": 4266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6303095690045822e-05,
+      "loss": 0.7832,
+      "step": 4267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6269000100547683e-05,
+      "loss": 0.75,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6234937044111152e-05,
+      "loss": 0.7466,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6200906533971205e-05,
+      "loss": 0.7474,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.616690858335014e-05,
+      "loss": 0.8149,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6132943205457606e-05,
+      "loss": 0.7249,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6099010413490623e-05,
+      "loss": 0.7583,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6065110220633496e-05,
+      "loss": 0.7317,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6031242640057907e-05,
+      "loss": 0.7935,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5997407684922862e-05,
+      "loss": 0.6744,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5963605368374677e-05,
+      "loss": 0.3595,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5929835703546993e-05,
+      "loss": 0.7822,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5896098703560758e-05,
+      "loss": 0.6831,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.586239438152424e-05,
+      "loss": 0.7336,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5828722750533008e-05,
+      "loss": 0.7349,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579508382366992e-05,
+      "loss": 0.7671,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5761477614005125e-05,
+      "loss": 0.7566,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5727904134596083e-05,
+      "loss": 0.7346,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5694363398487543e-05,
+      "loss": 0.7563,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.566085541871145e-05,
+      "loss": 0.7361,
+      "step": 4286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5627380208287114e-05,
+      "loss": 0.7515,
+      "step": 4287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.559393778022109e-05,
+      "loss": 0.7993,
+      "step": 4288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556052814750717e-05,
+      "loss": 0.7803,
+      "step": 4289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5527151323126422e-05,
+      "loss": 0.7397,
+      "step": 4290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.549380732004718e-05,
+      "loss": 0.7332,
+      "step": 4291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5460496151225e-05,
+      "loss": 0.3607,
+      "step": 4292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.542721782960268e-05,
+      "loss": 0.6925,
+      "step": 4293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539397236811029e-05,
+      "loss": 0.7668,
+      "step": 4294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.536075977966509e-05,
+      "loss": 0.7114,
+      "step": 4295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5327580077171587e-05,
+      "loss": 0.724,
+      "step": 4296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5294433273521525e-05,
+      "loss": 0.7603,
+      "step": 4297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526131938159383e-05,
+      "loss": 0.7429,
+      "step": 4298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.522823841425467e-05,
+      "loss": 0.7605,
+      "step": 4299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5195190384357404e-05,
+      "loss": 0.7632,
+      "step": 4300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5162175304742632e-05,
+      "loss": 0.7662,
+      "step": 4301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5129193188238066e-05,
+      "loss": 0.7295,
+      "step": 4302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5096244047658703e-05,
+      "loss": 0.7148,
+      "step": 4303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5063327895806667e-05,
+      "loss": 0.74,
+      "step": 4304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5030444745471294e-05,
+      "loss": 0.7134,
+      "step": 4305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4997594609429088e-05,
+      "loss": 0.7285,
+      "step": 4306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4964777500443728e-05,
+      "loss": 0.3564,
+      "step": 4307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4931993431266056e-05,
+      "loss": 0.7388,
+      "step": 4308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4899242414634096e-05,
+      "loss": 0.7778,
+      "step": 4309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4866524463272991e-05,
+      "loss": 0.7299,
+      "step": 4310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4833839589895071e-05,
+      "loss": 0.7029,
+      "step": 4311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4801187807199801e-05,
+      "loss": 0.6897,
+      "step": 4312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4768569127873789e-05,
+      "loss": 0.7529,
+      "step": 4313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4735983564590783e-05,
+      "loss": 0.7598,
+      "step": 4314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4703431130011668e-05,
+      "loss": 0.7192,
+      "step": 4315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.467091183678444e-05,
+      "loss": 0.7673,
+      "step": 4316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4638425697544233e-05,
+      "loss": 0.3701,
+      "step": 4317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4605972724913331e-05,
+      "loss": 0.3943,
+      "step": 4318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4573552931501044e-05,
+      "loss": 0.7605,
+      "step": 4319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4541166329903855e-05,
+      "loss": 0.7275,
+      "step": 4320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4508812932705363e-05,
+      "loss": 0.7178,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4476492752476222e-05,
+      "loss": 0.7366,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4444205801774202e-05,
+      "loss": 0.717,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4411952093144165e-05,
+      "loss": 0.7551,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4379731639118043e-05,
+      "loss": 0.6951,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4347544452214868e-05,
+      "loss": 0.7438,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4315390544940732e-05,
+      "loss": 0.7019,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4283269929788779e-05,
+      "loss": 0.7273,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4251182619239256e-05,
+      "loss": 0.7078,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4219128625759426e-05,
+      "loss": 0.7656,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4187107961803703e-05,
+      "loss": 0.7085,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.415512063981339e-05,
+      "loss": 0.7598,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4123166672216969e-05,
+      "loss": 0.7485,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.409124607142992e-05,
+      "loss": 0.6934,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405935884985473e-05,
+      "loss": 0.7327,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402750501988097e-05,
+      "loss": 0.7004,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3995684593885206e-05,
+      "loss": 0.7424,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3963897584231023e-05,
+      "loss": 0.7407,
+      "step": 4338
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3932144003269033e-05,
+      "loss": 0.7378,
+      "step": 4339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3900423863336842e-05,
+      "loss": 0.7405,
+      "step": 4340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3868737176759106e-05,
+      "loss": 0.3221,
+      "step": 4341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3837083955847418e-05,
+      "loss": 0.7065,
+      "step": 4342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3805464212900432e-05,
+      "loss": 0.7236,
+      "step": 4343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3773877960203741e-05,
+      "loss": 0.7334,
+      "step": 4344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3742325210029972e-05,
+      "loss": 0.7192,
+      "step": 4345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3710805974638696e-05,
+      "loss": 0.7202,
+      "step": 4346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3679320266276519e-05,
+      "loss": 0.7212,
+      "step": 4347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.364786809717692e-05,
+      "loss": 0.7949,
+      "step": 4348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3616449479560434e-05,
+      "loss": 0.6805,
+      "step": 4349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3585064425634542e-05,
+      "loss": 0.7517,
+      "step": 4350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3553712947593656e-05,
+      "loss": 0.7632,
+      "step": 4351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3522395057619187e-05,
+      "loss": 0.741,
+      "step": 4352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3491110767879444e-05,
+      "loss": 0.7505,
+      "step": 4353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.345986009052972e-05,
+      "loss": 0.7751,
+      "step": 4354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3428643037712229e-05,
+      "loss": 0.7795,
+      "step": 4355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 0.7578,
+      "step": 4356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3366309854177517e-05,
+      "loss": 0.3749,
+      "step": 4357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3335193747679386e-05,
+      "loss": 0.7654,
+      "step": 4358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.330411131415169e-05,
+      "loss": 0.7307,
+      "step": 4359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3273062565671256e-05,
+      "loss": 0.759,
+      "step": 4360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3242047514301858e-05,
+      "loss": 0.7881,
+      "step": 4361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3211066172094177e-05,
+      "loss": 0.7715,
+      "step": 4362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3180118551085763e-05,
+      "loss": 0.7622,
+      "step": 4363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3149204663301118e-05,
+      "loss": 0.6681,
+      "step": 4364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3118324520751557e-05,
+      "loss": 0.7607,
+      "step": 4365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.308747813543536e-05,
+      "loss": 0.7482,
+      "step": 4366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056665519337663e-05,
+      "loss": 0.7566,
+      "step": 4367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3025886684430467e-05,
+      "loss": 0.7327,
+      "step": 4368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299514164267267e-05,
+      "loss": 0.7439,
+      "step": 4369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.296443040601003e-05,
+      "loss": 0.736,
+      "step": 4370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.293375298637518e-05,
+      "loss": 0.6711,
+      "step": 4371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2903109395687596e-05,
+      "loss": 0.7598,
+      "step": 4372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2872499645853608e-05,
+      "loss": 0.6973,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2841923748766427e-05,
+      "loss": 0.7854,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2811381716306092e-05,
+      "loss": 0.7167,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2780873560339468e-05,
+      "loss": 0.731,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2750399292720283e-05,
+      "loss": 0.7693,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2719958925289088e-05,
+      "loss": 0.7163,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2689552469873267e-05,
+      "loss": 0.7581,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2659179938287035e-05,
+      "loss": 0.7432,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2628841342331388e-05,
+      "loss": 0.7134,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598536693794172e-05,
+      "loss": 0.7363,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2568266004450035e-05,
+      "loss": 0.7092,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2538029286060426e-05,
+      "loss": 0.7219,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2507826550373613e-05,
+      "loss": 0.7441,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2477657809124631e-05,
+      "loss": 0.7683,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2447523074035339e-05,
+      "loss": 0.7515,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2417422356814346e-05,
+      "loss": 0.707,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2387355669157075e-05,
+      "loss": 0.7598,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2357323022745715e-05,
+      "loss": 0.7812,
+      "step": 4390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2327324429249232e-05,
+      "loss": 0.7583,
+      "step": 4391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2297359900323347e-05,
+      "loss": 0.7141,
+      "step": 4392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2267429447610556e-05,
+      "loss": 0.7488,
+      "step": 4393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2237533082740126e-05,
+      "loss": 0.729,
+      "step": 4394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2207670817328066e-05,
+      "loss": 0.3694,
+      "step": 4395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2177842662977135e-05,
+      "loss": 0.7314,
+      "step": 4396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2148048631276877e-05,
+      "loss": 0.7207,
+      "step": 4397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2118288733803473e-05,
+      "loss": 0.7971,
+      "step": 4398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2088562982119955e-05,
+      "loss": 0.6865,
+      "step": 4399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2058871387776038e-05,
+      "loss": 0.7603,
+      "step": 4400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2029213962308172e-05,
+      "loss": 0.7671,
+      "step": 4401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1999590717239528e-05,
+      "loss": 0.7534,
+      "step": 4402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.197000166408001e-05,
+      "loss": 0.7246,
+      "step": 4403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19404468143262e-05,
+      "loss": 0.7395,
+      "step": 4404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1910926179461446e-05,
+      "loss": 0.7192,
+      "step": 4405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188143977095576e-05,
+      "loss": 0.7427,
+      "step": 4406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1851987600265869e-05,
+      "loss": 0.7068,
+      "step": 4407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1822569678835194e-05,
+      "loss": 0.7949,
+      "step": 4408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.179318601809386e-05,
+      "loss": 0.6877,
+      "step": 4409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1763836629458658e-05,
+      "loss": 0.7373,
+      "step": 4410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1734521524333086e-05,
+      "loss": 0.7185,
+      "step": 4411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1705240714107302e-05,
+      "loss": 0.7695,
+      "step": 4412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1675994210158181e-05,
+      "loss": 0.7488,
+      "step": 4413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.164678202384919e-05,
+      "loss": 0.7773,
+      "step": 4414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1617604166530528e-05,
+      "loss": 0.7224,
+      "step": 4415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1588460649539035e-05,
+      "loss": 0.7396,
+      "step": 4416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1559351484198211e-05,
+      "loss": 0.7515,
+      "step": 4417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1530276681818208e-05,
+      "loss": 0.76,
+      "step": 4418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1501236253695823e-05,
+      "loss": 0.3656,
+      "step": 4419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1472230211114499e-05,
+      "loss": 0.3533,
+      "step": 4420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1443258565344329e-05,
+      "loss": 0.3427,
+      "step": 4421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141432132764202e-05,
+      "loss": 0.7048,
+      "step": 4422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1385418509250933e-05,
+      "loss": 0.7483,
+      "step": 4423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1356550121401033e-05,
+      "loss": 0.3944,
+      "step": 4424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1327716175308934e-05,
+      "loss": 0.7429,
+      "step": 4425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.129891668217783e-05,
+      "loss": 0.3575,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1270151653197569e-05,
+      "loss": 0.6973,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124142109954459e-05,
+      "loss": 0.7405,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1212725032381933e-05,
+      "loss": 0.7495,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1184063462859262e-05,
+      "loss": 0.7284,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1155436402112785e-05,
+      "loss": 0.7629,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1126843861265346e-05,
+      "loss": 0.7615,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.109828585142637e-05,
+      "loss": 0.6992,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1069762383691862e-05,
+      "loss": 0.7483,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1041273469144397e-05,
+      "loss": 0.7129,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1012819118853147e-05,
+      "loss": 0.7532,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.098439934387383e-05,
+      "loss": 0.7717,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.095601415524874e-05,
+      "loss": 0.7207,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092766356400673e-05,
+      "loss": 0.7397,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0899347581163221e-05,
+      "loss": 0.6904,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0871066217720172e-05,
+      "loss": 0.7637,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.08428194846661e-05,
+      "loss": 0.3522,
+      "step": 4442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0814607392976061e-05,
+      "loss": 0.6924,
+      "step": 4443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0786429953611666e-05,
+      "loss": 0.7324,
+      "step": 4444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075828717752103e-05,
+      "loss": 0.7227,
+      "step": 4445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0730179075638868e-05,
+      "loss": 0.7324,
+      "step": 4446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0702105658886319e-05,
+      "loss": 0.7576,
+      "step": 4447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0674066938171123e-05,
+      "loss": 0.7188,
+      "step": 4448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0646062924387512e-05,
+      "loss": 0.7419,
+      "step": 4449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061809362841626e-05,
+      "loss": 0.7217,
+      "step": 4450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0590159061124605e-05,
+      "loss": 0.7634,
+      "step": 4451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0562259233366334e-05,
+      "loss": 0.7102,
+      "step": 4452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0534394155981709e-05,
+      "loss": 0.7705,
+      "step": 4453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0506563839797501e-05,
+      "loss": 0.7334,
+      "step": 4454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0478768295626983e-05,
+      "loss": 0.7271,
+      "step": 4455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0451007534269907e-05,
+      "loss": 0.7415,
+      "step": 4456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.04232815665125e-05,
+      "loss": 0.7583,
+      "step": 4457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0395590403127486e-05,
+      "loss": 0.7241,
+      "step": 4458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0367934054874073e-05,
+      "loss": 0.3705,
+      "step": 4459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.034031253249792e-05,
+      "loss": 0.7708,
+      "step": 4460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0312725846731175e-05,
+      "loss": 0.7808,
+      "step": 4461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.028517400829243e-05,
+      "loss": 0.6846,
+      "step": 4462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0257657027886768e-05,
+      "loss": 0.7297,
+      "step": 4463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.023017491620568e-05,
+      "loss": 0.7527,
+      "step": 4464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.020272768392715e-05,
+      "loss": 0.7214,
+      "step": 4465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0175315341715597e-05,
+      "loss": 0.7437,
+      "step": 4466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0147937900221883e-05,
+      "loss": 0.7693,
+      "step": 4467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0120595370083318e-05,
+      "loss": 0.7642,
+      "step": 4468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.009328776192363e-05,
+      "loss": 0.7537,
+      "step": 4469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0066015086352997e-05,
+      "loss": 0.7263,
+      "step": 4470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003877735396801e-05,
+      "loss": 0.7075,
+      "step": 4471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011574575351701e-05,
+      "loss": 0.7534,
+      "step": 4472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.984406761073494e-06,
+      "loss": 0.7292,
+      "step": 4473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.957273921689237e-06,
+      "loss": 0.7327,
+      "step": 4474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.930176067741216e-06,
+      "loss": 0.8042,
+      "step": 4475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758096e-06,
+      "loss": 0.7512,
+      "step": 4476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.876085358254939e-06,
+      "loss": 0.681,
+      "step": 4477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.849092523733216e-06,
+      "loss": 0.7959,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.822134716680831e-06,
+      "loss": 0.8101,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795211947571991e-06,
+      "loss": 0.6892,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.768324226867353e-06,
+      "loss": 0.7429,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74147156501396e-06,
+      "loss": 0.3969,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.714653972445198e-06,
+      "loss": 0.7441,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.687871459580844e-06,
+      "loss": 0.3707,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.661124036827063e-06,
+      "loss": 0.3823,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.634411714576353e-06,
+      "loss": 0.6797,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.607734503207589e-06,
+      "loss": 0.7417,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58109241308599e-06,
+      "loss": 0.7271,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55448545456319e-06,
+      "loss": 0.7532,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.527913637977092e-06,
+      "loss": 0.7307,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.501376973651999e-06,
+      "loss": 0.7202,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.474875471898526e-06,
+      "loss": 0.7822,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.448409143013626e-06,
+      "loss": 0.7358,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.421977997280596e-06,
+      "loss": 0.7146,
+      "step": 4494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.395582044969075e-06,
+      "loss": 0.7153,
+      "step": 4495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.369221296335006e-06,
+      "loss": 0.718,
+      "step": 4496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.342895761620662e-06,
+      "loss": 0.7207,
+      "step": 4497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.31660545105465e-06,
+      "loss": 0.7773,
+      "step": 4498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.290350374851864e-06,
+      "loss": 0.7888,
+      "step": 4499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.264130543213512e-06,
+      "loss": 0.7498,
+      "step": 4500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.237945966327133e-06,
+      "loss": 0.6638,
+      "step": 4501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.211796654366545e-06,
+      "loss": 0.7622,
+      "step": 4502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.185682617491863e-06,
+      "loss": 0.7004,
+      "step": 4503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15960386584952e-06,
+      "loss": 0.7178,
+      "step": 4504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.133560409572194e-06,
+      "loss": 0.7236,
+      "step": 4505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.107552258778907e-06,
+      "loss": 0.6797,
+      "step": 4506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081579423574916e-06,
+      "loss": 0.7322,
+      "step": 4507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.055641914051782e-06,
+      "loss": 0.7263,
+      "step": 4508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029739740287357e-06,
+      "loss": 0.7329,
+      "step": 4509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.00387291234569e-06,
+      "loss": 0.7697,
+      "step": 4510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.978041440277163e-06,
+      "loss": 0.696,
+      "step": 4511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.952245334118414e-06,
+      "loss": 0.7122,
+      "step": 4512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.926484603892327e-06,
+      "loss": 0.6703,
+      "step": 4513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900759259608038e-06,
+      "loss": 0.7299,
+      "step": 4514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.875069311260941e-06,
+      "loss": 0.7415,
+      "step": 4515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.849414768832687e-06,
+      "loss": 0.7583,
+      "step": 4516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.823795642291143e-06,
+      "loss": 0.7156,
+      "step": 4517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79821194159044e-06,
+      "loss": 0.7183,
+      "step": 4518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.772663676670934e-06,
+      "loss": 0.7397,
+      "step": 4519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.747150857459218e-06,
+      "loss": 0.7268,
+      "step": 4520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.72167349386811e-06,
+      "loss": 0.7456,
+      "step": 4521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.696231595796634e-06,
+      "loss": 0.3792,
+      "step": 4522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.670825173130082e-06,
+      "loss": 0.6877,
+      "step": 4523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-06,
+      "loss": 0.7485,
+      "step": 4524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.620118793483822e-06,
+      "loss": 0.7253,
+      "step": 4525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.5948188562057e-06,
+      "loss": 0.7139,
+      "step": 4526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.56955443373565e-06,
+      "loss": 0.7173,
+      "step": 4527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.544325535889996e-06,
+      "loss": 0.6948,
+      "step": 4528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.519132172471211e-06,
+      "loss": 0.7292,
+      "step": 4529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.493974353268019e-06,
+      "loss": 0.7776,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.46885208805529e-06,
+      "loss": 0.7346,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.443765386594094e-06,
+      "loss": 0.7603,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.418714258631688e-06,
+      "loss": 0.8091,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.3936987139015e-06,
+      "loss": 0.7112,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.36871876212314e-06,
+      "loss": 0.7207,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.343774413002381e-06,
+      "loss": 0.7412,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.318865676231158e-06,
+      "loss": 0.7046,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.293992561487596e-06,
+      "loss": 0.7217,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.269155078435931e-06,
+      "loss": 0.7344,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.24435323672661e-06,
+      "loss": 0.7444,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.219587045996203e-06,
+      "loss": 0.7991,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.194856515867432e-06,
+      "loss": 0.7522,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.170161655949138e-06,
+      "loss": 0.7432,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.14550247583633e-06,
+      "loss": 0.7332,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.12087898511018e-06,
+      "loss": 0.3711,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.096291193337934e-06,
+      "loss": 0.7236,
+      "step": 4546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.071739110073018e-06,
+      "loss": 0.7654,
+      "step": 4547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.047222744854943e-06,
+      "loss": 0.707,
+      "step": 4548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.022742107209369e-06,
+      "loss": 0.7058,
+      "step": 4549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.998297206648076e-06,
+      "loss": 0.7505,
+      "step": 4550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.973888052668943e-06,
+      "loss": 0.6951,
+      "step": 4551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.949514654755962e-06,
+      "loss": 0.7854,
+      "step": 4552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.92517702237925e-06,
+      "loss": 0.7354,
+      "step": 4553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.900875164995004e-06,
+      "loss": 0.7231,
+      "step": 4554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876609092045528e-06,
+      "loss": 0.718,
+      "step": 4555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.852378812959227e-06,
+      "loss": 0.6914,
+      "step": 4556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.828184337150613e-06,
+      "loss": 0.7224,
+      "step": 4557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.804025674020266e-06,
+      "loss": 0.6929,
+      "step": 4558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.779902832954833e-06,
+      "loss": 0.7725,
+      "step": 4559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.755815823327084e-06,
+      "loss": 0.8037,
+      "step": 4560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.731764654495832e-06,
+      "loss": 0.7439,
+      "step": 4561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.707749335806014e-06,
+      "loss": 0.7639,
+      "step": 4562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.683769876588576e-06,
+      "loss": 0.7122,
+      "step": 4563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.659826286160566e-06,
+      "loss": 0.7485,
+      "step": 4564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.635918573825096e-06,
+      "loss": 0.7344,
+      "step": 4565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 0.7104,
+      "step": 4566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.588210820574482e-06,
+      "loss": 0.7363,
+      "step": 4567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564410798195831e-06,
+      "loss": 0.7549,
+      "step": 4568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5406466909827116e-06,
+      "loss": 0.77,
+      "step": 4569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.516918508168469e-06,
+      "loss": 0.7126,
+      "step": 4570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.493226258972519e-06,
+      "loss": 0.6683,
+      "step": 4571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46956995260033e-06,
+      "loss": 0.6807,
+      "step": 4572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.445949598243363e-06,
+      "loss": 0.7603,
+      "step": 4573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.42236520507914e-06,
+      "loss": 0.7197,
+      "step": 4574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.398816782271223e-06,
+      "loss": 0.7131,
+      "step": 4575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.375304338969136e-06,
+      "loss": 0.339,
+      "step": 4576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.351827884308493e-06,
+      "loss": 0.3676,
+      "step": 4577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3283874274108764e-06,
+      "loss": 0.3546,
+      "step": 4578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.304982977383934e-06,
+      "loss": 0.7095,
+      "step": 4579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281614543321269e-06,
+      "loss": 0.8022,
+      "step": 4580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.258282134302519e-06,
+      "loss": 0.348,
+      "step": 4581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2349857593933205e-06,
+      "loss": 0.7224,
+      "step": 4582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21172542764531e-06,
+      "loss": 0.7559,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1885011480961164e-06,
+      "loss": 0.7252,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.165312929769352e-06,
+      "loss": 0.3572,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.142160781674645e-06,
+      "loss": 0.8252,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119044712807577e-06,
+      "loss": 0.7622,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.09596473214974e-06,
+      "loss": 0.7759,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.072920848668674e-06,
+      "loss": 0.3732,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.049913071317926e-06,
+      "loss": 0.7295,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.026941409036991e-06,
+      "loss": 0.7224,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.00400587075134e-06,
+      "loss": 0.7314,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.981106465372389e-06,
+      "loss": 0.7097,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.958243201797554e-06,
+      "loss": 0.7148,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.935416088910185e-06,
+      "loss": 0.3367,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9126251355795864e-06,
+      "loss": 0.7041,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.889870350661009e-06,
+      "loss": 0.769,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.867151742995659e-06,
+      "loss": 0.7544,
+      "step": 4598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.844469321410685e-06,
+      "loss": 0.7742,
+      "step": 4599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8218230947191705e-06,
+      "loss": 0.7732,
+      "step": 4600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7992130717201564e-06,
+      "loss": 0.7595,
+      "step": 4601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.776639261198581e-06,
+      "loss": 0.3413,
+      "step": 4602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7541016719253395e-06,
+      "loss": 0.377,
+      "step": 4603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.731600312657238e-06,
+      "loss": 0.7495,
+      "step": 4604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.709135192137017e-06,
+      "loss": 0.7441,
+      "step": 4605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68670631909335e-06,
+      "loss": 0.7764,
+      "step": 4606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6643137022407876e-06,
+      "loss": 0.7434,
+      "step": 4607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6419573502798374e-06,
+      "loss": 0.7852,
+      "step": 4608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.619637271896873e-06,
+      "loss": 0.7336,
+      "step": 4609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.597353475764201e-06,
+      "loss": 0.7527,
+      "step": 4610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5751059705400295e-06,
+      "loss": 0.7488,
+      "step": 4611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.552894764868456e-06,
+      "loss": 0.335,
+      "step": 4612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530719867379498e-06,
+      "loss": 0.7947,
+      "step": 4613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.508581286689031e-06,
+      "loss": 0.7434,
+      "step": 4614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.48647903139884e-06,
+      "loss": 0.6929,
+      "step": 4615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.464413110096601e-06,
+      "loss": 0.761,
+      "step": 4616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4423835313558555e-06,
+      "loss": 0.7771,
+      "step": 4617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420390303736035e-06,
+      "loss": 0.7462,
+      "step": 4618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.398433435782458e-06,
+      "loss": 0.7363,
+      "step": 4619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.37651293602628e-06,
+      "loss": 0.7368,
+      "step": 4620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.354628812984576e-06,
+      "loss": 0.7031,
+      "step": 4621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.332781075160243e-06,
+      "loss": 0.7664,
+      "step": 4622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.31096973104206e-06,
+      "loss": 0.7524,
+      "step": 4623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.28919478910468e-06,
+      "loss": 0.6824,
+      "step": 4624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.267456257808579e-06,
+      "loss": 0.7629,
+      "step": 4625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.245754145600091e-06,
+      "loss": 0.761,
+      "step": 4626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.224088460911437e-06,
+      "loss": 0.7192,
+      "step": 4627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.202459212160638e-06,
+      "loss": 0.7249,
+      "step": 4628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.180866407751595e-06,
+      "loss": 0.7256,
+      "step": 4629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.15931005607403e-06,
+      "loss": 0.6914,
+      "step": 4630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.137790165503499e-06,
+      "loss": 0.7285,
+      "step": 4631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1163067444013895e-06,
+      "loss": 0.7272,
+      "step": 4632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.094859801114949e-06,
+      "loss": 0.7616,
+      "step": 4633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.073449343977211e-06,
+      "loss": 0.3779,
+      "step": 4634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052075381307054e-06,
+      "loss": 0.6868,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.030737921409169e-06,
+      "loss": 0.7683,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0094369725740786e-06,
+      "loss": 0.752,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.988172543078097e-06,
+      "loss": 0.7048,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.96694464118337e-06,
+      "loss": 0.6936,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.945753275137844e-06,
+      "loss": 0.8026,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9245984531752784e-06,
+      "loss": 0.7289,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.903480183515198e-06,
+      "loss": 0.7209,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.882398474362949e-06,
+      "loss": 0.7136,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.861353333909692e-06,
+      "loss": 0.7581,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.840344770332351e-06,
+      "loss": 0.7288,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8193727917936536e-06,
+      "loss": 0.7561,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7984374064421035e-06,
+      "loss": 0.739,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.777538622412004e-06,
+      "loss": 0.7456,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756676447823406e-06,
+      "loss": 0.7141,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.735850890782157e-06,
+      "loss": 0.7554,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.715061959379875e-06,
+      "loss": 0.8209,
+      "step": 4651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.694309661693942e-06,
+      "loss": 0.7302,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.673594005787519e-06,
+      "loss": 0.8096,
+      "step": 4653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.652914999709535e-06,
+      "loss": 0.7727,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.63227265149463e-06,
+      "loss": 0.6985,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.611666969163243e-06,
+      "loss": 0.3666,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5910979607215805e-06,
+      "loss": 0.6902,
+      "step": 4657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.570565634161562e-06,
+      "loss": 0.713,
+      "step": 4658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.550069997460871e-06,
+      "loss": 0.8154,
+      "step": 4659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5296110585829506e-06,
+      "loss": 0.7659,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509188825476963e-06,
+      "loss": 0.7031,
+      "step": 4661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.488803306077806e-06,
+      "loss": 0.7048,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.468454508306131e-06,
+      "loss": 0.3209,
+      "step": 4663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.448142440068316e-06,
+      "loss": 0.7502,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.427867109256457e-06,
+      "loss": 0.7769,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4076285237483984e-06,
+      "loss": 0.7739,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.387426691407682e-06,
+      "loss": 0.6949,
+      "step": 4667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367261620083575e-06,
+      "loss": 0.7554,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.347133317611075e-06,
+      "loss": 0.7441,
+      "step": 4669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.327041791810905e-06,
+      "loss": 0.7944,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.306987050489442e-06,
+      "loss": 0.7834,
+      "step": 4671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2869691014388215e-06,
+      "loss": 0.76,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266987952436875e-06,
+      "loss": 0.74,
+      "step": 4673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.247043611247127e-06,
+      "loss": 0.7207,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2271360856188e-06,
+      "loss": 0.759,
+      "step": 4675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.20726538328683e-06,
+      "loss": 0.7646,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.187431511971819e-06,
+      "loss": 0.7732,
+      "step": 4677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.167634479380068e-06,
+      "loss": 0.7083,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.14787429320358e-06,
+      "loss": 0.7307,
+      "step": 4679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.128150961120026e-06,
+      "loss": 0.3501,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1084644907927526e-06,
+      "loss": 0.3782,
+      "step": 4681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.088814889870808e-06,
+      "loss": 0.6885,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0692021659888735e-06,
+      "loss": 0.7903,
+      "step": 4683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.049626326767365e-06,
+      "loss": 0.7787,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.030087379812298e-06,
+      "loss": 0.782,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0105853327154e-06,
+      "loss": 0.762,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.991120193054055e-06,
+      "loss": 0.7473,
+      "step": 4687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.97169196839129e-06,
+      "loss": 0.74,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9523006662757885e-06,
+      "loss": 0.7461,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.932946294241903e-06,
+      "loss": 0.752,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.913628859809638e-06,
+      "loss": 0.7063,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8943483704846475e-06,
+      "loss": 0.7664,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875104833758204e-06,
+      "loss": 0.7192,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.855898257107261e-06,
+      "loss": 0.7593,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.836728647994371e-06,
+      "loss": 0.701,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.817596013867764e-06,
+      "loss": 0.7391,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7985003621612735e-06,
+      "loss": 0.7114,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.779441700294362e-06,
+      "loss": 0.6733,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.760420035672164e-06,
+      "loss": 0.7458,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.741435375685377e-06,
+      "loss": 0.7942,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.722487727710368e-06,
+      "loss": 0.3546,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.703577099109091e-06,
+      "loss": 0.7188,
+      "step": 4702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.684703497229159e-06,
+      "loss": 0.7505,
+      "step": 4703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.665866929403739e-06,
+      "loss": 0.371,
+      "step": 4704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.647067402951655e-06,
+      "loss": 0.7961,
+      "step": 4705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.628304925177318e-06,
+      "loss": 0.7388,
+      "step": 4706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.609579503370754e-06,
+      "loss": 0.781,
+      "step": 4707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.590891144807574e-06,
+      "loss": 0.7129,
+      "step": 4708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.572239856749028e-06,
+      "loss": 0.6996,
+      "step": 4709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.553625646441928e-06,
+      "loss": 0.7231,
+      "step": 4710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5350485211186675e-06,
+      "loss": 0.7291,
+      "step": 4711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.516508487997284e-06,
+      "loss": 0.7854,
+      "step": 4712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.498005554281337e-06,
+      "loss": 0.7168,
+      "step": 4713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4795397271600315e-06,
+      "loss": 0.7717,
+      "step": 4714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.461111013808117e-06,
+      "loss": 0.7896,
+      "step": 4715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.442719421385922e-06,
+      "loss": 0.7759,
+      "step": 4716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.424364957039384e-06,
+      "loss": 0.7859,
+      "step": 4717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.406047627899989e-06,
+      "loss": 0.7415,
+      "step": 4718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3877674410848e-06,
+      "loss": 0.7328,
+      "step": 4719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.369524403696457e-06,
+      "loss": 0.7874,
+      "step": 4720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351318522823133e-06,
+      "loss": 0.3517,
+      "step": 4721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.333149805538606e-06,
+      "loss": 0.7732,
+      "step": 4722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.315018258902193e-06,
+      "loss": 0.72,
+      "step": 4723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.29692388995877e-06,
+      "loss": 0.7249,
+      "step": 4724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.278866705738782e-06,
+      "loss": 0.7166,
+      "step": 4725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.260846713258193e-06,
+      "loss": 0.7317,
+      "step": 4726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.242863919518558e-06,
+      "loss": 0.7756,
+      "step": 4727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.224918331506955e-06,
+      "loss": 0.7368,
+      "step": 4728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.207009956196007e-06,
+      "loss": 0.7437,
+      "step": 4729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.189138800543868e-06,
+      "loss": 0.7629,
+      "step": 4730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.171304871494264e-06,
+      "loss": 0.7244,
+      "step": 4731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.153508175976428e-06,
+      "loss": 0.804,
+      "step": 4732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.135748720905142e-06,
+      "loss": 0.7188,
+      "step": 4733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.118026513180695e-06,
+      "loss": 0.7678,
+      "step": 4734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100341559688936e-06,
+      "loss": 0.3256,
+      "step": 4735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.082693867301224e-06,
+      "loss": 0.6764,
+      "step": 4736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.065083442874418e-06,
+      "loss": 0.7335,
+      "step": 4737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.047510293250933e-06,
+      "loss": 0.7712,
+      "step": 4738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029974425258687e-06,
+      "loss": 0.6992,
+      "step": 4739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.012475845711106e-06,
+      "loss": 0.731,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99501456140714e-06,
+      "loss": 0.7218,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.977590579131241e-06,
+      "loss": 0.7296,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960203905653371e-06,
+      "loss": 0.759,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.942854547728992e-06,
+      "loss": 0.7291,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.925542512099056e-06,
+      "loss": 0.7583,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908267805490051e-06,
+      "loss": 0.6836,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.891030434613918e-06,
+      "loss": 0.7036,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.873830406168111e-06,
+      "loss": 0.707,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8566677268355945e-06,
+      "loss": 0.7598,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.839542403284768e-06,
+      "loss": 0.6914,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.822454442169576e-06,
+      "loss": 0.7219,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.805403850129408e-06,
+      "loss": 0.3804,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7883906337891626e-06,
+      "loss": 0.7859,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.771414799759165e-06,
+      "loss": 0.729,
+      "step": 4754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7544763546352834e-06,
+      "loss": 0.7283,
+      "step": 4755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7375753049987973e-06,
+      "loss": 0.7026,
+      "step": 4756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.720711657416509e-06,
+      "loss": 0.7578,
+      "step": 4757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.703885418440656e-06,
+      "loss": 0.7346,
+      "step": 4758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.687096594608941e-06,
+      "loss": 0.7568,
+      "step": 4759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6703451924445465e-06,
+      "loss": 0.7573,
+      "step": 4760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6536312184560993e-06,
+      "loss": 0.7412,
+      "step": 4761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6369546791377052e-06,
+      "loss": 0.7188,
+      "step": 4762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6203155809688914e-06,
+      "loss": 0.7634,
+      "step": 4763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6037139304146762e-06,
+      "loss": 0.75,
+      "step": 4764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.587149733925488e-06,
+      "loss": 0.7229,
+      "step": 4765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.570622997937234e-06,
+      "loss": 0.7329,
+      "step": 4766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5541337288712563e-06,
+      "loss": 0.7576,
+      "step": 4767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5376819331343403e-06,
+      "loss": 0.728,
+      "step": 4768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5212676171187064e-06,
+      "loss": 0.7847,
+      "step": 4769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5048907872020197e-06,
+      "loss": 0.7186,
+      "step": 4770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.488551449747357e-06,
+      "loss": 0.7336,
+      "step": 4771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4722496111032733e-06,
+      "loss": 0.74,
+      "step": 4772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.455985277603713e-06,
+      "loss": 0.6993,
+      "step": 4773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.439758455568065e-06,
+      "loss": 0.7617,
+      "step": 4774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4235691513011424e-06,
+      "loss": 0.7229,
+      "step": 4775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40741737109318e-06,
+      "loss": 0.7598,
+      "step": 4776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3913031212198354e-06,
+      "loss": 0.7515,
+      "step": 4777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.375226407942178e-06,
+      "loss": 0.7432,
+      "step": 4778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.359187237506689e-06,
+      "loss": 0.7228,
+      "step": 4779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3431856161452835e-06,
+      "loss": 0.7622,
+      "step": 4780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3272215500752658e-06,
+      "loss": 0.3536,
+      "step": 4781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.311295045499363e-06,
+      "loss": 0.7524,
+      "step": 4782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.295406108605692e-06,
+      "loss": 0.7157,
+      "step": 4783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.279554745567781e-06,
+      "loss": 0.7092,
+      "step": 4784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.26374096254457e-06,
+      "loss": 0.74,
+      "step": 4785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247964765680389e-06,
+      "loss": 0.7632,
+      "step": 4786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2322261611049453e-06,
+      "loss": 0.7732,
+      "step": 4787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2165251549333587e-06,
+      "loss": 0.7288,
+      "step": 4788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.20086175326616e-06,
+      "loss": 0.7327,
+      "step": 4789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1852359621892367e-06,
+      "loss": 0.7466,
+      "step": 4790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.169647787773866e-06,
+      "loss": 0.7327,
+      "step": 4791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1540972360767252e-06,
+      "loss": 0.392,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1385843131398698e-06,
+      "loss": 0.3425,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1231090249907337e-06,
+      "loss": 0.7974,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1076713776421185e-06,
+      "loss": 0.7744,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.092271377092215e-06,
+      "loss": 0.7463,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0769090293245705e-06,
+      "loss": 0.7102,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0615843403081103e-06,
+      "loss": 0.7373,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0462973159971374e-06,
+      "loss": 0.7693,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0310479623313127e-06,
+      "loss": 0.8352,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0158362852356626e-06,
+      "loss": 0.7417,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0006622906205708e-06,
+      "loss": 0.4098,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9855259843817542e-06,
+      "loss": 0.7178,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.970427372400353e-06,
+      "loss": 0.761,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9553664605427965e-06,
+      "loss": 0.7705,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9403432546609043e-06,
+      "loss": 0.7051,
+      "step": 4806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.925357760591818e-06,
+      "loss": 0.7102,
+      "step": 4807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.910409984158058e-06,
+      "loss": 0.7329,
+      "step": 4808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8954999311674558e-06,
+      "loss": 0.7197,
+      "step": 4809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.880627607413211e-06,
+      "loss": 0.7368,
+      "step": 4810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8657930186738567e-06,
+      "loss": 0.7654,
+      "step": 4811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8509961707132494e-06,
+      "loss": 0.7366,
+      "step": 4812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8362370692806116e-06,
+      "loss": 0.7424,
+      "step": 4813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.821515720110479e-06,
+      "loss": 0.6958,
+      "step": 4814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.806832128922732e-06,
+      "loss": 0.7627,
+      "step": 4815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7921863014225503e-06,
+      "loss": 0.7136,
+      "step": 4816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7775782433004717e-06,
+      "loss": 0.7419,
+      "step": 4817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7630079602323442e-06,
+      "loss": 0.7808,
+      "step": 4818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7484754578793513e-06,
+      "loss": 0.3738,
+      "step": 4819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733980741887987e-06,
+      "loss": 0.7654,
+      "step": 4820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7195238178900684e-06,
+      "loss": 0.7227,
+      "step": 4821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7051046915027246e-06,
+      "loss": 0.7236,
+      "step": 4822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6907233683284074e-06,
+      "loss": 0.3428,
+      "step": 4823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.676379853954858e-06,
+      "loss": 0.6764,
+      "step": 4824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.662074153955152e-06,
+      "loss": 0.6753,
+      "step": 4825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.647806273887665e-06,
+      "loss": 0.7001,
+      "step": 4826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6335762192960743e-06,
+      "loss": 0.6846,
+      "step": 4827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.619383995709368e-06,
+      "loss": 0.7705,
+      "step": 4828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.605229608641824e-06,
+      "loss": 0.7449,
+      "step": 4829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.59111306359302e-06,
+      "loss": 0.7312,
+      "step": 4830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.577034366047848e-06,
+      "loss": 0.709,
+      "step": 4831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5629935214764865e-06,
+      "loss": 0.7236,
+      "step": 4832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5489905353343723e-06,
+      "loss": 0.7451,
+      "step": 4833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.535025413062286e-06,
+      "loss": 0.7556,
+      "step": 4834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.521098160086266e-06,
+      "loss": 0.7147,
+      "step": 4835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5072087818176382e-06,
+      "loss": 0.7839,
+      "step": 4836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4933572836530305e-06,
+      "loss": 0.7412,
+      "step": 4837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.479543670974338e-06,
+      "loss": 0.7434,
+      "step": 4838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.465767949148734e-06,
+      "loss": 0.7322,
+      "step": 4839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.45203012352867e-06,
+      "loss": 0.7253,
+      "step": 4840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.438330199451877e-06,
+      "loss": 0.7024,
+      "step": 4841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.424668182241363e-06,
+      "loss": 0.7808,
+      "step": 4842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4110440772054156e-06,
+      "loss": 0.7302,
+      "step": 4843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3974578896375553e-06,
+      "loss": 0.7271,
+      "step": 4844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.383909624816616e-06,
+      "loss": 0.7881,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3703992880066638e-06,
+      "loss": 0.7102,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3569268844570445e-06,
+      "loss": 0.7751,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.343492419402371e-06,
+      "loss": 0.7352,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.330095898062479e-06,
+      "loss": 0.7585,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.316737325642515e-06,
+      "loss": 0.7241,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3034167073328284e-06,
+      "loss": 0.7466,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2901340483090782e-06,
+      "loss": 0.7734,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2768893537321147e-06,
+      "loss": 0.7563,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2636826287480873e-06,
+      "loss": 0.7334,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2505138784883586e-06,
+      "loss": 0.7361,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.237383108069546e-06,
+      "loss": 0.7122,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.224290322593525e-06,
+      "loss": 0.7632,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.211235527147404e-06,
+      "loss": 0.7595,
+      "step": 4858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1982187268035136e-06,
+      "loss": 0.7356,
+      "step": 4859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194314e-06,
+      "loss": 0.7395,
+      "step": 4860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1722991316380003e-06,
+      "loss": 0.7146,
+      "step": 4861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1593963468872434e-06,
+      "loss": 0.7708,
+      "step": 4862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1465315773804616e-06,
+      "loss": 0.7251,
+      "step": 4863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1337048281161563e-06,
+      "loss": 0.731,
+      "step": 4864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1209161040780747e-06,
+      "loss": 0.7039,
+      "step": 4865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1081654102351635e-06,
+      "loss": 0.7158,
+      "step": 4866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0954527515416156e-06,
+      "loss": 0.6707,
+      "step": 4867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.082778132936858e-06,
+      "loss": 0.7029,
+      "step": 4868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0701415593455064e-06,
+      "loss": 0.7441,
+      "step": 4869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0575430356774118e-06,
+      "loss": 0.7324,
+      "step": 4870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0449825668276244e-06,
+      "loss": 0.7466,
+      "step": 4871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.032460157676452e-06,
+      "loss": 0.3337,
+      "step": 4872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0199758130893475e-06,
+      "loss": 0.7715,
+      "step": 4873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0075295379170412e-06,
+      "loss": 0.709,
+      "step": 4874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.99512133699542e-06,
+      "loss": 0.7625,
+      "step": 4875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9827512151456173e-06,
+      "loss": 0.714,
+      "step": 4876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.970419177173932e-06,
+      "loss": 0.7551,
+      "step": 4877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.95812522787191e-06,
+      "loss": 0.6936,
+      "step": 4878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9458693720162513e-06,
+      "loss": 0.7539,
+      "step": 4879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.933651614368892e-06,
+      "loss": 0.749,
+      "step": 4880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.921471959676957e-06,
+      "loss": 0.7158,
+      "step": 4881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9093304126727494e-06,
+      "loss": 0.7637,
+      "step": 4882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8972269780737739e-06,
+      "loss": 0.6926,
+      "step": 4883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.885161660582746e-06,
+      "loss": 0.7263,
+      "step": 4884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731344648875383e-06,
+      "loss": 0.7056,
+      "step": 4885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8611453956612347e-06,
+      "loss": 0.7378,
+      "step": 4886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.849194457562109e-06,
+      "loss": 0.707,
+      "step": 4887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8372816552336026e-06,
+      "loss": 0.7163,
+      "step": 4888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8254069933043571e-06,
+      "loss": 0.6946,
+      "step": 4889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.81357047638816e-06,
+      "loss": 0.7651,
+      "step": 4890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8017721090840322e-06,
+      "loss": 0.7495,
+      "step": 4891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.790011895976118e-06,
+      "loss": 0.7656,
+      "step": 4892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7782898416337734e-06,
+      "loss": 0.7126,
+      "step": 4893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7666059506115107e-06,
+      "loss": 0.793,
+      "step": 4894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7549602274490318e-06,
+      "loss": 0.7209,
+      "step": 4895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7433526766711728e-06,
+      "loss": 0.7437,
+      "step": 4896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7317833027879927e-06,
+      "loss": 0.6958,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7202521102946622e-06,
+      "loss": 0.6935,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7087591036715534e-06,
+      "loss": 0.7087,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6973042873841826e-06,
+      "loss": 0.7217,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6858876658832233e-06,
+      "loss": 0.7144,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6745092436045494e-06,
+      "loss": 0.6852,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6631690249691356e-06,
+      "loss": 0.7231,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.651867014383146e-06,
+      "loss": 0.3657,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6406032162379125e-06,
+      "loss": 0.7126,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6293776349098677e-06,
+      "loss": 0.3519,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6181902747606558e-06,
+      "loss": 0.7148,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6070411401370334e-06,
+      "loss": 0.7217,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5959302353709127e-06,
+      "loss": 0.7507,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.584857564779363e-06,
+      "loss": 0.6968,
+      "step": 4910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5738231326645758e-06,
+      "loss": 0.7493,
+      "step": 4911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5628269433138997e-06,
+      "loss": 0.7744,
+      "step": 4912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5518690009998283e-06,
+      "loss": 0.7603,
+      "step": 4913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5409493099799887e-06,
+      "loss": 0.7389,
+      "step": 4914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.530067874497143e-06,
+      "loss": 0.7527,
+      "step": 4915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5192246987791981e-06,
+      "loss": 0.7356,
+      "step": 4916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5084197870391837e-06,
+      "loss": 0.7109,
+      "step": 4917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4976531434752749e-06,
+      "loss": 0.7864,
+      "step": 4918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4869247722707591e-06,
+      "loss": 0.7424,
+      "step": 4919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4762346775940793e-06,
+      "loss": 0.7542,
+      "step": 4920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.465582863598791e-06,
+      "loss": 0.7107,
+      "step": 4921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4549693344235726e-06,
+      "loss": 0.7725,
+      "step": 4922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.444394094192225e-06,
+      "loss": 0.7046,
+      "step": 4923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4338571470137063e-06,
+      "loss": 0.7239,
+      "step": 4924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4233584969820413e-06,
+      "loss": 0.7798,
+      "step": 4925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4128981481764115e-06,
+      "loss": 0.8101,
+      "step": 4926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4024761046611211e-06,
+      "loss": 0.7561,
+      "step": 4927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3920923704855649e-06,
+      "loss": 0.7349,
+      "step": 4928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3817469496842595e-06,
+      "loss": 0.7896,
+      "step": 4929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3714398462768563e-06,
+      "loss": 0.7717,
+      "step": 4930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3611710642681076e-06,
+      "loss": 0.7262,
+      "step": 4931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.350940607647866e-06,
+      "loss": 0.7423,
+      "step": 4932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3407484803910853e-06,
+      "loss": 0.712,
+      "step": 4933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3305946864578645e-06,
+      "loss": 0.7185,
+      "step": 4934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3204792297933588e-06,
+      "loss": 0.7227,
+      "step": 4935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3104021143278911e-06,
+      "loss": 0.7859,
+      "step": 4936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.300363343976818e-06,
+      "loss": 0.7507,
+      "step": 4937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2903629226406423e-06,
+      "loss": 0.7576,
+      "step": 4938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2804008542049438e-06,
+      "loss": 0.7385,
+      "step": 4939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2704771425404382e-06,
+      "loss": 0.7004,
+      "step": 4940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2605917915028742e-06,
+      "loss": 0.6865,
+      "step": 4941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2507448049331572e-06,
+      "loss": 0.7385,
+      "step": 4942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.24093618665726e-06,
+      "loss": 0.71,
+      "step": 4943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.231165940486234e-06,
+      "loss": 0.7544,
+      "step": 4944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.221434070216243e-06,
+      "loss": 0.7603,
+      "step": 4945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2117405796285286e-06,
+      "loss": 0.7161,
+      "step": 4946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2020854724894448e-06,
+      "loss": 0.717,
+      "step": 4947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.192468752550402e-06,
+      "loss": 0.7252,
+      "step": 4948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1828904235478889e-06,
+      "loss": 0.7471,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1733504892035064e-06,
+      "loss": 0.7454,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.163848953223934e-06,
+      "loss": 0.7095,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1543858193009182e-06,
+      "loss": 0.7292,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1449610911112741e-06,
+      "loss": 0.7744,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1355747723169385e-06,
+      "loss": 0.7361,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1262268665648723e-06,
+      "loss": 0.7407,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1169173774871478e-06,
+      "loss": 0.7466,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1076463087008938e-06,
+      "loss": 0.7432,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0984136638083177e-06,
+      "loss": 0.3854,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0892194463966942e-06,
+      "loss": 0.7554,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0800636600383663e-06,
+      "loss": 0.7212,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0709463082907545e-06,
+      "loss": 0.7319,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0618673946963365e-06,
+      "loss": 0.7725,
+      "step": 4962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0528269227826459e-06,
+      "loss": 0.7335,
+      "step": 4963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0438248960623064e-06,
+      "loss": 0.7327,
+      "step": 4964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0348613180329757e-06,
+      "loss": 0.6912,
+      "step": 4965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0259361921774013e-06,
+      "loss": 0.6921,
+      "step": 4966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0170495219633646e-06,
+      "loss": 0.762,
+      "step": 4967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.008201310843704e-06,
+      "loss": 0.7471,
+      "step": 4968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.993915622563577e-07,
+      "loss": 0.7078,
+      "step": 4969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.906202796242547e-07,
+      "loss": 0.7502,
+      "step": 4970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.818874663554357e-07,
+      "loss": 0.6841,
+      "step": 4971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.731931258429638e-07,
+      "loss": 0.7275,
+      "step": 4972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.645372614649595e-07,
+      "loss": 0.7272,
+      "step": 4973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.55919876584599e-07,
+      "loss": 0.7067,
+      "step": 4974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.473409745500927e-07,
+      "loss": 0.7461,
+      "step": 4975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.388005586947191e-07,
+      "loss": 0.7471,
+      "step": 4976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.302986323368013e-07,
+      "loss": 0.7219,
+      "step": 4977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.218351987796858e-07,
+      "loss": 0.7291,
+      "step": 4978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.134102613117756e-07,
+      "loss": 0.7734,
+      "step": 4979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.0502382320653e-07,
+      "loss": 0.3588,
+      "step": 4980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.966758877224201e-07,
+      "loss": 0.3478,
+      "step": 4981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.88366458102996e-07,
+      "loss": 0.751,
+      "step": 4982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.800955375768194e-07,
+      "loss": 0.7336,
+      "step": 4983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.718631293574753e-07,
+      "loss": 0.7573,
+      "step": 4984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.636692366436383e-07,
+      "loss": 0.7236,
+      "step": 4985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.555138626189618e-07,
+      "loss": 0.7646,
+      "step": 4986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.473970104521556e-07,
+      "loss": 0.7332,
+      "step": 4987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.393186832969746e-07,
+      "loss": 0.3658,
+      "step": 4988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.31278884292186e-07,
+      "loss": 0.7288,
+      "step": 4989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.232776165616018e-07,
+      "loss": 0.7397,
+      "step": 4990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.153148832140466e-07,
+      "loss": 0.7219,
+      "step": 4991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.07390687343379e-07,
+      "loss": 0.7292,
+      "step": 4992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.995050320285024e-07,
+      "loss": 0.7527,
+      "step": 4993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.916579203333108e-07,
+      "loss": 0.7568,
+      "step": 4994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.838493553067538e-07,
+      "loss": 0.7539,
+      "step": 4995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.760793399827937e-07,
+      "loss": 0.7563,
+      "step": 4996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.683478773804042e-07,
+      "loss": 0.7578,
+      "step": 4997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.606549705035937e-07,
+      "loss": 0.3639,
+      "step": 4998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.530006223413711e-07,
+      "loss": 0.704,
+      "step": 4999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.453848358678017e-07,
+      "loss": 0.7141,
+      "step": 5000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.378076140419187e-07,
+      "loss": 0.7195,
+      "step": 5001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.302689598078116e-07,
+      "loss": 0.7227,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.227688760945594e-07,
+      "loss": 0.699,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.153073658162646e-07,
+      "loss": 0.7512,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.078844318720524e-07,
+      "loss": 0.7483,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.00500077146038e-07,
+      "loss": 0.3422,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.931543045073708e-07,
+      "loss": 0.7468,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.858471168101787e-07,
+      "loss": 0.7859,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.785785168936243e-07,
+      "loss": 0.7654,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.713485075818815e-07,
+      "loss": 0.7312,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.641570916840922e-07,
+      "loss": 0.7141,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.570042719944436e-07,
+      "loss": 0.6892,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.498900512921014e-07,
+      "loss": 0.7185,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.428144323412544e-07,
+      "loss": 0.7588,
+      "step": 5014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.357774178910702e-07,
+      "loss": 0.3526,
+      "step": 5015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.287790106757396e-07,
+      "loss": 0.6328,
+      "step": 5016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.218192134144429e-07,
+      "loss": 0.717,
+      "step": 5017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.148980288113504e-07,
+      "loss": 0.7573,
+      "step": 5018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.080154595556331e-07,
+      "loss": 0.7483,
+      "step": 5019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.011715083214741e-07,
+      "loss": 0.7439,
+      "step": 5020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.943661777680354e-07,
+      "loss": 0.6534,
+      "step": 5021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.875994705394794e-07,
+      "loss": 0.6825,
+      "step": 5022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.808713892649587e-07,
+      "loss": 0.7546,
+      "step": 5023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.741819365586154e-07,
+      "loss": 0.7751,
+      "step": 5024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.675311150195928e-07,
+      "loss": 0.7319,
+      "step": 5025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.609189272320237e-07,
+      "loss": 0.7234,
+      "step": 5026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.543453757650086e-07,
+      "loss": 0.7542,
+      "step": 5027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.478104631726711e-07,
+      "loss": 0.7041,
+      "step": 5028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4131419199408e-07,
+      "loss": 0.7512,
+      "step": 5029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.348565647533388e-07,
+      "loss": 0.7583,
+      "step": 5030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.284375839594957e-07,
+      "loss": 0.7271,
+      "step": 5031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.220572521066002e-07,
+      "loss": 0.7263,
+      "step": 5032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.157155716736916e-07,
+      "loss": 0.7512,
+      "step": 5033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.094125451247655e-07,
+      "loss": 0.7141,
+      "step": 5034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.031481749088296e-07,
+      "loss": 0.6926,
+      "step": 5035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.969224634598591e-07,
+      "loss": 0.7451,
+      "step": 5036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.907354131968078e-07,
+      "loss": 0.7424,
+      "step": 5037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.845870265235975e-07,
+      "loss": 0.741,
+      "step": 5038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.784773058291503e-07,
+      "loss": 0.7412,
+      "step": 5039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7240625348735633e-07,
+      "loss": 0.7097,
+      "step": 5040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6637387185706206e-07,
+      "loss": 0.748,
+      "step": 5041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6038016328211476e-07,
+      "loss": 0.3777,
+      "step": 5042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.544251300913405e-07,
+      "loss": 0.7832,
+      "step": 5043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4850877459848837e-07,
+      "loss": 0.7439,
+      "step": 5044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.426310991023419e-07,
+      "loss": 0.7617,
+      "step": 5045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3679210588661866e-07,
+      "loss": 0.7062,
+      "step": 5046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3099179722002616e-07,
+      "loss": 0.7454,
+      "step": 5047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.252301753562171e-07,
+      "loss": 0.729,
+      "step": 5048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1950724253383423e-07,
+      "loss": 0.6981,
+      "step": 5049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1382300097647653e-07,
+      "loss": 0.7622,
+      "step": 5050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.081774528927218e-07,
+      "loss": 0.7632,
+      "step": 5051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.025706004760932e-07,
+      "loss": 0.7122,
+      "step": 5052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.970024459051147e-07,
+      "loss": 0.6948,
+      "step": 5053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.914729913432336e-07,
+      "loss": 0.3503,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8598223893887566e-07,
+      "loss": 0.7188,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.805301908254455e-07,
+      "loss": 0.741,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.751168491212931e-07,
+      "loss": 0.7344,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.697422159297248e-07,
+      "loss": 0.7628,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.644062933390147e-07,
+      "loss": 0.7197,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5910908342241533e-07,
+      "loss": 0.7412,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5385058823809156e-07,
+      "loss": 0.6951,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.486308098291979e-07,
+      "loss": 0.7561,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4344975022385653e-07,
+      "loss": 0.709,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383074114351237e-07,
+      "loss": 0.7302,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3320379546101234e-07,
+      "loss": 0.7512,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.281389042844918e-07,
+      "loss": 0.7522,
+      "step": 5066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.23112739873499e-07,
+      "loss": 0.6985,
+      "step": 5067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1812530418090513e-07,
+      "loss": 0.752,
+      "step": 5068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.13176599144549e-07,
+      "loss": 0.7146,
+      "step": 5069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0826662668720364e-07,
+      "loss": 0.6804,
+      "step": 5070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0339538871660965e-07,
+      "loss": 0.7144,
+      "step": 5071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.98562887125442e-07,
+      "loss": 0.6941,
+      "step": 5072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9376912379134314e-07,
+      "loss": 0.7178,
+      "step": 5073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.890141005768898e-07,
+      "loss": 0.349,
+      "step": 5074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.842978193296153e-07,
+      "loss": 0.7408,
+      "step": 5075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7962028188198706e-07,
+      "loss": 0.7664,
+      "step": 5076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7498149005144025e-07,
+      "loss": 0.7549,
+      "step": 5077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7038144564033305e-07,
+      "loss": 0.7114,
+      "step": 5078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.658201504359803e-07,
+      "loss": 0.7334,
+      "step": 5079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6129760621063093e-07,
+      "loss": 0.7573,
+      "step": 5080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5681381472151267e-07,
+      "loss": 0.7239,
+      "step": 5081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5236877771074306e-07,
+      "loss": 0.3814,
+      "step": 5082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.479624969054295e-07,
+      "loss": 0.7354,
+      "step": 5083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4359497401758024e-07,
+      "loss": 0.7527,
+      "step": 5084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3926621074417124e-07,
+      "loss": 0.7095,
+      "step": 5085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3497620876711257e-07,
+      "loss": 0.6946,
+      "step": 5086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3072496975325985e-07,
+      "loss": 0.6907,
+      "step": 5087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.265124953543918e-07,
+      "loss": 0.728,
+      "step": 5088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2233878720723245e-07,
+      "loss": 0.6976,
+      "step": 5089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.182038469334513e-07,
+      "loss": 0.769,
+      "step": 5090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.141076761396521e-07,
+      "loss": 0.7441,
+      "step": 5091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1005027641736175e-07,
+      "loss": 0.7114,
+      "step": 5092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0603164934306362e-07,
+      "loss": 0.7393,
+      "step": 5093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020517964781532e-07,
+      "loss": 0.3665,
+      "step": 5094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.981107193689713e-07,
+      "loss": 0.7385,
+      "step": 5095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9420841954681525e-07,
+      "loss": 0.7568,
+      "step": 5096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9034489852787218e-07,
+      "loss": 0.7373,
+      "step": 5097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.86520157813308e-07,
+      "loss": 0.7639,
+      "step": 5098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.827341988891673e-07,
+      "loss": 0.7451,
+      "step": 5099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7898702322648453e-07,
+      "loss": 0.7438,
+      "step": 5100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.752786322811839e-07,
+      "loss": 0.7092,
+      "step": 5101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.716090274941351e-07,
+      "loss": 0.7117,
+      "step": 5102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6797821029113093e-07,
+      "loss": 0.6902,
+      "step": 5103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6438618208290957e-07,
+      "loss": 0.7444,
+      "step": 5104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.608329442651213e-07,
+      "loss": 0.7134,
+      "step": 5105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731849821833954e-07,
+      "loss": 0.7402,
+      "step": 5106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5384284530810533e-07,
+      "loss": 0.6941,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5040598688482732e-07,
+      "loss": 0.7185,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4700792428388177e-07,
+      "loss": 0.6985,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.436486588255681e-07,
+      "loss": 0.7004,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.403281918150978e-07,
+      "loss": 0.7341,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3704652454261668e-07,
+      "loss": 0.7485,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3380365828318252e-07,
+      "loss": 0.7134,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3059959429679857e-07,
+      "loss": 0.7188,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.274343338283801e-07,
+      "loss": 0.7241,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2430787810776555e-07,
+      "loss": 0.6823,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2122022834970547e-07,
+      "loss": 0.7339,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1817138575389575e-07,
+      "loss": 0.6875,
+      "step": 5118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1516135150493324e-07,
+      "loss": 0.7233,
+      "step": 5119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1219012677234908e-07,
+      "loss": 0.7679,
+      "step": 5120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0925771271058649e-07,
+      "loss": 0.7534,
+      "step": 5121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0636411045902294e-07,
+      "loss": 0.7302,
+      "step": 5122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.035093211419369e-07,
+      "loss": 0.6917,
+      "step": 5123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0069334586854107e-07,
+      "loss": 0.3744,
+      "step": 5124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.791618573294913e-08,
+      "loss": 0.3546,
+      "step": 5125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.517784181422019e-08,
+      "loss": 0.7512,
+      "step": 5126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.247831517632532e-08,
+      "loss": 0.8042,
+      "step": 5127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.981760686811446e-08,
+      "loss": 0.7629,
+      "step": 5128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.719571792341618e-08,
+      "loss": 0.8154,
+      "step": 5129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.461264936093783e-08,
+      "loss": 0.7834,
+      "step": 5130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.206840218430989e-08,
+      "loss": 0.769,
+      "step": 5131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.956297738207497e-08,
+      "loss": 0.7258,
+      "step": 5132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.709637592770991e-08,
+      "loss": 0.7844,
+      "step": 5133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.466859877958144e-08,
+      "loss": 0.7559,
+      "step": 5134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.227964688099053e-08,
+      "loss": 0.7205,
+      "step": 5135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.992952116013918e-08,
+      "loss": 0.769,
+      "step": 5136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.761822253017469e-08,
+      "loss": 0.707,
+      "step": 5137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.534575188911207e-08,
+      "loss": 0.7256,
+      "step": 5138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.311211011990059e-08,
+      "loss": 0.7512,
+      "step": 5139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.09172980904238e-08,
+      "loss": 0.7278,
+      "step": 5140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8761316653455076e-08,
+      "loss": 0.7883,
+      "step": 5141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6644166646668826e-08,
+      "loss": 0.7441,
+      "step": 5142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.456584889269589e-08,
+      "loss": 0.7712,
+      "step": 5143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.252636419902368e-08,
+      "loss": 0.3389,
+      "step": 5144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.052571335808498e-08,
+      "loss": 0.739,
+      "step": 5145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.856389714723575e-08,
+      "loss": 0.7524,
+      "step": 5146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.66409163287107e-08,
+      "loss": 0.7817,
+      "step": 5147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.475677164966774e-08,
+      "loss": 0.7332,
+      "step": 5148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.291146384217681e-08,
+      "loss": 0.7621,
+      "step": 5149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.110499362323106e-08,
+      "loss": 0.7925,
+      "step": 5150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.933736169471347e-08,
+      "loss": 0.7443,
+      "step": 5151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7608568743419117e-08,
+      "loss": 0.7317,
+      "step": 5152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.591861544106623e-08,
+      "loss": 0.7344,
+      "step": 5153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4267502444274015e-08,
+      "loss": 0.7236,
+      "step": 5154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.265523039456264e-08,
+      "loss": 0.7363,
+      "step": 5155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1081799918375454e-08,
+      "loss": 0.7673,
+      "step": 5156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9547211627045657e-08,
+      "loss": 0.7175,
+      "step": 5157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.805146611684073e-08,
+      "loss": 0.7808,
+      "step": 5158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6594563968918018e-08,
+      "loss": 0.7471,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5176505749346936e-08,
+      "loss": 0.3408,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.379729200908676e-08,
+      "loss": 0.707,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2456923284042142e-08,
+      "loss": 0.761,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1155400094996504e-08,
+      "loss": 0.7715,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9892722947645326e-08,
+      "loss": 0.6584,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8668892332596167e-08,
+      "loss": 0.7605,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7483908725357545e-08,
+      "loss": 0.7361,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.633777258633895e-08,
+      "loss": 0.7556,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5230484360873044e-08,
+      "loss": 0.6896,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4162044479182346e-08,
+      "loss": 0.7852,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3132453356412555e-08,
+      "loss": 0.3531,
+      "step": 5170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2141711392588129e-08,
+      "loss": 0.6407,
+      "step": 5171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1189818972656696e-08,
+      "loss": 0.7183,
+      "step": 5172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0276776466477955e-08,
+      "loss": 0.7047,
+      "step": 5173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.402584228801469e-09,
+      "loss": 0.7285,
+      "step": 5174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.567242599299973e-09,
+      "loss": 0.7317,
+      "step": 5175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.770751902513862e-09,
+      "loss": 0.7349,
+      "step": 5176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.013112447940007e-09,
+      "loss": 0.7205,
+      "step": 5177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.294324529942941e-09,
+      "loss": 0.77,
+      "step": 5178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.614388427799266e-09,
+      "loss": 0.7906,
+      "step": 5179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.973304405697654e-09,
+      "loss": 0.7673,
+      "step": 5180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.371072712727742e-09,
+      "loss": 0.7571,
+      "step": 5181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.807693582869032e-09,
+      "loss": 0.6914,
+      "step": 5182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2831672350353005e-09,
+      "loss": 0.7114,
+      "step": 5183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7974938730190857e-09,
+      "loss": 0.791,
+      "step": 5184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3506736855360977e-09,
+      "loss": 0.7256,
+      "step": 5185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9427068461808083e-09,
+      "loss": 0.7954,
+      "step": 5186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5735935134708613e-09,
+      "loss": 0.7484,
+      "step": 5187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2433338308137644e-09,
+      "loss": 0.6531,
+      "step": 5188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.519279265512993e-10,
+      "loss": 0.7339,
+      "step": 5189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.993759138929079e-10,
+      "loss": 0.7153,
+      "step": 5190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.856778909601012e-10,
+      "loss": 0.7891,
+      "step": 5191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.108339407975613e-10,
+      "loss": 0.752,
+      "step": 5192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.74844131328733e-10,
+      "loss": 0.7507,
+      "step": 5193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.770851538912993e-11,
+      "loss": 0.749,
+      "step": 5194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9427130737437182e-11,
+      "loss": 0.7019,
+      "step": 5195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4332,
+      "step": 5196
+    },
+    {
+      "epoch": 1.0,
+      "step": 5196,
+      "total_flos": 2.480357566185472e+16,
+      "train_loss": 0.7582823450148702,
+      "train_runtime": 26578.0248,
+      "train_samples_per_second": 25.023,
+      "train_steps_per_second": 0.195
+    }
+  ],
+  "max_steps": 5196,
+  "num_train_epochs": 1,
+  "total_flos": 2.480357566185472e+16,
+  "trial_name": null,
+  "trial_params": null
+}